Show More
@@ -1,76 +1,76 b'' | |||
|
1 | 1 | # -*- coding: utf-8 -*- |
|
2 | 2 | """Release data for the IPython project. |
|
3 | 3 | |
|
4 |
$Id: Release.py 98 |
|
|
4 | $Id: Release.py 982 2005-12-30 23:57:07Z fperez $""" | |
|
5 | 5 | |
|
6 | 6 | #***************************************************************************** |
|
7 | 7 | # Copyright (C) 2001-2005 Fernando Perez <fperez@colorado.edu> |
|
8 | 8 | # |
|
9 | 9 | # Copyright (c) 2001 Janko Hauser <jhauser@zscout.de> and Nathaniel Gray |
|
10 | 10 | # <n8gray@caltech.edu> |
|
11 | 11 | # |
|
12 | 12 | # Distributed under the terms of the BSD License. The full license is in |
|
13 | 13 | # the file COPYING, distributed as part of this software. |
|
14 | 14 | #***************************************************************************** |
|
15 | 15 | |
|
16 | 16 | # Name of the package for release purposes. This is the name which labels |
|
17 | 17 | # the tarballs and RPMs made by distutils, so it's best to lowercase it. |
|
18 | 18 | name = 'ipython' |
|
19 | 19 | |
|
20 | 20 | # For versions with substrings (like 0.6.16.svn), use an extra . to separate |
|
21 | 21 | # the new substring. We have to avoid using either dashes or underscores, |
|
22 | 22 | # because bdist_rpm does not accept dashes (an RPM) convention, and |
|
23 | 23 | # bdist_deb does not accept underscores (a Debian convention). |
|
24 | 24 | |
|
25 |
version = '0.7.0.rc |
|
|
25 | version = '0.7.0.rc3' | |
|
26 | 26 | |
|
27 |
revision = '$Revision: 98 |
|
|
27 | revision = '$Revision: 982 $' | |
|
28 | 28 | |
|
29 | 29 | description = "An enhanced interactive Python shell." |
|
30 | 30 | |
|
31 | 31 | long_description = \ |
|
32 | 32 | """ |
|
33 | 33 | IPython provides a replacement for the interactive Python interpreter with |
|
34 | 34 | extra functionality. |
|
35 | 35 | |
|
36 | 36 | Main features: |
|
37 | 37 | |
|
38 | 38 | * Comprehensive object introspection. |
|
39 | 39 | |
|
40 | 40 | * Input history, persistent across sessions. |
|
41 | 41 | |
|
42 | 42 | * Caching of output results during a session with automatically generated |
|
43 | 43 | references. |
|
44 | 44 | |
|
45 | 45 | * Readline based name completion. |
|
46 | 46 | |
|
47 | 47 | * Extensible system of 'magic' commands for controlling the environment and |
|
48 | 48 | performing many tasks related either to IPython or the operating system. |
|
49 | 49 | |
|
50 | 50 | * Configuration system with easy switching between different setups (simpler |
|
51 | 51 | than changing $PYTHONSTARTUP environment variables every time). |
|
52 | 52 | |
|
53 | 53 | * Session logging and reloading. |
|
54 | 54 | |
|
55 | 55 | * Extensible syntax processing for special purpose situations. |
|
56 | 56 | |
|
57 | 57 | * Access to the system shell with user-extensible alias system. |
|
58 | 58 | |
|
59 | 59 | * Easily embeddable in other Python programs. |
|
60 | 60 | |
|
61 | 61 | * Integrated access to the pdb debugger and the Python profiler. """ |
|
62 | 62 | |
|
63 | 63 | license = 'BSD' |
|
64 | 64 | |
|
65 | 65 | authors = {'Fernando' : ('Fernando Perez','fperez@colorado.edu'), |
|
66 | 66 | 'Janko' : ('Janko Hauser','jhauser@zscout.de'), |
|
67 | 67 | 'Nathan' : ('Nathaniel Gray','n8gray@caltech.edu') |
|
68 | 68 | } |
|
69 | 69 | |
|
70 | 70 | url = 'http://ipython.scipy.org' |
|
71 | 71 | |
|
72 | 72 | download_url = 'http://ipython.scipy.org/dist' |
|
73 | 73 | |
|
74 | 74 | platforms = ['Linux','Mac OSX','Windows XP/2000/NT','Windows 95/98/ME'] |
|
75 | 75 | |
|
76 | 76 | keywords = ['Interactive','Interpreter','Shell'] |
@@ -1,2056 +1,2060 b'' | |||
|
1 | 1 | # -*- coding: utf-8 -*- |
|
2 | 2 | """ |
|
3 | 3 | IPython -- An enhanced Interactive Python |
|
4 | 4 | |
|
5 | 5 | Requires Python 2.1 or newer. |
|
6 | 6 | |
|
7 | 7 | This file contains all the classes and helper functions specific to IPython. |
|
8 | 8 | |
|
9 |
$Id: iplib.py 9 |
|
|
9 | $Id: iplib.py 982 2005-12-30 23:57:07Z fperez $ | |
|
10 | 10 | """ |
|
11 | 11 | |
|
12 | 12 | #***************************************************************************** |
|
13 | 13 | # Copyright (C) 2001 Janko Hauser <jhauser@zscout.de> and |
|
14 | 14 | # Copyright (C) 2001-2005 Fernando Perez. <fperez@colorado.edu> |
|
15 | 15 | # |
|
16 | 16 | # Distributed under the terms of the BSD License. The full license is in |
|
17 | 17 | # the file COPYING, distributed as part of this software. |
|
18 | 18 | # |
|
19 | 19 | # Note: this code originally subclassed code.InteractiveConsole from the |
|
20 | 20 | # Python standard library. Over time, all of that class has been copied |
|
21 | 21 | # verbatim here for modifications which could not be accomplished by |
|
22 | 22 | # subclassing. At this point, there are no dependencies at all on the code |
|
23 | 23 | # module anymore (it is not even imported). The Python License (sec. 2) |
|
24 | 24 | # allows for this, but it's always nice to acknowledge credit where credit is |
|
25 | 25 | # due. |
|
26 | 26 | #***************************************************************************** |
|
27 | 27 | |
|
28 | 28 | #**************************************************************************** |
|
29 | 29 | # Modules and globals |
|
30 | 30 | |
|
31 | 31 | from __future__ import generators # for 2.2 backwards-compatibility |
|
32 | 32 | |
|
33 | 33 | from IPython import Release |
|
34 | 34 | __author__ = '%s <%s>\n%s <%s>' % \ |
|
35 | 35 | ( Release.authors['Janko'] + Release.authors['Fernando'] ) |
|
36 | 36 | __license__ = Release.license |
|
37 | 37 | __version__ = Release.version |
|
38 | 38 | |
|
39 | 39 | # Python standard modules |
|
40 | 40 | import __main__ |
|
41 | 41 | import __builtin__ |
|
42 | 42 | import StringIO |
|
43 | 43 | import bdb |
|
44 | 44 | import cPickle as pickle |
|
45 | 45 | import codeop |
|
46 | 46 | import exceptions |
|
47 | 47 | import glob |
|
48 | 48 | import inspect |
|
49 | 49 | import keyword |
|
50 | 50 | import new |
|
51 | 51 | import os |
|
52 | 52 | import pdb |
|
53 | 53 | import pydoc |
|
54 | 54 | import re |
|
55 | 55 | import shutil |
|
56 | 56 | import string |
|
57 | 57 | import sys |
|
58 | 58 | import traceback |
|
59 | 59 | import types |
|
60 | 60 | |
|
61 | 61 | from pprint import pprint, pformat |
|
62 | 62 | |
|
63 | 63 | # IPython's own modules |
|
64 | 64 | import IPython |
|
65 | 65 | from IPython import OInspect,PyColorize,ultraTB |
|
66 | 66 | from IPython.ColorANSI import ColorScheme,ColorSchemeTable # too long names |
|
67 | 67 | from IPython.FakeModule import FakeModule |
|
68 | 68 | from IPython.Itpl import Itpl,itpl,printpl,ItplNS,itplns |
|
69 | 69 | from IPython.Logger import Logger |
|
70 | 70 | from IPython.Magic import Magic |
|
71 | 71 | from IPython.Prompts import CachedOutput |
|
72 | 72 | from IPython.Struct import Struct |
|
73 | 73 | from IPython.background_jobs import BackgroundJobManager |
|
74 | 74 | from IPython.usage import cmd_line_usage,interactive_usage |
|
75 | 75 | from IPython.genutils import * |
|
76 | 76 | |
|
77 | 77 | # store the builtin raw_input globally, and use this always, in case user code |
|
78 | 78 | # overwrites it (like wx.py.PyShell does) |
|
79 | 79 | raw_input_original = raw_input |
|
80 | 80 | |
|
81 | 81 | # compiled regexps for autoindent management |
|
82 | 82 | ini_spaces_re = re.compile(r'^(\s+)') |
|
83 | 83 | dedent_re = re.compile(r'^\s+raise|^\s+return|^\s+pass') |
|
84 | 84 | |
|
85 | 85 | #**************************************************************************** |
|
86 | 86 | # Some utility function definitions |
|
87 | 87 | |
|
88 | 88 | def softspace(file, newvalue): |
|
89 | 89 | """Copied from code.py, to remove the dependency""" |
|
90 | 90 | oldvalue = 0 |
|
91 | 91 | try: |
|
92 | 92 | oldvalue = file.softspace |
|
93 | 93 | except AttributeError: |
|
94 | 94 | pass |
|
95 | 95 | try: |
|
96 | 96 | file.softspace = newvalue |
|
97 | 97 | except (AttributeError, TypeError): |
|
98 | 98 | # "attribute-less object" or "read-only attributes" |
|
99 | 99 | pass |
|
100 | 100 | return oldvalue |
|
101 | 101 | |
|
102 | 102 | #**************************************************************************** |
|
103 | 103 | # These special functions get installed in the builtin namespace, to provide |
|
104 | 104 | # programmatic (pure python) access to magics, aliases and system calls. This |
|
105 | 105 | # is important for logging, user scripting, and more. |
|
106 | 106 | |
|
107 | 107 | # We are basically exposing, via normal python functions, the three mechanisms |
|
108 | 108 | # in which ipython offers special call modes (magics for internal control, |
|
109 | 109 | # aliases for direct system access via pre-selected names, and !cmd for |
|
110 | 110 | # calling arbitrary system commands). |
|
111 | 111 | |
|
112 | 112 | def ipmagic(arg_s): |
|
113 | 113 | """Call a magic function by name. |
|
114 | 114 | |
|
115 | 115 | Input: a string containing the name of the magic function to call and any |
|
116 | 116 | additional arguments to be passed to the magic. |
|
117 | 117 | |
|
118 | 118 | ipmagic('name -opt foo bar') is equivalent to typing at the ipython |
|
119 | 119 | prompt: |
|
120 | 120 | |
|
121 | 121 | In[1]: %name -opt foo bar |
|
122 | 122 | |
|
123 | 123 | To call a magic without arguments, simply use ipmagic('name'). |
|
124 | 124 | |
|
125 | 125 | This provides a proper Python function to call IPython's magics in any |
|
126 | 126 | valid Python code you can type at the interpreter, including loops and |
|
127 | 127 | compound statements. It is added by IPython to the Python builtin |
|
128 | 128 | namespace upon initialization.""" |
|
129 | 129 | |
|
130 | 130 | args = arg_s.split(' ',1) |
|
131 | 131 | magic_name = args[0] |
|
132 | 132 | if magic_name.startswith(__IPYTHON__.ESC_MAGIC): |
|
133 | 133 | magic_name = magic_name[1:] |
|
134 | 134 | try: |
|
135 | 135 | magic_args = args[1] |
|
136 | 136 | except IndexError: |
|
137 | 137 | magic_args = '' |
|
138 | 138 | fn = getattr(__IPYTHON__,'magic_'+magic_name,None) |
|
139 | 139 | if fn is None: |
|
140 | 140 | error("Magic function `%s` not found." % magic_name) |
|
141 | 141 | else: |
|
142 | 142 | magic_args = __IPYTHON__.var_expand(magic_args) |
|
143 | 143 | return fn(magic_args) |
|
144 | 144 | |
|
145 | 145 | def ipalias(arg_s): |
|
146 | 146 | """Call an alias by name. |
|
147 | 147 | |
|
148 | 148 | Input: a string containing the name of the alias to call and any |
|
149 | 149 | additional arguments to be passed to the magic. |
|
150 | 150 | |
|
151 | 151 | ipalias('name -opt foo bar') is equivalent to typing at the ipython |
|
152 | 152 | prompt: |
|
153 | 153 | |
|
154 | 154 | In[1]: name -opt foo bar |
|
155 | 155 | |
|
156 | 156 | To call an alias without arguments, simply use ipalias('name'). |
|
157 | 157 | |
|
158 | 158 | This provides a proper Python function to call IPython's aliases in any |
|
159 | 159 | valid Python code you can type at the interpreter, including loops and |
|
160 | 160 | compound statements. It is added by IPython to the Python builtin |
|
161 | 161 | namespace upon initialization.""" |
|
162 | 162 | |
|
163 | 163 | args = arg_s.split(' ',1) |
|
164 | 164 | alias_name = args[0] |
|
165 | 165 | try: |
|
166 | 166 | alias_args = args[1] |
|
167 | 167 | except IndexError: |
|
168 | 168 | alias_args = '' |
|
169 | 169 | if alias_name in __IPYTHON__.alias_table: |
|
170 | 170 | __IPYTHON__.call_alias(alias_name,alias_args) |
|
171 | 171 | else: |
|
172 | 172 | error("Alias `%s` not found." % alias_name) |
|
173 | 173 | |
|
174 | 174 | def ipsystem(arg_s): |
|
175 | 175 | """Make a system call, using IPython.""" |
|
176 | 176 | __IPYTHON__.system(arg_s) |
|
177 | 177 | |
|
178 | 178 | |
|
179 | 179 | #**************************************************************************** |
|
180 | 180 | # Local use exceptions |
|
181 | 181 | class SpaceInInput(exceptions.Exception): pass |
|
182 | 182 | |
|
183 | 183 | #**************************************************************************** |
|
184 | 184 | # Local use classes |
|
185 | 185 | class Bunch: pass |
|
186 | 186 | |
|
187 | 187 | class InputList(list): |
|
188 | 188 | """Class to store user input. |
|
189 | 189 | |
|
190 | 190 | It's basically a list, but slices return a string instead of a list, thus |
|
191 | 191 | allowing things like (assuming 'In' is an instance): |
|
192 | 192 | |
|
193 | 193 | exec In[4:7] |
|
194 | 194 | |
|
195 | 195 | or |
|
196 | 196 | |
|
197 | 197 | exec In[5:9] + In[14] + In[21:25]""" |
|
198 | 198 | |
|
199 | 199 | def __getslice__(self,i,j): |
|
200 | 200 | return ''.join(list.__getslice__(self,i,j)) |
|
201 | 201 | |
|
202 | 202 | class SyntaxTB(ultraTB.ListTB): |
|
203 | 203 | """Extension which holds some state: the last exception value""" |
|
204 | 204 | |
|
205 | 205 | def __init__(self,color_scheme = 'NoColor'): |
|
206 | 206 | ultraTB.ListTB.__init__(self,color_scheme) |
|
207 | 207 | self.last_syntax_error = None |
|
208 | 208 | |
|
209 | 209 | def __call__(self, etype, value, elist): |
|
210 | 210 | self.last_syntax_error = value |
|
211 | 211 | ultraTB.ListTB.__call__(self,etype,value,elist) |
|
212 | 212 | |
|
213 | 213 | def clear_err_state(self): |
|
214 | 214 | """Return the current error state and clear it""" |
|
215 | 215 | e = self.last_syntax_error |
|
216 | 216 | self.last_syntax_error = None |
|
217 | 217 | return e |
|
218 | 218 | |
|
219 | 219 | #**************************************************************************** |
|
220 | 220 | # Main IPython class |
|
221 | 221 | |
|
222 | 222 | # FIXME: the Magic class is a mixin for now, and will unfortunately remain so |
|
223 | 223 | # until a full rewrite is made. I've cleaned all cross-class uses of |
|
224 | 224 | # attributes and methods, but too much user code out there relies on the |
|
225 | 225 | # equlity %foo == __IP.magic_foo, so I can't actually remove the mixin usage. |
|
226 | 226 | # |
|
227 | 227 | # But at least now, all the pieces have been separated and we could, in |
|
228 | 228 | # principle, stop using the mixin. This will ease the transition to the |
|
229 | 229 | # chainsaw branch. |
|
230 | 230 | |
|
231 | 231 | # For reference, the following is the list of 'self.foo' uses in the Magic |
|
232 | 232 | # class as of 2005-12-28. These are names we CAN'T use in the main ipython |
|
233 | 233 | # class, to prevent clashes. |
|
234 | 234 | |
|
235 | 235 | # ['self.__class__', 'self.__dict__', 'self._inspect', 'self._ofind', |
|
236 | 236 | # 'self.arg_err', 'self.extract_input', 'self.format_', 'self.lsmagic', |
|
237 | 237 | # 'self.magic_', 'self.options_table', 'self.parse', 'self.shell', |
|
238 | 238 | # 'self.value'] |
|
239 | 239 | |
|
240 | 240 | class InteractiveShell(Magic): |
|
241 | 241 | """An enhanced console for Python.""" |
|
242 | 242 | |
|
243 | 243 | # class attribute to indicate whether the class supports threads or not. |
|
244 | 244 | # Subclasses with thread support should override this as needed. |
|
245 | 245 | isthreaded = False |
|
246 | 246 | |
|
247 | 247 | def __init__(self,name,usage=None,rc=Struct(opts=None,args=None), |
|
248 | 248 | user_ns = None,user_global_ns=None,banner2='', |
|
249 | 249 | custom_exceptions=((),None),embedded=False): |
|
250 | 250 | |
|
251 | 251 | # some minimal strict typechecks. For some core data structures, I |
|
252 | 252 | # want actual basic python types, not just anything that looks like |
|
253 | 253 | # one. This is especially true for namespaces. |
|
254 | 254 | for ns in (user_ns,user_global_ns): |
|
255 | 255 | if ns is not None and type(ns) != types.DictType: |
|
256 | 256 | raise TypeError,'namespace must be a dictionary' |
|
257 | 257 | |
|
258 | 258 | # Put a reference to self in builtins so that any form of embedded or |
|
259 | 259 | # imported code can test for being inside IPython. |
|
260 | 260 | __builtin__.__IPYTHON__ = self |
|
261 | 261 | |
|
262 | 262 | # And load into builtins ipmagic/ipalias/ipsystem as well |
|
263 | 263 | __builtin__.ipmagic = ipmagic |
|
264 | 264 | __builtin__.ipalias = ipalias |
|
265 | 265 | __builtin__.ipsystem = ipsystem |
|
266 | 266 | |
|
267 | 267 | # Add to __builtin__ other parts of IPython's public API |
|
268 | 268 | __builtin__.ip_set_hook = self.set_hook |
|
269 | 269 | |
|
270 | 270 | # Keep in the builtins a flag for when IPython is active. We set it |
|
271 | 271 | # with setdefault so that multiple nested IPythons don't clobber one |
|
272 | 272 | # another. Each will increase its value by one upon being activated, |
|
273 | 273 | # which also gives us a way to determine the nesting level. |
|
274 | 274 | __builtin__.__dict__.setdefault('__IPYTHON__active',0) |
|
275 | 275 | |
|
276 | 276 | # Do the intuitively correct thing for quit/exit: we remove the |
|
277 | 277 | # builtins if they exist, and our own prefilter routine will handle |
|
278 | 278 | # these special cases |
|
279 | 279 | try: |
|
280 | 280 | del __builtin__.exit, __builtin__.quit |
|
281 | 281 | except AttributeError: |
|
282 | 282 | pass |
|
283 | 283 | |
|
284 | 284 | # Store the actual shell's name |
|
285 | 285 | self.name = name |
|
286 | 286 | |
|
287 | 287 | # We need to know whether the instance is meant for embedding, since |
|
288 | 288 | # global/local namespaces need to be handled differently in that case |
|
289 | 289 | self.embedded = embedded |
|
290 | 290 | |
|
291 | 291 | # command compiler |
|
292 | 292 | self.compile = codeop.CommandCompiler() |
|
293 | 293 | |
|
294 | 294 | # User input buffer |
|
295 | 295 | self.buffer = [] |
|
296 | 296 | |
|
297 | 297 | # Default name given in compilation of code |
|
298 | 298 | self.filename = '<ipython console>' |
|
299 | 299 | |
|
300 | 300 | # Create the namespace where the user will operate. user_ns is |
|
301 | 301 | # normally the only one used, and it is passed to the exec calls as |
|
302 | 302 | # the locals argument. But we do carry a user_global_ns namespace |
|
303 | 303 | # given as the exec 'globals' argument, This is useful in embedding |
|
304 | 304 | # situations where the ipython shell opens in a context where the |
|
305 | 305 | # distinction between locals and globals is meaningful. |
|
306 | 306 | |
|
307 | 307 | # FIXME. For some strange reason, __builtins__ is showing up at user |
|
308 | 308 | # level as a dict instead of a module. This is a manual fix, but I |
|
309 | 309 | # should really track down where the problem is coming from. Alex |
|
310 | 310 | # Schmolck reported this problem first. |
|
311 | 311 | |
|
312 | 312 | # A useful post by Alex Martelli on this topic: |
|
313 | 313 | # Re: inconsistent value from __builtins__ |
|
314 | 314 | # Von: Alex Martelli <aleaxit@yahoo.com> |
|
315 | 315 | # Datum: Freitag 01 Oktober 2004 04:45:34 nachmittags/abends |
|
316 | 316 | # Gruppen: comp.lang.python |
|
317 | 317 | |
|
318 | 318 | # Michael Hohn <hohn@hooknose.lbl.gov> wrote: |
|
319 | 319 | # > >>> print type(builtin_check.get_global_binding('__builtins__')) |
|
320 | 320 | # > <type 'dict'> |
|
321 | 321 | # > >>> print type(__builtins__) |
|
322 | 322 | # > <type 'module'> |
|
323 | 323 | # > Is this difference in return value intentional? |
|
324 | 324 | |
|
325 | 325 | # Well, it's documented that '__builtins__' can be either a dictionary |
|
326 | 326 | # or a module, and it's been that way for a long time. Whether it's |
|
327 | 327 | # intentional (or sensible), I don't know. In any case, the idea is |
|
328 | 328 | # that if you need to access the built-in namespace directly, you |
|
329 | 329 | # should start with "import __builtin__" (note, no 's') which will |
|
330 | 330 | # definitely give you a module. Yeah, it's somewhatΒ confusing:-(. |
|
331 | 331 | |
|
332 | 332 | if user_ns is None: |
|
333 | 333 | # Set __name__ to __main__ to better match the behavior of the |
|
334 | 334 | # normal interpreter. |
|
335 | 335 | user_ns = {'__name__' :'__main__', |
|
336 | 336 | '__builtins__' : __builtin__, |
|
337 | 337 | } |
|
338 | 338 | |
|
339 | 339 | if user_global_ns is None: |
|
340 | 340 | user_global_ns = {} |
|
341 | 341 | |
|
342 | 342 | # Assign namespaces |
|
343 | 343 | # This is the namespace where all normal user variables live |
|
344 | 344 | self.user_ns = user_ns |
|
345 | 345 | # Embedded instances require a separate namespace for globals. |
|
346 | 346 | # Normally this one is unused by non-embedded instances. |
|
347 | 347 | self.user_global_ns = user_global_ns |
|
348 | 348 | # A namespace to keep track of internal data structures to prevent |
|
349 | 349 | # them from cluttering user-visible stuff. Will be updated later |
|
350 | 350 | self.internal_ns = {} |
|
351 | 351 | |
|
352 | 352 | # Namespace of system aliases. Each entry in the alias |
|
353 | 353 | # table must be a 2-tuple of the form (N,name), where N is the number |
|
354 | 354 | # of positional arguments of the alias. |
|
355 | 355 | self.alias_table = {} |
|
356 | 356 | |
|
357 | 357 | # A table holding all the namespaces IPython deals with, so that |
|
358 | 358 | # introspection facilities can search easily. |
|
359 | 359 | self.ns_table = {'user':user_ns, |
|
360 | 360 | 'user_global':user_global_ns, |
|
361 | 361 | 'alias':self.alias_table, |
|
362 | 362 | 'internal':self.internal_ns, |
|
363 | 363 | 'builtin':__builtin__.__dict__ |
|
364 | 364 | } |
|
365 | 365 | |
|
366 | 366 | # The user namespace MUST have a pointer to the shell itself. |
|
367 | 367 | self.user_ns[name] = self |
|
368 | 368 | |
|
369 | 369 | # We need to insert into sys.modules something that looks like a |
|
370 | 370 | # module but which accesses the IPython namespace, for shelve and |
|
371 | 371 | # pickle to work interactively. Normally they rely on getting |
|
372 | 372 | # everything out of __main__, but for embedding purposes each IPython |
|
373 | 373 | # instance has its own private namespace, so we can't go shoving |
|
374 | 374 | # everything into __main__. |
|
375 | 375 | |
|
376 | 376 | # note, however, that we should only do this for non-embedded |
|
377 | 377 | # ipythons, which really mimic the __main__.__dict__ with their own |
|
378 | 378 | # namespace. Embedded instances, on the other hand, should not do |
|
379 | 379 | # this because they need to manage the user local/global namespaces |
|
380 | 380 | # only, but they live within a 'normal' __main__ (meaning, they |
|
381 | 381 | # shouldn't overtake the execution environment of the script they're |
|
382 | 382 | # embedded in). |
|
383 | 383 | |
|
384 | 384 | if not embedded: |
|
385 | 385 | try: |
|
386 | 386 | main_name = self.user_ns['__name__'] |
|
387 | 387 | except KeyError: |
|
388 | 388 | raise KeyError,'user_ns dictionary MUST have a "__name__" key' |
|
389 | 389 | else: |
|
390 | 390 | #print "pickle hack in place" # dbg |
|
391 | 391 | sys.modules[main_name] = FakeModule(self.user_ns) |
|
392 | 392 | |
|
393 | 393 | # List of input with multi-line handling. |
|
394 | 394 | # Fill its zero entry, user counter starts at 1 |
|
395 | 395 | self.input_hist = InputList(['\n']) |
|
396 | 396 | |
|
397 | 397 | # list of visited directories |
|
398 | 398 | try: |
|
399 | 399 | self.dir_hist = [os.getcwd()] |
|
400 | 400 | except IOError, e: |
|
401 | 401 | self.dir_hist = [] |
|
402 | 402 | |
|
403 | 403 | # dict of output history |
|
404 | 404 | self.output_hist = {} |
|
405 | 405 | |
|
406 | 406 | # dict of things NOT to alias (keywords, builtins and some magics) |
|
407 | 407 | no_alias = {} |
|
408 | 408 | no_alias_magics = ['cd','popd','pushd','dhist','alias','unalias'] |
|
409 | 409 | for key in keyword.kwlist + no_alias_magics: |
|
410 | 410 | no_alias[key] = 1 |
|
411 | 411 | no_alias.update(__builtin__.__dict__) |
|
412 | 412 | self.no_alias = no_alias |
|
413 | 413 | |
|
414 | 414 | # make global variables for user access to these |
|
415 | 415 | self.user_ns['_ih'] = self.input_hist |
|
416 | 416 | self.user_ns['_oh'] = self.output_hist |
|
417 | 417 | self.user_ns['_dh'] = self.dir_hist |
|
418 | 418 | |
|
419 | 419 | # user aliases to input and output histories |
|
420 | 420 | self.user_ns['In'] = self.input_hist |
|
421 | 421 | self.user_ns['Out'] = self.output_hist |
|
422 | 422 | |
|
423 | 423 | # Object variable to store code object waiting execution. This is |
|
424 | 424 | # used mainly by the multithreaded shells, but it can come in handy in |
|
425 | 425 | # other situations. No need to use a Queue here, since it's a single |
|
426 | 426 | # item which gets cleared once run. |
|
427 | 427 | self.code_to_run = None |
|
428 | 428 | |
|
429 | 429 | # Job manager (for jobs run as background threads) |
|
430 | 430 | self.jobs = BackgroundJobManager() |
|
431 | 431 | # Put the job manager into builtins so it's always there. |
|
432 | 432 | __builtin__.jobs = self.jobs |
|
433 | 433 | |
|
434 | 434 | # escapes for automatic behavior on the command line |
|
435 | 435 | self.ESC_SHELL = '!' |
|
436 | 436 | self.ESC_HELP = '?' |
|
437 | 437 | self.ESC_MAGIC = '%' |
|
438 | 438 | self.ESC_QUOTE = ',' |
|
439 | 439 | self.ESC_QUOTE2 = ';' |
|
440 | 440 | self.ESC_PAREN = '/' |
|
441 | 441 | |
|
442 | 442 | # And their associated handlers |
|
443 | 443 | self.esc_handlers = {self.ESC_PAREN : self.handle_auto, |
|
444 | 444 | self.ESC_QUOTE : self.handle_auto, |
|
445 | 445 | self.ESC_QUOTE2 : self.handle_auto, |
|
446 | 446 | self.ESC_MAGIC : self.handle_magic, |
|
447 | 447 | self.ESC_HELP : self.handle_help, |
|
448 | 448 | self.ESC_SHELL : self.handle_shell_escape, |
|
449 | 449 | } |
|
450 | 450 | |
|
451 | 451 | # class initializations |
|
452 | 452 | Magic.__init__(self,self) |
|
453 | 453 | |
|
454 | 454 | # Python source parser/formatter for syntax highlighting |
|
455 | 455 | pyformat = PyColorize.Parser().format |
|
456 | 456 | self.pycolorize = lambda src: pyformat(src,'str',self.rc['colors']) |
|
457 | 457 | |
|
458 | 458 | # hooks holds pointers used for user-side customizations |
|
459 | 459 | self.hooks = Struct() |
|
460 | 460 | |
|
461 | 461 | # Set all default hooks, defined in the IPython.hooks module. |
|
462 | 462 | hooks = IPython.hooks |
|
463 | 463 | for hook_name in hooks.__all__: |
|
464 | 464 | self.set_hook(hook_name,getattr(hooks,hook_name)) |
|
465 | 465 | |
|
466 | 466 | # Flag to mark unconditional exit |
|
467 | 467 | self.exit_now = False |
|
468 | 468 | |
|
469 | 469 | self.usage_min = """\ |
|
470 | 470 | An enhanced console for Python. |
|
471 | 471 | Some of its features are: |
|
472 | 472 | - Readline support if the readline library is present. |
|
473 | 473 | - Tab completion in the local namespace. |
|
474 | 474 | - Logging of input, see command-line options. |
|
475 | 475 | - System shell escape via ! , eg !ls. |
|
476 | 476 | - Magic commands, starting with a % (like %ls, %pwd, %cd, etc.) |
|
477 | 477 | - Keeps track of locally defined variables via %who, %whos. |
|
478 | 478 | - Show object information with a ? eg ?x or x? (use ?? for more info). |
|
479 | 479 | """ |
|
480 | 480 | if usage: self.usage = usage |
|
481 | 481 | else: self.usage = self.usage_min |
|
482 | 482 | |
|
483 | 483 | # Storage |
|
484 | 484 | self.rc = rc # This will hold all configuration information |
|
485 | 485 | self.pager = 'less' |
|
486 | 486 | # temporary files used for various purposes. Deleted at exit. |
|
487 | 487 | self.tempfiles = [] |
|
488 | 488 | |
|
489 | 489 | # Keep track of readline usage (later set by init_readline) |
|
490 | 490 | self.has_readline = False |
|
491 | 491 | |
|
492 | 492 | # template for logfile headers. It gets resolved at runtime by the |
|
493 | 493 | # logstart method. |
|
494 | 494 | self.loghead_tpl = \ |
|
495 | 495 | """#log# Automatic Logger file. *** THIS MUST BE THE FIRST LINE *** |
|
496 | 496 | #log# DO NOT CHANGE THIS LINE OR THE TWO BELOW |
|
497 | 497 | #log# opts = %s |
|
498 | 498 | #log# args = %s |
|
499 | 499 | #log# It is safe to make manual edits below here. |
|
500 | 500 | #log#----------------------------------------------------------------------- |
|
501 | 501 | """ |
|
502 | 502 | # for pushd/popd management |
|
503 | 503 | try: |
|
504 | 504 | self.home_dir = get_home_dir() |
|
505 | 505 | except HomeDirError,msg: |
|
506 | 506 | fatal(msg) |
|
507 | 507 | |
|
508 | 508 | self.dir_stack = [os.getcwd().replace(self.home_dir,'~')] |
|
509 | 509 | |
|
510 | 510 | # Functions to call the underlying shell. |
|
511 | 511 | |
|
512 | 512 | # utility to expand user variables via Itpl |
|
513 | 513 | self.var_expand = lambda cmd: str(ItplNS(cmd.replace('#','\#'), |
|
514 | 514 | self.user_ns)) |
|
515 | 515 | # The first is similar to os.system, but it doesn't return a value, |
|
516 | 516 | # and it allows interpolation of variables in the user's namespace. |
|
517 | 517 | self.system = lambda cmd: shell(self.var_expand(cmd), |
|
518 | 518 | header='IPython system call: ', |
|
519 | 519 | verbose=self.rc.system_verbose) |
|
520 | 520 | # These are for getoutput and getoutputerror: |
|
521 | 521 | self.getoutput = lambda cmd: \ |
|
522 | 522 | getoutput(self.var_expand(cmd), |
|
523 | 523 | header='IPython system call: ', |
|
524 | 524 | verbose=self.rc.system_verbose) |
|
525 | 525 | self.getoutputerror = lambda cmd: \ |
|
526 | 526 | getoutputerror(str(ItplNS(cmd.replace('#','\#'), |
|
527 | 527 | self.user_ns)), |
|
528 | 528 | header='IPython system call: ', |
|
529 | 529 | verbose=self.rc.system_verbose) |
|
530 | 530 | |
|
531 | 531 | # RegExp for splitting line contents into pre-char//first |
|
532 | 532 | # word-method//rest. For clarity, each group in on one line. |
|
533 | 533 | |
|
534 | 534 | # WARNING: update the regexp if the above escapes are changed, as they |
|
535 | 535 | # are hardwired in. |
|
536 | 536 | |
|
537 | 537 | # Don't get carried away with trying to make the autocalling catch too |
|
538 | 538 | # much: it's better to be conservative rather than to trigger hidden |
|
539 | 539 | # evals() somewhere and end up causing side effects. |
|
540 | 540 | |
|
541 | 541 | self.line_split = re.compile(r'^([\s*,;/])' |
|
542 | 542 | r'([\?\w\.]+\w*\s*)' |
|
543 | 543 | r'(\(?.*$)') |
|
544 | 544 | |
|
545 | 545 | # Original re, keep around for a while in case changes break something |
|
546 | 546 | #self.line_split = re.compile(r'(^[\s*!\?%,/]?)' |
|
547 | 547 | # r'(\s*[\?\w\.]+\w*\s*)' |
|
548 | 548 | # r'(\(?.*$)') |
|
549 | 549 | |
|
550 | 550 | # RegExp to identify potential function names |
|
551 | 551 | self.re_fun_name = re.compile(r'[a-zA-Z_]([a-zA-Z0-9_.]*) *$') |
|
552 | 552 | # RegExp to exclude strings with this start from autocalling |
|
553 | 553 | self.re_exclude_auto = re.compile('^[!=()<>,\*/\+-]|^is ') |
|
554 | 554 | |
|
555 | 555 | # try to catch also methods for stuff in lists/tuples/dicts: off |
|
556 | 556 | # (experimental). For this to work, the line_split regexp would need |
|
557 | 557 | # to be modified so it wouldn't break things at '['. That line is |
|
558 | 558 | # nasty enough that I shouldn't change it until I can test it _well_. |
|
559 | 559 | #self.re_fun_name = re.compile (r'[a-zA-Z_]([a-zA-Z0-9_.\[\]]*) ?$') |
|
560 | 560 | |
|
561 | 561 | # keep track of where we started running (mainly for crash post-mortem) |
|
562 | 562 | self.starting_dir = os.getcwd() |
|
563 | 563 | |
|
564 | 564 | # Various switches which can be set |
|
565 | 565 | self.CACHELENGTH = 5000 # this is cheap, it's just text |
|
566 | 566 | self.BANNER = "Python %(version)s on %(platform)s\n" % sys.__dict__ |
|
567 | 567 | self.banner2 = banner2 |
|
568 | 568 | |
|
569 | 569 | # TraceBack handlers: |
|
570 | 570 | |
|
571 | 571 | # Syntax error handler. |
|
572 | 572 | self.SyntaxTB = SyntaxTB(color_scheme='NoColor') |
|
573 | 573 | |
|
574 | 574 | # The interactive one is initialized with an offset, meaning we always |
|
575 | 575 | # want to remove the topmost item in the traceback, which is our own |
|
576 | 576 | # internal code. Valid modes: ['Plain','Context','Verbose'] |
|
577 | 577 | self.InteractiveTB = ultraTB.AutoFormattedTB(mode = 'Plain', |
|
578 | 578 | color_scheme='NoColor', |
|
579 | 579 | tb_offset = 1) |
|
580 | 580 | |
|
581 | 581 | # IPython itself shouldn't crash. This will produce a detailed |
|
582 | 582 | # post-mortem if it does. But we only install the crash handler for |
|
583 | 583 | # non-threaded shells, the threaded ones use a normal verbose reporter |
|
584 | 584 | # and lose the crash handler. This is because exceptions in the main |
|
585 | 585 | # thread (such as in GUI code) propagate directly to sys.excepthook, |
|
586 | 586 | # and there's no point in printing crash dumps for every user exception. |
|
587 | 587 | if self.isthreaded: |
|
588 | 588 | sys.excepthook = ultraTB.FormattedTB() |
|
589 | 589 | else: |
|
590 | 590 | from IPython import CrashHandler |
|
591 | 591 | sys.excepthook = CrashHandler.CrashHandler(self) |
|
592 | 592 | |
|
593 | 593 | # The instance will store a pointer to this, so that runtime code |
|
594 | 594 | # (such as magics) can access it. This is because during the |
|
595 | 595 | # read-eval loop, it gets temporarily overwritten (to deal with GUI |
|
596 | 596 | # frameworks). |
|
597 | 597 | self.sys_excepthook = sys.excepthook |
|
598 | 598 | |
|
599 | 599 | # and add any custom exception handlers the user may have specified |
|
600 | 600 | self.set_custom_exc(*custom_exceptions) |
|
601 | 601 | |
|
602 | 602 | # Object inspector |
|
603 | 603 | self.inspector = OInspect.Inspector(OInspect.InspectColors, |
|
604 | 604 | PyColorize.ANSICodeColors, |
|
605 | 605 | 'NoColor') |
|
606 | 606 | # indentation management |
|
607 | 607 | self.autoindent = False |
|
608 | 608 | self.indent_current_nsp = 0 |
|
609 | 609 | self.indent_current = '' # actual indent string |
|
610 | 610 | |
|
611 | 611 | # Make some aliases automatically |
|
612 | 612 | # Prepare list of shell aliases to auto-define |
|
613 | 613 | if os.name == 'posix': |
|
614 | 614 | auto_alias = ('mkdir mkdir', 'rmdir rmdir', |
|
615 | 615 | 'mv mv -i','rm rm -i','cp cp -i', |
|
616 | 616 | 'cat cat','less less','clear clear', |
|
617 | 617 | # a better ls |
|
618 | 618 | 'ls ls -F', |
|
619 | 619 | # long ls |
|
620 | 620 | 'll ls -lF', |
|
621 | 621 | # color ls |
|
622 | 622 | 'lc ls -F -o --color', |
|
623 | 623 | # ls normal files only |
|
624 | 624 | 'lf ls -F -o --color %l | grep ^-', |
|
625 | 625 | # ls symbolic links |
|
626 | 626 | 'lk ls -F -o --color %l | grep ^l', |
|
627 | 627 | # directories or links to directories, |
|
628 | 628 | 'ldir ls -F -o --color %l | grep /$', |
|
629 | 629 | # things which are executable |
|
630 | 630 | 'lx ls -F -o --color %l | grep ^-..x', |
|
631 | 631 | ) |
|
632 | 632 | elif os.name in ['nt','dos']: |
|
633 | 633 | auto_alias = ('dir dir /on', 'ls dir /on', |
|
634 | 634 | 'ddir dir /ad /on', 'ldir dir /ad /on', |
|
635 | 635 | 'mkdir mkdir','rmdir rmdir','echo echo', |
|
636 | 636 | 'ren ren','cls cls','copy copy') |
|
637 | 637 | else: |
|
638 | 638 | auto_alias = () |
|
639 | 639 | self.auto_alias = map(lambda s:s.split(None,1),auto_alias) |
|
640 | 640 | # Call the actual (public) initializer |
|
641 | 641 | self.init_auto_alias() |
|
642 | 642 | # end __init__ |
|
643 | 643 | |
|
644 | 644 | def post_config_initialization(self): |
|
645 | 645 | """Post configuration init method |
|
646 | 646 | |
|
647 | 647 | This is called after the configuration files have been processed to |
|
648 | 648 | 'finalize' the initialization.""" |
|
649 | 649 | |
|
650 | 650 | rc = self.rc |
|
651 | 651 | |
|
652 | 652 | # Load readline proper |
|
653 | 653 | if rc.readline: |
|
654 | 654 | self.init_readline() |
|
655 | 655 | |
|
656 | 656 | # log system |
|
657 | 657 | self.logger = Logger(self,logfname='ipython_log.py',logmode='rotate') |
|
658 | 658 | # local shortcut, this is used a LOT |
|
659 | 659 | self.log = self.logger.log |
|
660 | 660 | |
|
661 | 661 | # Initialize cache, set in/out prompts and printing system |
|
662 | 662 | self.outputcache = CachedOutput(self, |
|
663 | 663 | rc.cache_size, |
|
664 | 664 | rc.pprint, |
|
665 | 665 | input_sep = rc.separate_in, |
|
666 | 666 | output_sep = rc.separate_out, |
|
667 | 667 | output_sep2 = rc.separate_out2, |
|
668 | 668 | ps1 = rc.prompt_in1, |
|
669 | 669 | ps2 = rc.prompt_in2, |
|
670 | 670 | ps_out = rc.prompt_out, |
|
671 | 671 | pad_left = rc.prompts_pad_left) |
|
672 | 672 | |
|
673 | 673 | # user may have over-ridden the default print hook: |
|
674 | 674 | try: |
|
675 | 675 | self.outputcache.__class__.display = self.hooks.display |
|
676 | 676 | except AttributeError: |
|
677 | 677 | pass |
|
678 | 678 | |
|
679 | 679 | # I don't like assigning globally to sys, because it means when embedding |
|
680 | 680 | # instances, each embedded instance overrides the previous choice. But |
|
681 | 681 | # sys.displayhook seems to be called internally by exec, so I don't see a |
|
682 | 682 | # way around it. |
|
683 | 683 | sys.displayhook = self.outputcache |
|
684 | 684 | |
|
685 | 685 | # Set user colors (don't do it in the constructor above so that it |
|
686 | 686 | # doesn't crash if colors option is invalid) |
|
687 | 687 | self.magic_colors(rc.colors) |
|
688 | 688 | |
|
689 | 689 | # Set calling of pdb on exceptions |
|
690 | 690 | self.call_pdb = rc.pdb |
|
691 | 691 | |
|
692 | 692 | # Load user aliases |
|
693 | 693 | for alias in rc.alias: |
|
694 | 694 | self.magic_alias(alias) |
|
695 | 695 | |
|
696 | 696 | # dynamic data that survives through sessions |
|
697 | 697 | # XXX make the filename a config option? |
|
698 | 698 | persist_base = 'persist' |
|
699 | 699 | if rc.profile: |
|
700 | 700 | persist_base += '_%s' % rc.profile |
|
701 | 701 | self.persist_fname = os.path.join(rc.ipythondir,persist_base) |
|
702 | 702 | |
|
703 | 703 | try: |
|
704 | 704 | self.persist = pickle.load(file(self.persist_fname)) |
|
705 | 705 | except: |
|
706 | 706 | self.persist = {} |
|
707 | 707 | |
|
708 | 708 | |
|
709 | 709 | for (key, value) in [(k[2:],v) for (k,v) in self.persist.items() if k.startswith('S:')]: |
|
710 | 710 | try: |
|
711 | 711 | obj = pickle.loads(value) |
|
712 | 712 | except: |
|
713 | 713 | |
|
714 | 714 | print "Unable to restore variable '%s', ignoring (use %%store -d to forget!)" % key |
|
715 | 715 | print "The error was:",sys.exc_info()[0] |
|
716 | 716 | continue |
|
717 | 717 | |
|
718 | 718 | |
|
719 | 719 | self.user_ns[key] = obj |
|
720 | 720 | |
|
721 | 721 | |
|
722 | 722 | |
|
723 | 723 | |
|
724 | 724 | def set_hook(self,name,hook): |
|
725 | 725 | """set_hook(name,hook) -> sets an internal IPython hook. |
|
726 | 726 | |
|
727 | 727 | IPython exposes some of its internal API as user-modifiable hooks. By |
|
728 | 728 | resetting one of these hooks, you can modify IPython's behavior to |
|
729 | 729 | call at runtime your own routines.""" |
|
730 | 730 | |
|
731 | 731 | # At some point in the future, this should validate the hook before it |
|
732 | 732 | # accepts it. Probably at least check that the hook takes the number |
|
733 | 733 | # of args it's supposed to. |
|
734 | 734 | setattr(self.hooks,name,new.instancemethod(hook,self,self.__class__)) |
|
735 | 735 | |
|
736 | 736 | def set_custom_exc(self,exc_tuple,handler): |
|
737 | 737 | """set_custom_exc(exc_tuple,handler) |
|
738 | 738 | |
|
739 | 739 | Set a custom exception handler, which will be called if any of the |
|
740 | 740 | exceptions in exc_tuple occur in the mainloop (specifically, in the |
|
741 | 741 | runcode() method. |
|
742 | 742 | |
|
743 | 743 | Inputs: |
|
744 | 744 | |
|
745 | 745 | - exc_tuple: a *tuple* of valid exceptions to call the defined |
|
746 | 746 | handler for. It is very important that you use a tuple, and NOT A |
|
747 | 747 | LIST here, because of the way Python's except statement works. If |
|
748 | 748 | you only want to trap a single exception, use a singleton tuple: |
|
749 | 749 | |
|
750 | 750 | exc_tuple == (MyCustomException,) |
|
751 | 751 | |
|
752 | 752 | - handler: this must be defined as a function with the following |
|
753 | 753 | basic interface: def my_handler(self,etype,value,tb). |
|
754 | 754 | |
|
755 | 755 | This will be made into an instance method (via new.instancemethod) |
|
756 | 756 | of IPython itself, and it will be called if any of the exceptions |
|
757 | 757 | listed in the exc_tuple are caught. If the handler is None, an |
|
758 | 758 | internal basic one is used, which just prints basic info. |
|
759 | 759 | |
|
760 | 760 | WARNING: by putting in your own exception handler into IPython's main |
|
761 | 761 | execution loop, you run a very good chance of nasty crashes. This |
|
762 | 762 | facility should only be used if you really know what you are doing.""" |
|
763 | 763 | |
|
764 | 764 | assert type(exc_tuple)==type(()) , \ |
|
765 | 765 | "The custom exceptions must be given AS A TUPLE." |
|
766 | 766 | |
|
767 | 767 | def dummy_handler(self,etype,value,tb): |
|
768 | 768 | print '*** Simple custom exception handler ***' |
|
769 | 769 | print 'Exception type :',etype |
|
770 | 770 | print 'Exception value:',value |
|
771 | 771 | print 'Traceback :',tb |
|
772 | 772 | print 'Source code :','\n'.join(self.buffer) |
|
773 | 773 | |
|
774 | 774 | if handler is None: handler = dummy_handler |
|
775 | 775 | |
|
776 | 776 | self.CustomTB = new.instancemethod(handler,self,self.__class__) |
|
777 | 777 | self.custom_exceptions = exc_tuple |
|
778 | 778 | |
|
779 | 779 | def set_custom_completer(self,completer,pos=0): |
|
780 | 780 | """set_custom_completer(completer,pos=0) |
|
781 | 781 | |
|
782 | 782 | Adds a new custom completer function. |
|
783 | 783 | |
|
784 | 784 | The position argument (defaults to 0) is the index in the completers |
|
785 | 785 | list where you want the completer to be inserted.""" |
|
786 | 786 | |
|
787 | 787 | newcomp = new.instancemethod(completer,self.Completer, |
|
788 | 788 | self.Completer.__class__) |
|
789 | 789 | self.Completer.matchers.insert(pos,newcomp) |
|
790 | 790 | |
|
791 | 791 | def _get_call_pdb(self): |
|
792 | 792 | return self._call_pdb |
|
793 | 793 | |
|
794 | 794 | def _set_call_pdb(self,val): |
|
795 | 795 | |
|
796 | 796 | if val not in (0,1,False,True): |
|
797 | 797 | raise ValueError,'new call_pdb value must be boolean' |
|
798 | 798 | |
|
799 | 799 | # store value in instance |
|
800 | 800 | self._call_pdb = val |
|
801 | 801 | |
|
802 | 802 | # notify the actual exception handlers |
|
803 | 803 | self.InteractiveTB.call_pdb = val |
|
804 | 804 | if self.isthreaded: |
|
805 | 805 | try: |
|
806 | 806 | self.sys_excepthook.call_pdb = val |
|
807 | 807 | except: |
|
808 | 808 | warn('Failed to activate pdb for threaded exception handler') |
|
809 | 809 | |
|
810 | 810 | call_pdb = property(_get_call_pdb,_set_call_pdb,None, |
|
811 | 811 | 'Control auto-activation of pdb at exceptions') |
|
812 | 812 | |
|
813 | 813 | def complete(self,text): |
|
814 | 814 | """Return a sorted list of all possible completions on text. |
|
815 | 815 | |
|
816 | 816 | Inputs: |
|
817 | 817 | |
|
818 | 818 | - text: a string of text to be completed on. |
|
819 | 819 | |
|
820 | 820 | This is a wrapper around the completion mechanism, similar to what |
|
821 | 821 | readline does at the command line when the TAB key is hit. By |
|
822 | 822 | exposing it as a method, it can be used by other non-readline |
|
823 | 823 | environments (such as GUIs) for text completion. |
|
824 | 824 | |
|
825 | 825 | Simple usage example: |
|
826 | 826 | |
|
827 | 827 | In [1]: x = 'hello' |
|
828 | 828 | |
|
829 | 829 | In [2]: __IP.complete('x.l') |
|
830 | 830 | Out[2]: ['x.ljust', 'x.lower', 'x.lstrip']""" |
|
831 | 831 | |
|
832 | 832 | complete = self.Completer.complete |
|
833 | 833 | state = 0 |
|
834 | 834 | # use a dict so we get unique keys, since ipyhton's multiple |
|
835 | 835 | # completers can return duplicates. |
|
836 | 836 | comps = {} |
|
837 | 837 | while True: |
|
838 | 838 | newcomp = complete(text,state) |
|
839 | 839 | if newcomp is None: |
|
840 | 840 | break |
|
841 | 841 | comps[newcomp] = 1 |
|
842 | 842 | state += 1 |
|
843 | 843 | outcomps = comps.keys() |
|
844 | 844 | outcomps.sort() |
|
845 | 845 | return outcomps |
|
846 | 846 | |
|
847 | 847 | def set_completer_frame(self, frame): |
|
848 | 848 | if frame: |
|
849 | 849 | self.Completer.namespace = frame.f_locals |
|
850 | 850 | self.Completer.global_namespace = frame.f_globals |
|
851 | 851 | else: |
|
852 | 852 | self.Completer.namespace = self.user_ns |
|
853 | 853 | self.Completer.global_namespace = self.user_global_ns |
|
854 | 854 | |
|
855 | 855 | def init_auto_alias(self): |
|
856 | 856 | """Define some aliases automatically. |
|
857 | 857 | |
|
858 | 858 | These are ALL parameter-less aliases""" |
|
859 | 859 | for alias,cmd in self.auto_alias: |
|
860 | 860 | self.alias_table[alias] = (0,cmd) |
|
861 | 861 | |
|
862 | 862 | def alias_table_validate(self,verbose=0): |
|
863 | 863 | """Update information about the alias table. |
|
864 | 864 | |
|
865 | 865 | In particular, make sure no Python keywords/builtins are in it.""" |
|
866 | 866 | |
|
867 | 867 | no_alias = self.no_alias |
|
868 | 868 | for k in self.alias_table.keys(): |
|
869 | 869 | if k in no_alias: |
|
870 | 870 | del self.alias_table[k] |
|
871 | 871 | if verbose: |
|
872 | 872 | print ("Deleting alias <%s>, it's a Python " |
|
873 | 873 | "keyword or builtin." % k) |
|
874 | 874 | |
|
875 | 875 | def set_autoindent(self,value=None): |
|
876 | 876 | """Set the autoindent flag, checking for readline support. |
|
877 | 877 | |
|
878 | 878 | If called with no arguments, it acts as a toggle.""" |
|
879 | 879 | |
|
880 | 880 | if not self.has_readline: |
|
881 | 881 | if os.name == 'posix': |
|
882 | 882 | warn("The auto-indent feature requires the readline library") |
|
883 | 883 | self.autoindent = 0 |
|
884 | 884 | return |
|
885 | 885 | if value is None: |
|
886 | 886 | self.autoindent = not self.autoindent |
|
887 | 887 | else: |
|
888 | 888 | self.autoindent = value |
|
889 | 889 | |
|
890 | 890 | def rc_set_toggle(self,rc_field,value=None): |
|
891 | 891 | """Set or toggle a field in IPython's rc config. structure. |
|
892 | 892 | |
|
893 | 893 | If called with no arguments, it acts as a toggle. |
|
894 | 894 | |
|
895 | 895 | If called with a non-existent field, the resulting AttributeError |
|
896 | 896 | exception will propagate out.""" |
|
897 | 897 | |
|
898 | 898 | rc_val = getattr(self.rc,rc_field) |
|
899 | 899 | if value is None: |
|
900 | 900 | value = not rc_val |
|
901 | 901 | setattr(self.rc,rc_field,value) |
|
902 | 902 | |
|
903 | 903 | def user_setup(self,ipythondir,rc_suffix,mode='install'): |
|
904 | 904 | """Install the user configuration directory. |
|
905 | 905 | |
|
906 | 906 | Can be called when running for the first time or to upgrade the user's |
|
907 | 907 | .ipython/ directory with the mode parameter. Valid modes are 'install' |
|
908 | 908 | and 'upgrade'.""" |
|
909 | 909 | |
|
910 | 910 | def wait(): |
|
911 | 911 | try: |
|
912 | 912 | raw_input("Please press <RETURN> to start IPython.") |
|
913 | 913 | except EOFError: |
|
914 | 914 | print >> Term.cout |
|
915 | 915 | print '*'*70 |
|
916 | 916 | |
|
917 | 917 | cwd = os.getcwd() # remember where we started |
|
918 | 918 | glb = glob.glob |
|
919 | 919 | print '*'*70 |
|
920 | 920 | if mode == 'install': |
|
921 | 921 | print \ |
|
922 | 922 | """Welcome to IPython. I will try to create a personal configuration directory |
|
923 | 923 | where you can customize many aspects of IPython's functionality in:\n""" |
|
924 | 924 | else: |
|
925 | 925 | print 'I am going to upgrade your configuration in:' |
|
926 | 926 | |
|
927 | 927 | print ipythondir |
|
928 | 928 | |
|
929 | 929 | rcdirend = os.path.join('IPython','UserConfig') |
|
930 | 930 | cfg = lambda d: os.path.join(d,rcdirend) |
|
931 | 931 | try: |
|
932 | 932 | rcdir = filter(os.path.isdir,map(cfg,sys.path))[0] |
|
933 | 933 | except IOError: |
|
934 | 934 | warning = """ |
|
935 | 935 | Installation error. IPython's directory was not found. |
|
936 | 936 | |
|
937 | 937 | Check the following: |
|
938 | 938 | |
|
939 | 939 | The ipython/IPython directory should be in a directory belonging to your |
|
940 | 940 | PYTHONPATH environment variable (that is, it should be in a directory |
|
941 | 941 | belonging to sys.path). You can copy it explicitly there or just link to it. |
|
942 | 942 | |
|
943 | 943 | IPython will proceed with builtin defaults. |
|
944 | 944 | """ |
|
945 | 945 | warn(warning) |
|
946 | 946 | wait() |
|
947 | 947 | return |
|
948 | 948 | |
|
949 | 949 | if mode == 'install': |
|
950 | 950 | try: |
|
951 | 951 | shutil.copytree(rcdir,ipythondir) |
|
952 | 952 | os.chdir(ipythondir) |
|
953 | 953 | rc_files = glb("ipythonrc*") |
|
954 | 954 | for rc_file in rc_files: |
|
955 | 955 | os.rename(rc_file,rc_file+rc_suffix) |
|
956 | 956 | except: |
|
957 | 957 | warning = """ |
|
958 | 958 | |
|
959 | 959 | There was a problem with the installation: |
|
960 | 960 | %s |
|
961 | 961 | Try to correct it or contact the developers if you think it's a bug. |
|
962 | 962 | IPython will proceed with builtin defaults.""" % sys.exc_info()[1] |
|
963 | 963 | warn(warning) |
|
964 | 964 | wait() |
|
965 | 965 | return |
|
966 | 966 | |
|
967 | 967 | elif mode == 'upgrade': |
|
968 | 968 | try: |
|
969 | 969 | os.chdir(ipythondir) |
|
970 | 970 | except: |
|
971 | 971 | print """ |
|
972 | 972 | Can not upgrade: changing to directory %s failed. Details: |
|
973 | 973 | %s |
|
974 | 974 | """ % (ipythondir,sys.exc_info()[1]) |
|
975 | 975 | wait() |
|
976 | 976 | return |
|
977 | 977 | else: |
|
978 | 978 | sources = glb(os.path.join(rcdir,'[A-Za-z]*')) |
|
979 | 979 | for new_full_path in sources: |
|
980 | 980 | new_filename = os.path.basename(new_full_path) |
|
981 | 981 | if new_filename.startswith('ipythonrc'): |
|
982 | 982 | new_filename = new_filename + rc_suffix |
|
983 | 983 | # The config directory should only contain files, skip any |
|
984 | 984 | # directories which may be there (like CVS) |
|
985 | 985 | if os.path.isdir(new_full_path): |
|
986 | 986 | continue |
|
987 | 987 | if os.path.exists(new_filename): |
|
988 | 988 | old_file = new_filename+'.old' |
|
989 | 989 | if os.path.exists(old_file): |
|
990 | 990 | os.remove(old_file) |
|
991 | 991 | os.rename(new_filename,old_file) |
|
992 | 992 | shutil.copy(new_full_path,new_filename) |
|
993 | 993 | else: |
|
994 | 994 | raise ValueError,'unrecognized mode for install:',`mode` |
|
995 | 995 | |
|
996 | 996 | # Fix line-endings to those native to each platform in the config |
|
997 | 997 | # directory. |
|
998 | 998 | try: |
|
999 | 999 | os.chdir(ipythondir) |
|
1000 | 1000 | except: |
|
1001 | 1001 | print """ |
|
1002 | 1002 | Problem: changing to directory %s failed. |
|
1003 | 1003 | Details: |
|
1004 | 1004 | %s |
|
1005 | 1005 | |
|
1006 | 1006 | Some configuration files may have incorrect line endings. This should not |
|
1007 | 1007 | cause any problems during execution. """ % (ipythondir,sys.exc_info()[1]) |
|
1008 | 1008 | wait() |
|
1009 | 1009 | else: |
|
1010 | 1010 | for fname in glb('ipythonrc*'): |
|
1011 | 1011 | try: |
|
1012 | 1012 | native_line_ends(fname,backup=0) |
|
1013 | 1013 | except IOError: |
|
1014 | 1014 | pass |
|
1015 | 1015 | |
|
1016 | 1016 | if mode == 'install': |
|
1017 | 1017 | print """ |
|
1018 | 1018 | Successful installation! |
|
1019 | 1019 | |
|
1020 | 1020 | Please read the sections 'Initial Configuration' and 'Quick Tips' in the |
|
1021 | 1021 | IPython manual (there are both HTML and PDF versions supplied with the |
|
1022 | 1022 | distribution) to make sure that your system environment is properly configured |
|
1023 | 1023 | to take advantage of IPython's features.""" |
|
1024 | 1024 | else: |
|
1025 | 1025 | print """ |
|
1026 | 1026 | Successful upgrade! |
|
1027 | 1027 | |
|
1028 | 1028 | All files in your directory: |
|
1029 | 1029 | %(ipythondir)s |
|
1030 | 1030 | which would have been overwritten by the upgrade were backed up with a .old |
|
1031 | 1031 | extension. If you had made particular customizations in those files you may |
|
1032 | 1032 | want to merge them back into the new files.""" % locals() |
|
1033 | 1033 | wait() |
|
1034 | 1034 | os.chdir(cwd) |
|
1035 | 1035 | # end user_setup() |
|
1036 | 1036 | |
|
1037 | 1037 | def atexit_operations(self): |
|
1038 | 1038 | """This will be executed at the time of exit. |
|
1039 | 1039 | |
|
1040 | 1040 | Saving of persistent data should be performed here. """ |
|
1041 | 1041 | |
|
1042 | 1042 | # input history |
|
1043 | 1043 | self.savehist() |
|
1044 | 1044 | |
|
1045 | 1045 | # Cleanup all tempfiles left around |
|
1046 | 1046 | for tfile in self.tempfiles: |
|
1047 | 1047 | try: |
|
1048 | 1048 | os.unlink(tfile) |
|
1049 | 1049 | except OSError: |
|
1050 | 1050 | pass |
|
1051 | 1051 | |
|
1052 | 1052 | # save the "persistent data" catch-all dictionary |
|
1053 | 1053 | try: |
|
1054 | 1054 | pickle.dump(self.persist, open(self.persist_fname,"w")) |
|
1055 | 1055 | except: |
|
1056 | 1056 | print "*** ERROR *** persistent data saving failed." |
|
1057 | 1057 | |
|
1058 | 1058 | def savehist(self): |
|
1059 | 1059 | """Save input history to a file (via readline library).""" |
|
1060 | 1060 | try: |
|
1061 | 1061 | self.readline.write_history_file(self.histfile) |
|
1062 | 1062 | except: |
|
1063 | 1063 | print 'Unable to save IPython command history to file: ' + \ |
|
1064 | 1064 | `self.histfile` |
|
1065 | 1065 | |
|
1066 | 1066 | def pre_readline(self): |
|
1067 | 1067 | """readline hook to be used at the start of each line. |
|
1068 | 1068 | |
|
1069 | 1069 | Currently it handles auto-indent only.""" |
|
1070 | 1070 | |
|
1071 | 1071 | self.readline.insert_text(self.indent_current) |
|
1072 | 1072 | |
|
1073 | 1073 | def init_readline(self): |
|
1074 | 1074 | """Command history completion/saving/reloading.""" |
|
1075 | 1075 | try: |
|
1076 | 1076 | import readline |
|
1077 | 1077 | except ImportError: |
|
1078 | 1078 | self.has_readline = 0 |
|
1079 | 1079 | self.readline = None |
|
1080 | 1080 | # no point in bugging windows users with this every time: |
|
1081 | 1081 | if os.name == 'posix': |
|
1082 | 1082 | warn('Readline services not available on this platform.') |
|
1083 | 1083 | else: |
|
1084 | 1084 | import atexit |
|
1085 | 1085 | from IPython.completer import IPCompleter |
|
1086 | 1086 | self.Completer = IPCompleter(self, |
|
1087 | 1087 | self.user_ns, |
|
1088 | 1088 | self.user_global_ns, |
|
1089 | 1089 | self.rc.readline_omit__names, |
|
1090 | 1090 | self.alias_table) |
|
1091 | 1091 | |
|
1092 | 1092 | # Platform-specific configuration |
|
1093 | 1093 | if os.name == 'nt': |
|
1094 | 1094 | self.readline_startup_hook = readline.set_pre_input_hook |
|
1095 | 1095 | else: |
|
1096 | 1096 | self.readline_startup_hook = readline.set_startup_hook |
|
1097 | 1097 | |
|
1098 | 1098 | # Load user's initrc file (readline config) |
|
1099 | 1099 | inputrc_name = os.environ.get('INPUTRC') |
|
1100 | 1100 | if inputrc_name is None: |
|
1101 | 1101 | home_dir = get_home_dir() |
|
1102 | 1102 | if home_dir is not None: |
|
1103 | 1103 | inputrc_name = os.path.join(home_dir,'.inputrc') |
|
1104 | 1104 | if os.path.isfile(inputrc_name): |
|
1105 | 1105 | try: |
|
1106 | 1106 | readline.read_init_file(inputrc_name) |
|
1107 | 1107 | except: |
|
1108 | 1108 | warn('Problems reading readline initialization file <%s>' |
|
1109 | 1109 | % inputrc_name) |
|
1110 | 1110 | |
|
1111 | 1111 | self.has_readline = 1 |
|
1112 | 1112 | self.readline = readline |
|
1113 | 1113 | # save this in sys so embedded copies can restore it properly |
|
1114 | 1114 | sys.ipcompleter = self.Completer.complete |
|
1115 | 1115 | readline.set_completer(self.Completer.complete) |
|
1116 | 1116 | |
|
1117 | 1117 | # Configure readline according to user's prefs |
|
1118 | 1118 | for rlcommand in self.rc.readline_parse_and_bind: |
|
1119 | 1119 | readline.parse_and_bind(rlcommand) |
|
1120 | 1120 | |
|
1121 | 1121 | # remove some chars from the delimiters list |
|
1122 | 1122 | delims = readline.get_completer_delims() |
|
1123 | 1123 | delims = delims.translate(string._idmap, |
|
1124 | 1124 | self.rc.readline_remove_delims) |
|
1125 | 1125 | readline.set_completer_delims(delims) |
|
1126 | 1126 | # otherwise we end up with a monster history after a while: |
|
1127 | 1127 | readline.set_history_length(1000) |
|
1128 | 1128 | try: |
|
1129 | 1129 | #print '*** Reading readline history' # dbg |
|
1130 | 1130 | readline.read_history_file(self.histfile) |
|
1131 | 1131 | except IOError: |
|
1132 | 1132 | pass # It doesn't exist yet. |
|
1133 | 1133 | |
|
1134 | 1134 | atexit.register(self.atexit_operations) |
|
1135 | 1135 | del atexit |
|
1136 | 1136 | |
|
1137 | 1137 | # Configure auto-indent for all platforms |
|
1138 | 1138 | self.set_autoindent(self.rc.autoindent) |
|
1139 | 1139 | |
|
1140 | 1140 | def _should_recompile(self,e): |
|
1141 | 1141 | """Utility routine for edit_syntax_error""" |
|
1142 | 1142 | |
|
1143 | 1143 | if e.filename in ('<ipython console>','<input>','<string>', |
|
1144 | 1144 | '<console>'): |
|
1145 | 1145 | return False |
|
1146 | 1146 | try: |
|
1147 | 1147 | if not ask_yes_no('Return to editor to correct syntax error? ' |
|
1148 | 1148 | '[Y/n] ','y'): |
|
1149 | 1149 | return False |
|
1150 | 1150 | except EOFError: |
|
1151 | 1151 | return False |
|
1152 | 1152 | self.hooks.fix_error_editor(e.filename,e.lineno,e.offset,e.msg) |
|
1153 | 1153 | return True |
|
1154 | 1154 | |
|
1155 | 1155 | def edit_syntax_error(self): |
|
1156 | 1156 | """The bottom half of the syntax error handler called in the main loop. |
|
1157 | 1157 | |
|
1158 | 1158 | Loop until syntax error is fixed or user cancels. |
|
1159 | 1159 | """ |
|
1160 | 1160 | |
|
1161 | 1161 | while self.SyntaxTB.last_syntax_error: |
|
1162 | 1162 | # copy and clear last_syntax_error |
|
1163 | 1163 | err = self.SyntaxTB.clear_err_state() |
|
1164 | 1164 | if not self._should_recompile(err): |
|
1165 | 1165 | return |
|
1166 | 1166 | try: |
|
1167 | 1167 | # may set last_syntax_error again if a SyntaxError is raised |
|
1168 | 1168 | self.safe_execfile(err.filename,self.shell.user_ns) |
|
1169 | 1169 | except: |
|
1170 | 1170 | self.showtraceback() |
|
1171 | 1171 | else: |
|
1172 | 1172 | f = file(err.filename) |
|
1173 | 1173 | try: |
|
1174 | 1174 | sys.displayhook(f.read()) |
|
1175 | 1175 | finally: |
|
1176 | 1176 | f.close() |
|
1177 | 1177 | |
|
1178 | 1178 | def showsyntaxerror(self, filename=None): |
|
1179 | 1179 | """Display the syntax error that just occurred. |
|
1180 | 1180 | |
|
1181 | 1181 | This doesn't display a stack trace because there isn't one. |
|
1182 | 1182 | |
|
1183 | 1183 | If a filename is given, it is stuffed in the exception instead |
|
1184 | 1184 | of what was there before (because Python's parser always uses |
|
1185 | 1185 | "<string>" when reading from a string). |
|
1186 | 1186 | """ |
|
1187 | 1187 | etype, value, last_traceback = sys.exc_info() |
|
1188 | 1188 | if filename and etype is SyntaxError: |
|
1189 | 1189 | # Work hard to stuff the correct filename in the exception |
|
1190 | 1190 | try: |
|
1191 | 1191 | msg, (dummy_filename, lineno, offset, line) = value |
|
1192 | 1192 | except: |
|
1193 | 1193 | # Not the format we expect; leave it alone |
|
1194 | 1194 | pass |
|
1195 | 1195 | else: |
|
1196 | 1196 | # Stuff in the right filename |
|
1197 | 1197 | try: |
|
1198 | 1198 | # Assume SyntaxError is a class exception |
|
1199 | 1199 | value = SyntaxError(msg, (filename, lineno, offset, line)) |
|
1200 | 1200 | except: |
|
1201 | 1201 | # If that failed, assume SyntaxError is a string |
|
1202 | 1202 | value = msg, (filename, lineno, offset, line) |
|
1203 | 1203 | self.SyntaxTB(etype,value,[]) |
|
1204 | 1204 | |
|
1205 | 1205 | def debugger(self): |
|
1206 | 1206 | """Call the pdb debugger.""" |
|
1207 | 1207 | |
|
1208 | 1208 | if not self.rc.pdb: |
|
1209 | 1209 | return |
|
1210 | 1210 | pdb.pm() |
|
1211 | 1211 | |
|
1212 | 1212 | def showtraceback(self,exc_tuple = None,filename=None): |
|
1213 | 1213 | """Display the exception that just occurred.""" |
|
1214 | 1214 | |
|
1215 | 1215 | # Though this won't be called by syntax errors in the input line, |
|
1216 | 1216 | # there may be SyntaxError cases whith imported code. |
|
1217 | 1217 | if exc_tuple is None: |
|
1218 | 1218 | type, value, tb = sys.exc_info() |
|
1219 | 1219 | else: |
|
1220 | 1220 | type, value, tb = exc_tuple |
|
1221 | 1221 | if type is SyntaxError: |
|
1222 | 1222 | self.showsyntaxerror(filename) |
|
1223 | 1223 | else: |
|
1224 | 1224 | self.InteractiveTB() |
|
1225 | 1225 | if self.InteractiveTB.call_pdb and self.has_readline: |
|
1226 | 1226 | # pdb mucks up readline, fix it back |
|
1227 | 1227 | self.readline.set_completer(self.Completer.complete) |
|
1228 | 1228 | |
|
1229 | 1229 | def mainloop(self,banner=None): |
|
1230 | 1230 | """Creates the local namespace and starts the mainloop. |
|
1231 | 1231 | |
|
1232 | 1232 | If an optional banner argument is given, it will override the |
|
1233 | 1233 | internally created default banner.""" |
|
1234 | 1234 | |
|
1235 | 1235 | if self.rc.c: # Emulate Python's -c option |
|
1236 | 1236 | self.exec_init_cmd() |
|
1237 | 1237 | if banner is None: |
|
1238 | 1238 | if self.rc.banner: |
|
1239 | 1239 | banner = self.BANNER+self.banner2 |
|
1240 | 1240 | else: |
|
1241 | 1241 | banner = '' |
|
1242 | 1242 | self.interact(banner) |
|
1243 | 1243 | |
|
1244 | 1244 | def exec_init_cmd(self): |
|
1245 | 1245 | """Execute a command given at the command line. |
|
1246 | 1246 | |
|
1247 | 1247 | This emulates Python's -c option.""" |
|
1248 | 1248 | |
|
1249 | 1249 | sys.argv = ['-c'] |
|
1250 | 1250 | self.push(self.rc.c) |
|
1251 | 1251 | |
|
1252 | 1252 | def embed_mainloop(self,header='',local_ns=None,global_ns=None,stack_depth=0): |
|
1253 | 1253 | """Embeds IPython into a running python program. |
|
1254 | 1254 | |
|
1255 | 1255 | Input: |
|
1256 | 1256 | |
|
1257 | 1257 | - header: An optional header message can be specified. |
|
1258 | 1258 | |
|
1259 | 1259 | - local_ns, global_ns: working namespaces. If given as None, the |
|
1260 | 1260 | IPython-initialized one is updated with __main__.__dict__, so that |
|
1261 | 1261 | program variables become visible but user-specific configuration |
|
1262 | 1262 | remains possible. |
|
1263 | 1263 | |
|
1264 | 1264 | - stack_depth: specifies how many levels in the stack to go to |
|
1265 | 1265 | looking for namespaces (when local_ns and global_ns are None). This |
|
1266 | 1266 | allows an intermediate caller to make sure that this function gets |
|
1267 | 1267 | the namespace from the intended level in the stack. By default (0) |
|
1268 | 1268 | it will get its locals and globals from the immediate caller. |
|
1269 | 1269 | |
|
1270 | 1270 | Warning: it's possible to use this in a program which is being run by |
|
1271 | 1271 | IPython itself (via %run), but some funny things will happen (a few |
|
1272 | 1272 | globals get overwritten). In the future this will be cleaned up, as |
|
1273 | 1273 | there is no fundamental reason why it can't work perfectly.""" |
|
1274 | 1274 | |
|
1275 | 1275 | # Get locals and globals from caller |
|
1276 | 1276 | if local_ns is None or global_ns is None: |
|
1277 | 1277 | call_frame = sys._getframe(stack_depth).f_back |
|
1278 | 1278 | |
|
1279 | 1279 | if local_ns is None: |
|
1280 | 1280 | local_ns = call_frame.f_locals |
|
1281 | 1281 | if global_ns is None: |
|
1282 | 1282 | global_ns = call_frame.f_globals |
|
1283 | 1283 | |
|
1284 | 1284 | # Update namespaces and fire up interpreter |
|
1285 | 1285 | self.user_ns = local_ns |
|
1286 | 1286 | self.user_global_ns = global_ns |
|
1287 | 1287 | |
|
1288 | 1288 | # Patch for global embedding to make sure that things don't overwrite |
|
1289 | 1289 | # user globals accidentally. Thanks to Richard <rxe@renre-europe.com> |
|
1290 | 1290 | # FIXME. Test this a bit more carefully (the if.. is new) |
|
1291 | 1291 | if local_ns is None and global_ns is None: |
|
1292 | 1292 | self.user_global_ns.update(__main__.__dict__) |
|
1293 | 1293 | |
|
1294 | 1294 | # make sure the tab-completer has the correct frame information, so it |
|
1295 | 1295 | # actually completes using the frame's locals/globals |
|
1296 | 1296 | self.set_completer_frame(call_frame) |
|
1297 | 1297 | |
|
1298 | 1298 | self.interact(header) |
|
1299 | 1299 | |
|
1300 | 1300 | def interact(self, banner=None): |
|
1301 | 1301 | """Closely emulate the interactive Python console. |
|
1302 | 1302 | |
|
1303 | 1303 | The optional banner argument specify the banner to print |
|
1304 | 1304 | before the first interaction; by default it prints a banner |
|
1305 | 1305 | similar to the one printed by the real Python interpreter, |
|
1306 | 1306 | followed by the current class name in parentheses (so as not |
|
1307 | 1307 | to confuse this with the real interpreter -- since it's so |
|
1308 | 1308 | close!). |
|
1309 | 1309 | |
|
1310 | 1310 | """ |
|
1311 | 1311 | cprt = 'Type "copyright", "credits" or "license" for more information.' |
|
1312 | 1312 | if banner is None: |
|
1313 | 1313 | self.write("Python %s on %s\n%s\n(%s)\n" % |
|
1314 | 1314 | (sys.version, sys.platform, cprt, |
|
1315 | 1315 | self.__class__.__name__)) |
|
1316 | 1316 | else: |
|
1317 | 1317 | self.write(banner) |
|
1318 | 1318 | |
|
1319 | 1319 | more = 0 |
|
1320 | 1320 | |
|
1321 | 1321 | # Mark activity in the builtins |
|
1322 | 1322 | __builtin__.__dict__['__IPYTHON__active'] += 1 |
|
1323 | 1323 | |
|
1324 | 1324 | # exit_now is set by a call to %Exit or %Quit |
|
1325 | 1325 | while not self.exit_now: |
|
1326 | 1326 | try: |
|
1327 | 1327 | if more: |
|
1328 | 1328 | prompt = self.outputcache.prompt2 |
|
1329 | 1329 | if self.autoindent: |
|
1330 | 1330 | self.readline_startup_hook(self.pre_readline) |
|
1331 | 1331 | else: |
|
1332 | 1332 | prompt = self.outputcache.prompt1 |
|
1333 | 1333 | try: |
|
1334 | 1334 | line = self.raw_input(prompt,more) |
|
1335 | 1335 | if self.autoindent: |
|
1336 | 1336 | self.readline_startup_hook(None) |
|
1337 | 1337 | except EOFError: |
|
1338 | 1338 | if self.autoindent: |
|
1339 | 1339 | self.readline_startup_hook(None) |
|
1340 | 1340 | self.write("\n") |
|
1341 | 1341 | self.exit() |
|
1342 | 1342 | else: |
|
1343 | 1343 | more = self.push(line) |
|
1344 | 1344 | |
|
1345 | 1345 | if (self.SyntaxTB.last_syntax_error and |
|
1346 | 1346 | self.rc.autoedit_syntax): |
|
1347 | 1347 | self.edit_syntax_error() |
|
1348 | 1348 | |
|
1349 | 1349 | except KeyboardInterrupt: |
|
1350 | 1350 | self.write("\nKeyboardInterrupt\n") |
|
1351 | 1351 | self.resetbuffer() |
|
1352 | 1352 | more = 0 |
|
1353 | 1353 | # keep cache in sync with the prompt counter: |
|
1354 | 1354 | self.outputcache.prompt_count -= 1 |
|
1355 | 1355 | |
|
1356 | 1356 | if self.autoindent: |
|
1357 | 1357 | self.indent_current_nsp = 0 |
|
1358 | 1358 | self.indent_current = ' '* self.indent_current_nsp |
|
1359 | 1359 | |
|
1360 | 1360 | except bdb.BdbQuit: |
|
1361 | 1361 | warn("The Python debugger has exited with a BdbQuit exception.\n" |
|
1362 | 1362 | "Because of how pdb handles the stack, it is impossible\n" |
|
1363 | 1363 | "for IPython to properly format this particular exception.\n" |
|
1364 | 1364 | "IPython will resume normal operation.") |
|
1365 | 1365 | |
|
1366 | 1366 | # We are off again... |
|
1367 | 1367 | __builtin__.__dict__['__IPYTHON__active'] -= 1 |
|
1368 | 1368 | |
|
1369 | 1369 | def excepthook(self, type, value, tb): |
|
1370 | 1370 | """One more defense for GUI apps that call sys.excepthook. |
|
1371 | 1371 | |
|
1372 | 1372 | GUI frameworks like wxPython trap exceptions and call |
|
1373 | 1373 | sys.excepthook themselves. I guess this is a feature that |
|
1374 | 1374 | enables them to keep running after exceptions that would |
|
1375 | 1375 | otherwise kill their mainloop. This is a bother for IPython |
|
1376 | 1376 | which excepts to catch all of the program exceptions with a try: |
|
1377 | 1377 | except: statement. |
|
1378 | 1378 | |
|
1379 | 1379 | Normally, IPython sets sys.excepthook to a CrashHandler instance, so if |
|
1380 | 1380 | any app directly invokes sys.excepthook, it will look to the user like |
|
1381 | 1381 | IPython crashed. In order to work around this, we can disable the |
|
1382 | 1382 | CrashHandler and replace it with this excepthook instead, which prints a |
|
1383 | 1383 | regular traceback using our InteractiveTB. In this fashion, apps which |
|
1384 | 1384 | call sys.excepthook will generate a regular-looking exception from |
|
1385 | 1385 | IPython, and the CrashHandler will only be triggered by real IPython |
|
1386 | 1386 | crashes. |
|
1387 | 1387 | |
|
1388 | 1388 | This hook should be used sparingly, only in places which are not likely |
|
1389 | 1389 | to be true IPython errors. |
|
1390 | 1390 | """ |
|
1391 | 1391 | |
|
1392 | 1392 | self.InteractiveTB(type, value, tb, tb_offset=0) |
|
1393 | 1393 | if self.InteractiveTB.call_pdb and self.has_readline: |
|
1394 | 1394 | self.readline.set_completer(self.Completer.complete) |
|
1395 | 1395 | |
|
1396 | 1396 | def call_alias(self,alias,rest=''): |
|
1397 | 1397 | """Call an alias given its name and the rest of the line. |
|
1398 | 1398 | |
|
1399 | 1399 | This function MUST be given a proper alias, because it doesn't make |
|
1400 | 1400 | any checks when looking up into the alias table. The caller is |
|
1401 | 1401 | responsible for invoking it only with a valid alias.""" |
|
1402 | 1402 | |
|
1403 | 1403 | #print 'ALIAS: <%s>+<%s>' % (alias,rest) # dbg |
|
1404 | 1404 | nargs,cmd = self.alias_table[alias] |
|
1405 | 1405 | # Expand the %l special to be the user's input line |
|
1406 | 1406 | if cmd.find('%l') >= 0: |
|
1407 | 1407 | cmd = cmd.replace('%l',rest) |
|
1408 | 1408 | rest = '' |
|
1409 | 1409 | if nargs==0: |
|
1410 | 1410 | # Simple, argument-less aliases |
|
1411 | 1411 | cmd = '%s %s' % (cmd,rest) |
|
1412 | 1412 | else: |
|
1413 | 1413 | # Handle aliases with positional arguments |
|
1414 | 1414 | args = rest.split(None,nargs) |
|
1415 | 1415 | if len(args)< nargs: |
|
1416 | 1416 | error('Alias <%s> requires %s arguments, %s given.' % |
|
1417 | 1417 | (alias,nargs,len(args))) |
|
1418 | 1418 | return |
|
1419 | 1419 | cmd = '%s %s' % (cmd % tuple(args[:nargs]),' '.join(args[nargs:])) |
|
1420 | 1420 | # Now call the macro, evaluating in the user's namespace |
|
1421 | 1421 | try: |
|
1422 | 1422 | self.system(cmd) |
|
1423 | 1423 | except: |
|
1424 | 1424 | self.showtraceback() |
|
1425 | 1425 | |
|
1426 | 1426 | def autoindent_update(self,line): |
|
1427 | 1427 | """Keep track of the indent level.""" |
|
1428 | 1428 | if self.autoindent: |
|
1429 | 1429 | if line: |
|
1430 | 1430 | ini_spaces = ini_spaces_re.match(line) |
|
1431 | 1431 | if ini_spaces: |
|
1432 | 1432 | nspaces = ini_spaces.end() |
|
1433 | 1433 | else: |
|
1434 | 1434 | nspaces = 0 |
|
1435 | 1435 | self.indent_current_nsp = nspaces |
|
1436 | 1436 | |
|
1437 | 1437 | if line[-1] == ':': |
|
1438 | 1438 | self.indent_current_nsp += 4 |
|
1439 | 1439 | elif dedent_re.match(line): |
|
1440 | 1440 | self.indent_current_nsp -= 4 |
|
1441 | 1441 | else: |
|
1442 | 1442 | self.indent_current_nsp = 0 |
|
1443 | 1443 | |
|
1444 | 1444 | # indent_current is the actual string to be inserted |
|
1445 | 1445 | # by the readline hooks for indentation |
|
1446 | 1446 | self.indent_current = ' '* self.indent_current_nsp |
|
1447 | 1447 | |
|
1448 | 1448 | def runlines(self,lines): |
|
1449 | 1449 | """Run a string of one or more lines of source. |
|
1450 | 1450 | |
|
1451 | 1451 | This method is capable of running a string containing multiple source |
|
1452 | 1452 | lines, as if they had been entered at the IPython prompt. Since it |
|
1453 | 1453 | exposes IPython's processing machinery, the given strings can contain |
|
1454 | 1454 | magic calls (%magic), special shell access (!cmd), etc.""" |
|
1455 | 1455 | |
|
1456 | 1456 | # We must start with a clean buffer, in case this is run from an |
|
1457 | 1457 | # interactive IPython session (via a magic, for example). |
|
1458 | 1458 | self.resetbuffer() |
|
1459 | 1459 | lines = lines.split('\n') |
|
1460 | 1460 | more = 0 |
|
1461 | 1461 | for line in lines: |
|
1462 | 1462 | # skip blank lines so we don't mess up the prompt counter, but do |
|
1463 | 1463 | # NOT skip even a blank line if we are in a code block (more is |
|
1464 | 1464 | # true) |
|
1465 | 1465 | if line or more: |
|
1466 | 1466 | more = self.push(self.prefilter(line,more)) |
|
1467 | 1467 | # IPython's runsource returns None if there was an error |
|
1468 | 1468 | # compiling the code. This allows us to stop processing right |
|
1469 | 1469 | # away, so the user gets the error message at the right place. |
|
1470 | 1470 | if more is None: |
|
1471 | 1471 | break |
|
1472 | 1472 | # final newline in case the input didn't have it, so that the code |
|
1473 | 1473 | # actually does get executed |
|
1474 | 1474 | if more: |
|
1475 | 1475 | self.push('\n') |
|
1476 | 1476 | |
|
1477 | 1477 | def runsource(self, source, filename='<input>', symbol='single'): |
|
1478 | 1478 | """Compile and run some source in the interpreter. |
|
1479 | 1479 | |
|
1480 | 1480 | Arguments are as for compile_command(). |
|
1481 | 1481 | |
|
1482 | 1482 | One several things can happen: |
|
1483 | 1483 | |
|
1484 | 1484 | 1) The input is incorrect; compile_command() raised an |
|
1485 | 1485 | exception (SyntaxError or OverflowError). A syntax traceback |
|
1486 | 1486 | will be printed by calling the showsyntaxerror() method. |
|
1487 | 1487 | |
|
1488 | 1488 | 2) The input is incomplete, and more input is required; |
|
1489 | 1489 | compile_command() returned None. Nothing happens. |
|
1490 | 1490 | |
|
1491 | 1491 | 3) The input is complete; compile_command() returned a code |
|
1492 | 1492 | object. The code is executed by calling self.runcode() (which |
|
1493 | 1493 | also handles run-time exceptions, except for SystemExit). |
|
1494 | 1494 | |
|
1495 | 1495 | The return value is: |
|
1496 | 1496 | |
|
1497 | 1497 | - True in case 2 |
|
1498 | 1498 | |
|
1499 | 1499 | - False in the other cases, unless an exception is raised, where |
|
1500 | 1500 | None is returned instead. This can be used by external callers to |
|
1501 | 1501 | know whether to continue feeding input or not. |
|
1502 | 1502 | |
|
1503 | 1503 | The return value can be used to decide whether to use sys.ps1 or |
|
1504 | 1504 | sys.ps2 to prompt the next line.""" |
|
1505 | 1505 | |
|
1506 | 1506 | try: |
|
1507 | 1507 | code = self.compile(source,filename,symbol) |
|
1508 | 1508 | except (OverflowError, SyntaxError, ValueError): |
|
1509 | 1509 | # Case 1 |
|
1510 | 1510 | self.showsyntaxerror(filename) |
|
1511 | 1511 | return None |
|
1512 | 1512 | |
|
1513 | 1513 | if code is None: |
|
1514 | 1514 | # Case 2 |
|
1515 | 1515 | return True |
|
1516 | 1516 | |
|
1517 | 1517 | # Case 3 |
|
1518 | 1518 | # We store the code object so that threaded shells and |
|
1519 | 1519 | # custom exception handlers can access all this info if needed. |
|
1520 | 1520 | # The source corresponding to this can be obtained from the |
|
1521 | 1521 | # buffer attribute as '\n'.join(self.buffer). |
|
1522 | 1522 | self.code_to_run = code |
|
1523 | 1523 | # now actually execute the code object |
|
1524 | 1524 | if self.runcode(code) == 0: |
|
1525 | 1525 | return False |
|
1526 | 1526 | else: |
|
1527 | 1527 | return None |
|
1528 | 1528 | |
|
1529 | 1529 | def runcode(self,code_obj): |
|
1530 | 1530 | """Execute a code object. |
|
1531 | 1531 | |
|
1532 | 1532 | When an exception occurs, self.showtraceback() is called to display a |
|
1533 | 1533 | traceback. |
|
1534 | 1534 | |
|
1535 | 1535 | Return value: a flag indicating whether the code to be run completed |
|
1536 | 1536 | successfully: |
|
1537 | 1537 | |
|
1538 | 1538 | - 0: successful execution. |
|
1539 | 1539 | - 1: an error occurred. |
|
1540 | 1540 | """ |
|
1541 | 1541 | |
|
1542 | 1542 | # Set our own excepthook in case the user code tries to call it |
|
1543 | 1543 | # directly, so that the IPython crash handler doesn't get triggered |
|
1544 | 1544 | old_excepthook,sys.excepthook = sys.excepthook, self.excepthook |
|
1545 | 1545 | |
|
1546 | 1546 | # we save the original sys.excepthook in the instance, in case config |
|
1547 | 1547 | # code (such as magics) needs access to it. |
|
1548 | 1548 | self.sys_excepthook = old_excepthook |
|
1549 | 1549 | outflag = 1 # happens in more places, so it's easier as default |
|
1550 | 1550 | try: |
|
1551 | 1551 | try: |
|
1552 | 1552 | # Embedded instances require separate global/local namespaces |
|
1553 | 1553 | # so they can see both the surrounding (local) namespace and |
|
1554 | 1554 | # the module-level globals when called inside another function. |
|
1555 | 1555 | if self.embedded: |
|
1556 | 1556 | exec code_obj in self.user_global_ns, self.user_ns |
|
1557 | 1557 | # Normal (non-embedded) instances should only have a single |
|
1558 | 1558 | # namespace for user code execution, otherwise functions won't |
|
1559 | 1559 | # see interactive top-level globals. |
|
1560 | 1560 | else: |
|
1561 | 1561 | exec code_obj in self.user_ns |
|
1562 | 1562 | finally: |
|
1563 | 1563 | # Reset our crash handler in place |
|
1564 | 1564 | sys.excepthook = old_excepthook |
|
1565 | 1565 | except SystemExit: |
|
1566 | 1566 | self.resetbuffer() |
|
1567 | 1567 | self.showtraceback() |
|
1568 | 1568 | warn("Type exit or quit to exit IPython " |
|
1569 | 1569 | "(%Exit or %Quit do so unconditionally).",level=1) |
|
1570 | 1570 | except self.custom_exceptions: |
|
1571 | 1571 | etype,value,tb = sys.exc_info() |
|
1572 | 1572 | self.CustomTB(etype,value,tb) |
|
1573 | 1573 | except: |
|
1574 | 1574 | self.showtraceback() |
|
1575 | 1575 | else: |
|
1576 | 1576 | outflag = 0 |
|
1577 | 1577 | if softspace(sys.stdout, 0): |
|
1578 | 1578 | |
|
1579 | 1579 | # Flush out code object which has been run (and source) |
|
1580 | 1580 | self.code_to_run = None |
|
1581 | 1581 | return outflag |
|
1582 | 1582 | |
|
1583 | 1583 | def push(self, line): |
|
1584 | 1584 | """Push a line to the interpreter. |
|
1585 | 1585 | |
|
1586 | 1586 | The line should not have a trailing newline; it may have |
|
1587 | 1587 | internal newlines. The line is appended to a buffer and the |
|
1588 | 1588 | interpreter's runsource() method is called with the |
|
1589 | 1589 | concatenated contents of the buffer as source. If this |
|
1590 | 1590 | indicates that the command was executed or invalid, the buffer |
|
1591 | 1591 | is reset; otherwise, the command is incomplete, and the buffer |
|
1592 | 1592 | is left as it was after the line was appended. The return |
|
1593 | 1593 | value is 1 if more input is required, 0 if the line was dealt |
|
1594 | 1594 | with in some way (this is the same as runsource()). |
|
1595 | 1595 | """ |
|
1596 | 1596 | |
|
1597 | 1597 | # autoindent management should be done here, and not in the |
|
1598 | 1598 | # interactive loop, since that one is only seen by keyboard input. We |
|
1599 | 1599 | # need this done correctly even for code run via runlines (which uses |
|
1600 | 1600 | # push). |
|
1601 | ||
|
1602 | print 'push line: <%s>' % line # dbg | |
|
1601 | 1603 | self.autoindent_update(line) |
|
1602 | 1604 | |
|
1603 | 1605 | self.buffer.append(line) |
|
1604 | 1606 | more = self.runsource('\n'.join(self.buffer), self.filename) |
|
1605 | 1607 | if not more: |
|
1606 | 1608 | self.resetbuffer() |
|
1607 | 1609 | return more |
|
1608 | 1610 | |
|
1609 | 1611 | def resetbuffer(self): |
|
1610 | 1612 | """Reset the input buffer.""" |
|
1611 | 1613 | self.buffer[:] = [] |
|
1612 | 1614 | |
|
1613 | 1615 | def raw_input(self,prompt='',continue_prompt=False): |
|
1614 | 1616 | """Write a prompt and read a line. |
|
1615 | 1617 | |
|
1616 | 1618 | The returned line does not include the trailing newline. |
|
1617 | 1619 | When the user enters the EOF key sequence, EOFError is raised. |
|
1618 | 1620 | |
|
1619 | 1621 | Optional inputs: |
|
1620 | 1622 | |
|
1621 | 1623 | - prompt(''): a string to be printed to prompt the user. |
|
1622 | 1624 | |
|
1623 | 1625 | - continue_prompt(False): whether this line is the first one or a |
|
1624 | 1626 | continuation in a sequence of inputs. |
|
1625 | 1627 | """ |
|
1626 | 1628 | |
|
1627 | 1629 | line = raw_input_original(prompt) |
|
1628 | 1630 | # Try to be reasonably smart about not re-indenting pasted input more |
|
1629 | 1631 | # than necessary. We do this by trimming out the auto-indent initial |
|
1630 | 1632 | # spaces, if the user's actual input started itself with whitespace. |
|
1631 | 1633 | if self.autoindent: |
|
1632 | 1634 | line2 = line[self.indent_current_nsp:] |
|
1633 | 1635 | if line2[0:1] in (' ','\t'): |
|
1634 | 1636 | line = line2 |
|
1635 | 1637 | return self.prefilter(line,continue_prompt) |
|
1636 | 1638 | |
|
1637 | 1639 | def split_user_input(self,line): |
|
1638 | 1640 | """Split user input into pre-char, function part and rest.""" |
|
1639 | 1641 | |
|
1640 | 1642 | lsplit = self.line_split.match(line) |
|
1641 | 1643 | if lsplit is None: # no regexp match returns None |
|
1642 | 1644 | try: |
|
1643 | 1645 | iFun,theRest = line.split(None,1) |
|
1644 | 1646 | except ValueError: |
|
1645 | 1647 | iFun,theRest = line,'' |
|
1646 | 1648 | pre = re.match('^(\s*)(.*)',line).groups()[0] |
|
1647 | 1649 | else: |
|
1648 | 1650 | pre,iFun,theRest = lsplit.groups() |
|
1649 | 1651 | |
|
1650 | 1652 | #print 'line:<%s>' % line # dbg |
|
1651 | 1653 | #print 'pre <%s> iFun <%s> rest <%s>' % (pre,iFun.strip(),theRest) # dbg |
|
1652 | 1654 | return pre,iFun.strip(),theRest |
|
1653 | 1655 | |
|
1654 | 1656 | def _prefilter(self, line, continue_prompt): |
|
1655 | 1657 | """Calls different preprocessors, depending on the form of line.""" |
|
1656 | 1658 | |
|
1657 | 1659 | # All handlers *must* return a value, even if it's blank (''). |
|
1658 | 1660 | |
|
1659 | 1661 | # Lines are NOT logged here. Handlers should process the line as |
|
1660 | 1662 | # needed, update the cache AND log it (so that the input cache array |
|
1661 | 1663 | # stays synced). |
|
1662 | 1664 | |
|
1663 | 1665 | # This function is _very_ delicate, and since it's also the one which |
|
1664 | 1666 | # determines IPython's response to user input, it must be as efficient |
|
1665 | 1667 | # as possible. For this reason it has _many_ returns in it, trying |
|
1666 | 1668 | # always to exit as quickly as it can figure out what it needs to do. |
|
1667 | 1669 | |
|
1668 | 1670 | # This function is the main responsible for maintaining IPython's |
|
1669 | 1671 | # behavior respectful of Python's semantics. So be _very_ careful if |
|
1670 | 1672 | # making changes to anything here. |
|
1671 | 1673 | |
|
1672 | 1674 | #..................................................................... |
|
1673 | 1675 | # Code begins |
|
1674 | 1676 | |
|
1675 | 1677 | #if line.startswith('%crash'): raise RuntimeError,'Crash now!' # dbg |
|
1676 | 1678 | |
|
1677 | 1679 | # save the line away in case we crash, so the post-mortem handler can |
|
1678 | 1680 | # record it |
|
1679 | 1681 | self._last_input_line = line |
|
1680 | 1682 | |
|
1681 | 1683 | #print '***line: <%s>' % line # dbg |
|
1682 | 1684 | |
|
1683 | 1685 | # the input history needs to track even empty lines |
|
1684 | 1686 | if not line.strip(): |
|
1685 | 1687 | if not continue_prompt: |
|
1686 | 1688 | self.outputcache.prompt_count -= 1 |
|
1687 | 1689 | return self.handle_normal(line,continue_prompt) |
|
1688 | 1690 | #return self.handle_normal('',continue_prompt) |
|
1689 | 1691 | |
|
1690 | 1692 | # print '***cont',continue_prompt # dbg |
|
1691 | 1693 | # special handlers are only allowed for single line statements |
|
1692 | 1694 | if continue_prompt and not self.rc.multi_line_specials: |
|
1693 | 1695 | return self.handle_normal(line,continue_prompt) |
|
1694 | 1696 | |
|
1695 | 1697 | # For the rest, we need the structure of the input |
|
1696 | 1698 | pre,iFun,theRest = self.split_user_input(line) |
|
1697 | 1699 | #print 'pre <%s> iFun <%s> rest <%s>' % (pre,iFun,theRest) # dbg |
|
1698 | 1700 | |
|
1699 | 1701 | # First check for explicit escapes in the last/first character |
|
1700 | 1702 | handler = None |
|
1701 | 1703 | if line[-1] == self.ESC_HELP: |
|
1702 | 1704 | handler = self.esc_handlers.get(line[-1]) # the ? can be at the end |
|
1703 | 1705 | if handler is None: |
|
1704 | 1706 | # look at the first character of iFun, NOT of line, so we skip |
|
1705 | 1707 | # leading whitespace in multiline input |
|
1706 | 1708 | handler = self.esc_handlers.get(iFun[0:1]) |
|
1707 | 1709 | if handler is not None: |
|
1708 | 1710 | return handler(line,continue_prompt,pre,iFun,theRest) |
|
1709 | 1711 | # Emacs ipython-mode tags certain input lines |
|
1710 | 1712 | if line.endswith('# PYTHON-MODE'): |
|
1711 | 1713 | return self.handle_emacs(line,continue_prompt) |
|
1712 | 1714 | |
|
1713 | 1715 | # Next, check if we can automatically execute this thing |
|
1714 | 1716 | |
|
1715 | 1717 | # Allow ! in multi-line statements if multi_line_specials is on: |
|
1716 | 1718 | if continue_prompt and self.rc.multi_line_specials and \ |
|
1717 | 1719 | iFun.startswith(self.ESC_SHELL): |
|
1718 | 1720 | return self.handle_shell_escape(line,continue_prompt, |
|
1719 | 1721 | pre=pre,iFun=iFun, |
|
1720 | 1722 | theRest=theRest) |
|
1721 | 1723 | |
|
1722 | 1724 | # Let's try to find if the input line is a magic fn |
|
1723 | 1725 | oinfo = None |
|
1724 | 1726 | if hasattr(self,'magic_'+iFun): |
|
1725 | 1727 | # WARNING: _ofind uses getattr(), so it can consume generators and |
|
1726 | 1728 | # cause other side effects. |
|
1727 | 1729 | oinfo = self._ofind(iFun) # FIXME - _ofind is part of Magic |
|
1728 | 1730 | if oinfo['ismagic']: |
|
1729 | 1731 | # Be careful not to call magics when a variable assignment is |
|
1730 | 1732 | # being made (ls='hi', for example) |
|
1731 | 1733 | if self.rc.automagic and \ |
|
1732 | 1734 | (len(theRest)==0 or theRest[0] not in '!=()<>,') and \ |
|
1733 | 1735 | (self.rc.multi_line_specials or not continue_prompt): |
|
1734 | 1736 | return self.handle_magic(line,continue_prompt, |
|
1735 | 1737 | pre,iFun,theRest) |
|
1736 | 1738 | else: |
|
1737 | 1739 | return self.handle_normal(line,continue_prompt) |
|
1738 | 1740 | |
|
1739 | 1741 | # If the rest of the line begins with an (in)equality, assginment or |
|
1740 | 1742 | # function call, we should not call _ofind but simply execute it. |
|
1741 | 1743 | # This avoids spurious geattr() accesses on objects upon assignment. |
|
1742 | 1744 | # |
|
1743 | 1745 | # It also allows users to assign to either alias or magic names true |
|
1744 | 1746 | # python variables (the magic/alias systems always take second seat to |
|
1745 | 1747 | # true python code). |
|
1746 | 1748 | if theRest and theRest[0] in '!=()': |
|
1747 | 1749 | return self.handle_normal(line,continue_prompt) |
|
1748 | 1750 | |
|
1749 | 1751 | if oinfo is None: |
|
1750 | 1752 | # let's try to ensure that _oinfo is ONLY called when autocall is |
|
1751 | 1753 | # on. Since it has inevitable potential side effects, at least |
|
1752 | 1754 | # having autocall off should be a guarantee to the user that no |
|
1753 | 1755 | # weird things will happen. |
|
1754 | 1756 | |
|
1755 | 1757 | if self.rc.autocall: |
|
1756 | 1758 | oinfo = self._ofind(iFun) # FIXME - _ofind is part of Magic |
|
1757 | 1759 | else: |
|
1758 | 1760 | # in this case, all that's left is either an alias or |
|
1759 | 1761 | # processing the line normally. |
|
1760 | 1762 | if iFun in self.alias_table: |
|
1761 | 1763 | return self.handle_alias(line,continue_prompt, |
|
1762 | 1764 | pre,iFun,theRest) |
|
1763 | 1765 | else: |
|
1764 | 1766 | return self.handle_normal(line,continue_prompt) |
|
1765 | 1767 | |
|
1766 | 1768 | if not oinfo['found']: |
|
1767 | 1769 | return self.handle_normal(line,continue_prompt) |
|
1768 | 1770 | else: |
|
1769 | 1771 | #print 'iFun <%s> rest <%s>' % (iFun,theRest) # dbg |
|
1770 | 1772 | if oinfo['isalias']: |
|
1771 | 1773 | return self.handle_alias(line,continue_prompt, |
|
1772 | 1774 | pre,iFun,theRest) |
|
1773 | 1775 | |
|
1774 | 1776 | if self.rc.autocall and \ |
|
1775 | 1777 | not self.re_exclude_auto.match(theRest) and \ |
|
1776 | 1778 | self.re_fun_name.match(iFun) and \ |
|
1777 | 1779 | callable(oinfo['obj']) : |
|
1778 | 1780 | #print 'going auto' # dbg |
|
1779 | 1781 | return self.handle_auto(line,continue_prompt,pre,iFun,theRest) |
|
1780 | 1782 | else: |
|
1781 | 1783 | #print 'was callable?', callable(oinfo['obj']) # dbg |
|
1782 | 1784 | return self.handle_normal(line,continue_prompt) |
|
1783 | 1785 | |
|
1784 | 1786 | # If we get here, we have a normal Python line. Log and return. |
|
1785 | 1787 | return self.handle_normal(line,continue_prompt) |
|
1786 | 1788 | |
|
1787 | 1789 | def _prefilter_dumb(self, line, continue_prompt): |
|
1788 | 1790 | """simple prefilter function, for debugging""" |
|
1789 | 1791 | return self.handle_normal(line,continue_prompt) |
|
1790 | 1792 | |
|
1791 | 1793 | # Set the default prefilter() function (this can be user-overridden) |
|
1792 | 1794 | prefilter = _prefilter |
|
1793 | 1795 | |
|
1794 | 1796 | def handle_normal(self,line,continue_prompt=None, |
|
1795 | 1797 | pre=None,iFun=None,theRest=None): |
|
1796 | 1798 | """Handle normal input lines. Use as a template for handlers.""" |
|
1797 | 1799 | |
|
1798 | 1800 | # With autoindent on, we need some way to exit the input loop, and I |
|
1799 | 1801 | # don't want to force the user to have to backspace all the way to |
|
1800 | 1802 | # clear the line. The rule will be in this case, that either two |
|
1801 | 1803 | # lines of pure whitespace in a row, or a line of pure whitespace but |
|
1802 | 1804 | # of a size different to the indent level, will exit the input loop. |
|
1803 | 1805 | |
|
1804 | 1806 | if (continue_prompt and self.autoindent and isspace(line) and |
|
1805 | 1807 | (line != self.indent_current or isspace(self.buffer[-1]))): |
|
1806 | 1808 | line = '' |
|
1807 | 1809 | |
|
1808 | 1810 | self.log(line,continue_prompt) |
|
1809 | 1811 | return line |
|
1810 | 1812 | |
|
1811 | 1813 | def handle_alias(self,line,continue_prompt=None, |
|
1812 | 1814 | pre=None,iFun=None,theRest=None): |
|
1813 | 1815 | """Handle alias input lines. """ |
|
1814 | 1816 | |
|
1815 | line_out = 'ipalias("%s %s")' % (iFun,esc_quotes(theRest)) | |
|
1817 | # pre is needed, because it carries the leading whitespace. Otherwise | |
|
1818 | # aliases won't work in indented sections. | |
|
1819 | line_out = '%sipalias("%s %s")' % (pre,iFun,esc_quotes(theRest)) | |
|
1816 | 1820 | self.log(line_out,continue_prompt) |
|
1817 | 1821 | return line_out |
|
1818 | 1822 | |
|
1819 | 1823 | def handle_shell_escape(self, line, continue_prompt=None, |
|
1820 | 1824 | pre=None,iFun=None,theRest=None): |
|
1821 | 1825 | """Execute the line in a shell, empty return value""" |
|
1822 | 1826 | |
|
1823 | 1827 | #print 'line in :', `line` # dbg |
|
1824 | 1828 | # Example of a special handler. Others follow a similar pattern. |
|
1825 | 1829 | if continue_prompt: # multi-line statements |
|
1826 | 1830 | if iFun.startswith('!!'): |
|
1827 | 1831 | print 'SyntaxError: !! is not allowed in multiline statements' |
|
1828 | 1832 | return pre |
|
1829 | 1833 | else: |
|
1830 | 1834 | cmd = ("%s %s" % (iFun[1:],theRest)) |
|
1831 | line_out = 'ipsystem(r"""%s"""[:-1])' % (cmd + "_") | |
|
1835 | line_out = '%sipsystem(r"""%s"""[:-1])' % (pre,cmd + "_") | |
|
1832 | 1836 | else: # single-line input |
|
1833 | 1837 | if line.startswith('!!'): |
|
1834 | 1838 | # rewrite iFun/theRest to properly hold the call to %sx and |
|
1835 | 1839 | # the actual command to be executed, so handle_magic can work |
|
1836 | 1840 | # correctly |
|
1837 | 1841 | theRest = '%s %s' % (iFun[2:],theRest) |
|
1838 | 1842 | iFun = 'sx' |
|
1839 | 1843 | return self.handle_magic('%ssx %s' % (self.ESC_MAGIC,line[2:]), |
|
1840 | 1844 | continue_prompt,pre,iFun,theRest) |
|
1841 | 1845 | else: |
|
1842 | 1846 | cmd=line[1:] |
|
1843 | line_out = 'ipsystem(r"""%s"""[:-1])' % (cmd +"_") | |
|
1847 | line_out = '%sipsystem(r"""%s"""[:-1])' % (pre,cmd +"_") | |
|
1844 | 1848 | # update cache/log and return |
|
1845 | 1849 | self.log(line_out,continue_prompt) |
|
1846 | 1850 | return line_out |
|
1847 | 1851 | |
|
1848 | 1852 | def handle_magic(self, line, continue_prompt=None, |
|
1849 | 1853 | pre=None,iFun=None,theRest=None): |
|
1850 | 1854 | """Execute magic functions. |
|
1851 | 1855 | |
|
1852 | 1856 | Also log them with a prepended # so the log is clean Python.""" |
|
1853 | 1857 | |
|
1854 | 1858 | cmd = '%sipmagic("%s")' % (pre,esc_quotes('%s %s' % (iFun,theRest))) |
|
1855 | 1859 | self.log(cmd,continue_prompt) |
|
1856 | 1860 | #print 'in handle_magic, cmd=<%s>' % cmd # dbg |
|
1857 | 1861 | return cmd |
|
1858 | 1862 | |
|
1859 | 1863 | def handle_auto(self, line, continue_prompt=None, |
|
1860 | 1864 | pre=None,iFun=None,theRest=None): |
|
1861 | 1865 | """Hande lines which can be auto-executed, quoting if requested.""" |
|
1862 | 1866 | |
|
1863 | 1867 | #print 'pre <%s> iFun <%s> rest <%s>' % (pre,iFun,theRest) # dbg |
|
1864 | 1868 | |
|
1865 | 1869 | # This should only be active for single-line input! |
|
1866 | 1870 | if continue_prompt: |
|
1867 | 1871 | return line |
|
1868 | 1872 | |
|
1869 | 1873 | if pre == self.ESC_QUOTE: |
|
1870 | 1874 | # Auto-quote splitting on whitespace |
|
1871 | 1875 | newcmd = '%s("%s")' % (iFun,'", "'.join(theRest.split()) ) |
|
1872 | 1876 | elif pre == self.ESC_QUOTE2: |
|
1873 | 1877 | # Auto-quote whole string |
|
1874 | 1878 | newcmd = '%s("%s")' % (iFun,theRest) |
|
1875 | 1879 | else: |
|
1876 | 1880 | # Auto-paren |
|
1877 | 1881 | if theRest[0:1] in ('=','['): |
|
1878 | 1882 | # Don't autocall in these cases. They can be either |
|
1879 | 1883 | # rebindings of an existing callable's name, or item access |
|
1880 | 1884 | # for an object which is BOTH callable and implements |
|
1881 | 1885 | # __getitem__. |
|
1882 | 1886 | return '%s %s' % (iFun,theRest) |
|
1883 | 1887 | if theRest.endswith(';'): |
|
1884 | 1888 | newcmd = '%s(%s);' % (iFun.rstrip(),theRest[:-1]) |
|
1885 | 1889 | else: |
|
1886 | 1890 | newcmd = '%s(%s)' % (iFun.rstrip(),theRest) |
|
1887 | 1891 | |
|
1888 | 1892 | print >>Term.cout, self.outputcache.prompt1.auto_rewrite() + newcmd |
|
1889 | 1893 | # log what is now valid Python, not the actual user input (without the |
|
1890 | 1894 | # final newline) |
|
1891 | 1895 | self.log(newcmd,continue_prompt) |
|
1892 | 1896 | return newcmd |
|
1893 | 1897 | |
|
1894 | 1898 | def handle_help(self, line, continue_prompt=None, |
|
1895 | 1899 | pre=None,iFun=None,theRest=None): |
|
1896 | 1900 | """Try to get some help for the object. |
|
1897 | 1901 | |
|
1898 | 1902 | obj? or ?obj -> basic information. |
|
1899 | 1903 | obj?? or ??obj -> more details. |
|
1900 | 1904 | """ |
|
1901 | 1905 | |
|
1902 | 1906 | # We need to make sure that we don't process lines which would be |
|
1903 | 1907 | # otherwise valid python, such as "x=1 # what?" |
|
1904 | 1908 | try: |
|
1905 | 1909 | codeop.compile_command(line) |
|
1906 | 1910 | except SyntaxError: |
|
1907 | 1911 | # We should only handle as help stuff which is NOT valid syntax |
|
1908 | 1912 | if line[0]==self.ESC_HELP: |
|
1909 | 1913 | line = line[1:] |
|
1910 | 1914 | elif line[-1]==self.ESC_HELP: |
|
1911 | 1915 | line = line[:-1] |
|
1912 | 1916 | self.log('#?'+line) |
|
1913 | 1917 | if line: |
|
1914 | 1918 | self.magic_pinfo(line) |
|
1915 | 1919 | else: |
|
1916 | 1920 | page(self.usage,screen_lines=self.rc.screen_length) |
|
1917 | 1921 | return '' # Empty string is needed here! |
|
1918 | 1922 | except: |
|
1919 | 1923 | # Pass any other exceptions through to the normal handler |
|
1920 | 1924 | return self.handle_normal(line,continue_prompt) |
|
1921 | 1925 | else: |
|
1922 | 1926 | # If the code compiles ok, we should handle it normally |
|
1923 | 1927 | return self.handle_normal(line,continue_prompt) |
|
1924 | 1928 | |
|
1925 | 1929 | def handle_emacs(self,line,continue_prompt=None, |
|
1926 | 1930 | pre=None,iFun=None,theRest=None): |
|
1927 | 1931 | """Handle input lines marked by python-mode.""" |
|
1928 | 1932 | |
|
1929 | 1933 | # Currently, nothing is done. Later more functionality can be added |
|
1930 | 1934 | # here if needed. |
|
1931 | 1935 | |
|
1932 | 1936 | # The input cache shouldn't be updated |
|
1933 | 1937 | |
|
1934 | 1938 | return line |
|
1935 | 1939 | |
|
1936 | 1940 | def write(self,data): |
|
1937 | 1941 | """Write a string to the default output""" |
|
1938 | 1942 | Term.cout.write(data) |
|
1939 | 1943 | |
|
1940 | 1944 | def write_err(self,data): |
|
1941 | 1945 | """Write a string to the default error output""" |
|
1942 | 1946 | Term.cerr.write(data) |
|
1943 | 1947 | |
|
1944 | 1948 | def exit(self): |
|
1945 | 1949 | """Handle interactive exit. |
|
1946 | 1950 | |
|
1947 | 1951 | This method sets the exit_now attribute.""" |
|
1948 | 1952 | |
|
1949 | 1953 | if self.rc.confirm_exit: |
|
1950 | 1954 | if ask_yes_no('Do you really want to exit ([y]/n)?','y'): |
|
1951 | 1955 | self.exit_now = True |
|
1952 | 1956 | else: |
|
1953 | 1957 | self.exit_now = True |
|
1954 | 1958 | return self.exit_now |
|
1955 | 1959 | |
|
1956 | 1960 | def safe_execfile(self,fname,*where,**kw): |
|
1957 | 1961 | fname = os.path.expanduser(fname) |
|
1958 | 1962 | |
|
1959 | 1963 | # find things also in current directory |
|
1960 | 1964 | dname = os.path.dirname(fname) |
|
1961 | 1965 | if not sys.path.count(dname): |
|
1962 | 1966 | sys.path.append(dname) |
|
1963 | 1967 | |
|
1964 | 1968 | try: |
|
1965 | 1969 | xfile = open(fname) |
|
1966 | 1970 | except: |
|
1967 | 1971 | print >> Term.cerr, \ |
|
1968 | 1972 | 'Could not open file <%s> for safe execution.' % fname |
|
1969 | 1973 | return None |
|
1970 | 1974 | |
|
1971 | 1975 | kw.setdefault('islog',0) |
|
1972 | 1976 | kw.setdefault('quiet',1) |
|
1973 | 1977 | kw.setdefault('exit_ignore',0) |
|
1974 | 1978 | first = xfile.readline() |
|
1975 | 1979 | loghead = str(self.loghead_tpl).split('\n',1)[0].strip() |
|
1976 | 1980 | xfile.close() |
|
1977 | 1981 | # line by line execution |
|
1978 | 1982 | if first.startswith(loghead) or kw['islog']: |
|
1979 | 1983 | print 'Loading log file <%s> one line at a time...' % fname |
|
1980 | 1984 | if kw['quiet']: |
|
1981 | 1985 | stdout_save = sys.stdout |
|
1982 | 1986 | sys.stdout = StringIO.StringIO() |
|
1983 | 1987 | try: |
|
1984 | 1988 | globs,locs = where[0:2] |
|
1985 | 1989 | except: |
|
1986 | 1990 | try: |
|
1987 | 1991 | globs = locs = where[0] |
|
1988 | 1992 | except: |
|
1989 | 1993 | globs = locs = globals() |
|
1990 | 1994 | badblocks = [] |
|
1991 | 1995 | |
|
1992 | 1996 | # we also need to identify indented blocks of code when replaying |
|
1993 | 1997 | # logs and put them together before passing them to an exec |
|
1994 | 1998 | # statement. This takes a bit of regexp and look-ahead work in the |
|
1995 | 1999 | # file. It's easiest if we swallow the whole thing in memory |
|
1996 | 2000 | # first, and manually walk through the lines list moving the |
|
1997 | 2001 | # counter ourselves. |
|
1998 | 2002 | indent_re = re.compile('\s+\S') |
|
1999 | 2003 | xfile = open(fname) |
|
2000 | 2004 | filelines = xfile.readlines() |
|
2001 | 2005 | xfile.close() |
|
2002 | 2006 | nlines = len(filelines) |
|
2003 | 2007 | lnum = 0 |
|
2004 | 2008 | while lnum < nlines: |
|
2005 | 2009 | line = filelines[lnum] |
|
2006 | 2010 | lnum += 1 |
|
2007 | 2011 | # don't re-insert logger status info into cache |
|
2008 | 2012 | if line.startswith('#log#'): |
|
2009 | 2013 | continue |
|
2010 | 2014 | else: |
|
2011 | 2015 | # build a block of code (maybe a single line) for execution |
|
2012 | 2016 | block = line |
|
2013 | 2017 | try: |
|
2014 | 2018 | next = filelines[lnum] # lnum has already incremented |
|
2015 | 2019 | except: |
|
2016 | 2020 | next = None |
|
2017 | 2021 | while next and indent_re.match(next): |
|
2018 | 2022 | block += next |
|
2019 | 2023 | lnum += 1 |
|
2020 | 2024 | try: |
|
2021 | 2025 | next = filelines[lnum] |
|
2022 | 2026 | except: |
|
2023 | 2027 | next = None |
|
2024 | 2028 | # now execute the block of one or more lines |
|
2025 | 2029 | try: |
|
2026 | 2030 | exec block in globs,locs |
|
2027 | 2031 | except SystemExit: |
|
2028 | 2032 | pass |
|
2029 | 2033 | except: |
|
2030 | 2034 | badblocks.append(block.rstrip()) |
|
2031 | 2035 | if kw['quiet']: # restore stdout |
|
2032 | 2036 | sys.stdout.close() |
|
2033 | 2037 | sys.stdout = stdout_save |
|
2034 | 2038 | print 'Finished replaying log file <%s>' % fname |
|
2035 | 2039 | if badblocks: |
|
2036 | 2040 | print >> sys.stderr, ('\nThe following lines/blocks in file ' |
|
2037 | 2041 | '<%s> reported errors:' % fname) |
|
2038 | 2042 | |
|
2039 | 2043 | for badline in badblocks: |
|
2040 | 2044 | print >> sys.stderr, badline |
|
2041 | 2045 | else: # regular file execution |
|
2042 | 2046 | try: |
|
2043 | 2047 | execfile(fname,*where) |
|
2044 | 2048 | except SyntaxError: |
|
2045 | 2049 | etype,evalue = sys.exc_info()[:2] |
|
2046 | 2050 | self.SyntaxTB(etype,evalue,[]) |
|
2047 | 2051 | warn('Failure executing file: <%s>' % fname) |
|
2048 | 2052 | except SystemExit,status: |
|
2049 | 2053 | if not kw['exit_ignore']: |
|
2050 | 2054 | self.InteractiveTB() |
|
2051 | 2055 | warn('Failure executing file: <%s>' % fname) |
|
2052 | 2056 | except: |
|
2053 | 2057 | self.InteractiveTB() |
|
2054 | 2058 | warn('Failure executing file: <%s>' % fname) |
|
2055 | 2059 | |
|
2056 | 2060 | #************************* end of file <iplib.py> ***************************** |
@@ -1,4678 +1,4682 b'' | |||
|
1 | 1 | 2005-12-30 Fernando Perez <Fernando.Perez@colorado.edu> |
|
2 | 2 | |
|
3 | * IPython/iplib.py (handle_shell_escape): fix bug where both | |
|
4 | aliases and system accesses where broken for indented code (such | |
|
5 | as loops). | |
|
6 | ||
|
3 | 7 | * IPython/genutils.py (shell): fix small but critical bug for |
|
4 | 8 | win32 system access. |
|
5 | 9 | |
|
6 | 10 | 2005-12-29 Fernando Perez <Fernando.Perez@colorado.edu> |
|
7 | 11 | |
|
8 | 12 | * IPython/iplib.py (showtraceback): remove use of the |
|
9 | 13 | sys.last_{type/value/traceback} structures, which are non |
|
10 | 14 | thread-safe. |
|
11 | 15 | (_prefilter): change control flow to ensure that we NEVER |
|
12 | 16 | introspect objects when autocall is off. This will guarantee that |
|
13 | 17 | having an input line of the form 'x.y', where access to attribute |
|
14 | 18 | 'y' has side effects, doesn't trigger the side effect TWICE. It |
|
15 | 19 | is important to note that, with autocall on, these side effects |
|
16 | 20 | can still happen. |
|
17 | 21 | (ipsystem): new builtin, to complete the ip{magic/alias/system} |
|
18 | 22 | trio. IPython offers these three kinds of special calls which are |
|
19 | 23 | not python code, and it's a good thing to have their call method |
|
20 | 24 | be accessible as pure python functions (not just special syntax at |
|
21 | 25 | the command line). It gives us a better internal implementation |
|
22 | 26 | structure, as well as exposing these for user scripting more |
|
23 | 27 | cleanly. |
|
24 | 28 | |
|
25 | 29 | * IPython/macro.py (Macro.__init__): moved macros to a standalone |
|
26 | 30 | file. Now that they'll be more likely to be used with the |
|
27 | 31 | persistance system (%store), I want to make sure their module path |
|
28 | 32 | doesn't change in the future, so that we don't break things for |
|
29 | 33 | users' persisted data. |
|
30 | 34 | |
|
31 | 35 | * IPython/iplib.py (autoindent_update): move indentation |
|
32 | 36 | management into the _text_ processing loop, not the keyboard |
|
33 | 37 | interactive one. This is necessary to correctly process non-typed |
|
34 | 38 | multiline input (such as macros). |
|
35 | 39 | |
|
36 | 40 | * IPython/Magic.py (Magic.format_latex): patch by Stefan van der |
|
37 | 41 | Walt <stefan-AT-sun.ac.za> to fix latex formatting of docstrings, |
|
38 | 42 | which was producing problems in the resulting manual. |
|
39 | 43 | (magic_whos): improve reporting of instances (show their class, |
|
40 | 44 | instead of simply printing 'instance' which isn't terribly |
|
41 | 45 | informative). |
|
42 | 46 | |
|
43 | 47 | * IPython/genutils.py (shell): commit Jorgen Stenarson's patch |
|
44 | 48 | (minor mods) to support network shares under win32. |
|
45 | 49 | |
|
46 | 50 | * IPython/winconsole.py (get_console_size): add new winconsole |
|
47 | 51 | module and fixes to page_dumb() to improve its behavior under |
|
48 | 52 | win32. Contributed by Alexander Belchenko <bialix-AT-ukr.net>. |
|
49 | 53 | |
|
50 | 54 | * IPython/Magic.py (Macro): simplified Macro class to just |
|
51 | 55 | subclass list. We've had only 2.2 compatibility for a very long |
|
52 | 56 | time, yet I was still avoiding subclassing the builtin types. No |
|
53 | 57 | more (I'm also starting to use properties, though I won't shift to |
|
54 | 58 | 2.3-specific features quite yet). |
|
55 | 59 | (magic_store): added Ville's patch for lightweight variable |
|
56 | 60 | persistence, after a request on the user list by Matt Wilkie |
|
57 | 61 | <maphew-AT-gmail.com>. The new %store magic's docstring has full |
|
58 | 62 | details. |
|
59 | 63 | |
|
60 | 64 | * IPython/iplib.py (InteractiveShell.post_config_initialization): |
|
61 | 65 | changed the default logfile name from 'ipython.log' to |
|
62 | 66 | 'ipython_log.py'. These logs are real python files, and now that |
|
63 | 67 | we have much better multiline support, people are more likely to |
|
64 | 68 | want to use them as such. Might as well name them correctly. |
|
65 | 69 | |
|
66 | 70 | * IPython/Magic.py: substantial cleanup. While we can't stop |
|
67 | 71 | using magics as mixins, due to the existing customizations 'out |
|
68 | 72 | there' which rely on the mixin naming conventions, at least I |
|
69 | 73 | cleaned out all cross-class name usage. So once we are OK with |
|
70 | 74 | breaking compatibility, the two systems can be separated. |
|
71 | 75 | |
|
72 | 76 | * IPython/Logger.py: major cleanup. This one is NOT a mixin |
|
73 | 77 | anymore, and the class is a fair bit less hideous as well. New |
|
74 | 78 | features were also introduced: timestamping of input, and logging |
|
75 | 79 | of output results. These are user-visible with the -t and -o |
|
76 | 80 | options to %logstart. Closes |
|
77 | 81 | http://www.scipy.net/roundup/ipython/issue11 and a request by |
|
78 | 82 | William Stein (SAGE developer - http://modular.ucsd.edu/sage). |
|
79 | 83 | |
|
80 | 84 | 2005-12-28 Fernando Perez <Fernando.Perez@colorado.edu> |
|
81 | 85 | |
|
82 | 86 | * IPython/iplib.py (handle_shell_escape): add Ville's patch to |
|
83 | 87 | better hadnle backslashes in paths. See the thread 'More Windows |
|
84 | 88 | questions part 2 - \/ characters revisited' on the iypthon user |
|
85 | 89 | list: |
|
86 | 90 | http://scipy.net/pipermail/ipython-user/2005-June/000907.html |
|
87 | 91 | |
|
88 | 92 | (InteractiveShell.__init__): fix tab-completion bug in threaded shells. |
|
89 | 93 | |
|
90 | 94 | (InteractiveShell.__init__): change threaded shells to not use the |
|
91 | 95 | ipython crash handler. This was causing more problems than not, |
|
92 | 96 | as exceptions in the main thread (GUI code, typically) would |
|
93 | 97 | always show up as a 'crash', when they really weren't. |
|
94 | 98 | |
|
95 | 99 | The colors and exception mode commands (%colors/%xmode) have been |
|
96 | 100 | synchronized to also take this into account, so users can get |
|
97 | 101 | verbose exceptions for their threaded code as well. I also added |
|
98 | 102 | support for activating pdb inside this exception handler as well, |
|
99 | 103 | so now GUI authors can use IPython's enhanced pdb at runtime. |
|
100 | 104 | |
|
101 | 105 | * IPython/ipmaker.py (make_IPython): make the autoedit_syntax flag |
|
102 | 106 | true by default, and add it to the shipped ipythonrc file. Since |
|
103 | 107 | this asks the user before proceeding, I think it's OK to make it |
|
104 | 108 | true by default. |
|
105 | 109 | |
|
106 | 110 | * IPython/Magic.py (magic_exit): make new exit/quit magics instead |
|
107 | 111 | of the previous special-casing of input in the eval loop. I think |
|
108 | 112 | this is cleaner, as they really are commands and shouldn't have |
|
109 | 113 | a special role in the middle of the core code. |
|
110 | 114 | |
|
111 | 115 | 2005-12-27 Fernando Perez <Fernando.Perez@colorado.edu> |
|
112 | 116 | |
|
113 | 117 | * IPython/iplib.py (edit_syntax_error): added support for |
|
114 | 118 | automatically reopening the editor if the file had a syntax error |
|
115 | 119 | in it. Thanks to scottt who provided the patch at: |
|
116 | 120 | http://www.scipy.net/roundup/ipython/issue36 (slightly modified |
|
117 | 121 | version committed). |
|
118 | 122 | |
|
119 | 123 | * IPython/iplib.py (handle_normal): add suport for multi-line |
|
120 | 124 | input with emtpy lines. This fixes |
|
121 | 125 | http://www.scipy.net/roundup/ipython/issue43 and a similar |
|
122 | 126 | discussion on the user list. |
|
123 | 127 | |
|
124 | 128 | WARNING: a behavior change is necessarily introduced to support |
|
125 | 129 | blank lines: now a single blank line with whitespace does NOT |
|
126 | 130 | break the input loop, which means that when autoindent is on, by |
|
127 | 131 | default hitting return on the next (indented) line does NOT exit. |
|
128 | 132 | |
|
129 | 133 | Instead, to exit a multiline input you can either have: |
|
130 | 134 | |
|
131 | 135 | - TWO whitespace lines (just hit return again), or |
|
132 | 136 | - a single whitespace line of a different length than provided |
|
133 | 137 | by the autoindent (add or remove a space). |
|
134 | 138 | |
|
135 | 139 | * IPython/completer.py (MagicCompleter.__init__): new 'completer' |
|
136 | 140 | module to better organize all readline-related functionality. |
|
137 | 141 | I've deleted FlexCompleter and put all completion clases here. |
|
138 | 142 | |
|
139 | 143 | * IPython/iplib.py (raw_input): improve indentation management. |
|
140 | 144 | It is now possible to paste indented code with autoindent on, and |
|
141 | 145 | the code is interpreted correctly (though it still looks bad on |
|
142 | 146 | screen, due to the line-oriented nature of ipython). |
|
143 | 147 | (MagicCompleter.complete): change behavior so that a TAB key on an |
|
144 | 148 | otherwise empty line actually inserts a tab, instead of completing |
|
145 | 149 | on the entire global namespace. This makes it easier to use the |
|
146 | 150 | TAB key for indentation. After a request by Hans Meine |
|
147 | 151 | <hans_meine-AT-gmx.net> |
|
148 | 152 | (_prefilter): add support so that typing plain 'exit' or 'quit' |
|
149 | 153 | does a sensible thing. Originally I tried to deviate as little as |
|
150 | 154 | possible from the default python behavior, but even that one may |
|
151 | 155 | change in this direction (thread on python-dev to that effect). |
|
152 | 156 | Regardless, ipython should do the right thing even if CPython's |
|
153 | 157 | '>>>' prompt doesn't. |
|
154 | 158 | (InteractiveShell): removed subclassing code.InteractiveConsole |
|
155 | 159 | class. By now we'd overridden just about all of its methods: I've |
|
156 | 160 | copied the remaining two over, and now ipython is a standalone |
|
157 | 161 | class. This will provide a clearer picture for the chainsaw |
|
158 | 162 | branch refactoring. |
|
159 | 163 | |
|
160 | 164 | 2005-12-26 Fernando Perez <Fernando.Perez@colorado.edu> |
|
161 | 165 | |
|
162 | 166 | * IPython/ultraTB.py (VerboseTB.text): harden reporting against |
|
163 | 167 | failures for objects which break when dir() is called on them. |
|
164 | 168 | |
|
165 | 169 | * IPython/FlexCompleter.py (Completer.__init__): Added support for |
|
166 | 170 | distinct local and global namespaces in the completer API. This |
|
167 | 171 | change allows us top properly handle completion with distinct |
|
168 | 172 | scopes, including in embedded instances (this had never really |
|
169 | 173 | worked correctly). |
|
170 | 174 | |
|
171 | 175 | Note: this introduces a change in the constructor for |
|
172 | 176 | MagicCompleter, as a new global_namespace parameter is now the |
|
173 | 177 | second argument (the others were bumped one position). |
|
174 | 178 | |
|
175 | 179 | 2005-12-25 Fernando Perez <Fernando.Perez@colorado.edu> |
|
176 | 180 | |
|
177 | 181 | * IPython/iplib.py (embed_mainloop): fix tab-completion in |
|
178 | 182 | embedded instances (which can be done now thanks to Vivian's |
|
179 | 183 | frame-handling fixes for pdb). |
|
180 | 184 | (InteractiveShell.__init__): Fix namespace handling problem in |
|
181 | 185 | embedded instances. We were overwriting __main__ unconditionally, |
|
182 | 186 | and this should only be done for 'full' (non-embedded) IPython; |
|
183 | 187 | embedded instances must respect the caller's __main__. Thanks to |
|
184 | 188 | a bug report by Yaroslav Bulatov <yaroslavvb-AT-gmail.com> |
|
185 | 189 | |
|
186 | 190 | 2005-12-24 Fernando Perez <Fernando.Perez@colorado.edu> |
|
187 | 191 | |
|
188 | 192 | * setup.py: added download_url to setup(). This registers the |
|
189 | 193 | download address at PyPI, which is not only useful to humans |
|
190 | 194 | browsing the site, but is also picked up by setuptools (the Eggs |
|
191 | 195 | machinery). Thanks to Ville and R. Kern for the info/discussion |
|
192 | 196 | on this. |
|
193 | 197 | |
|
194 | 198 | 2005-12-23 Fernando Perez <Fernando.Perez@colorado.edu> |
|
195 | 199 | |
|
196 | 200 | * IPython/Debugger.py (Pdb.__init__): Major pdb mode enhancements. |
|
197 | 201 | This brings a lot of nice functionality to the pdb mode, which now |
|
198 | 202 | has tab-completion, syntax highlighting, and better stack handling |
|
199 | 203 | than before. Many thanks to Vivian De Smedt |
|
200 | 204 | <vivian-AT-vdesmedt.com> for the original patches. |
|
201 | 205 | |
|
202 | 206 | 2005-12-08 Fernando Perez <Fernando.Perez@colorado.edu> |
|
203 | 207 | |
|
204 | 208 | * IPython/Shell.py (IPShellGTK.mainloop): fix mainloop() calling |
|
205 | 209 | sequence to consistently accept the banner argument. The |
|
206 | 210 | inconsistency was tripping SAGE, thanks to Gary Zablackis |
|
207 | 211 | <gzabl-AT-yahoo.com> for the report. |
|
208 | 212 | |
|
209 | 213 | 2005-11-15 Fernando Perez <Fernando.Perez@colorado.edu> |
|
210 | 214 | |
|
211 | 215 | * IPython/iplib.py (InteractiveShell.post_config_initialization): |
|
212 | 216 | Fix bug where a naked 'alias' call in the ipythonrc file would |
|
213 | 217 | cause a crash. Bug reported by Jorgen Stenarson. |
|
214 | 218 | |
|
215 | 219 | 2005-11-15 Fernando Perez <Fernando.Perez@colorado.edu> |
|
216 | 220 | |
|
217 | 221 | * IPython/ipmaker.py (make_IPython): cleanups which should improve |
|
218 | 222 | startup time. |
|
219 | 223 | |
|
220 | 224 | * IPython/iplib.py (runcode): my globals 'fix' for embedded |
|
221 | 225 | instances had introduced a bug with globals in normal code. Now |
|
222 | 226 | it's working in all cases. |
|
223 | 227 | |
|
224 | 228 | * IPython/Magic.py (magic_psearch): Finish wildcard cleanup and |
|
225 | 229 | API changes. A new ipytonrc option, 'wildcards_case_sensitive' |
|
226 | 230 | has been introduced to set the default case sensitivity of the |
|
227 | 231 | searches. Users can still select either mode at runtime on a |
|
228 | 232 | per-search basis. |
|
229 | 233 | |
|
230 | 234 | 2005-11-13 Fernando Perez <Fernando.Perez@colorado.edu> |
|
231 | 235 | |
|
232 | 236 | * IPython/wildcard.py (NameSpace.__init__): fix resolution of |
|
233 | 237 | attributes in wildcard searches for subclasses. Modified version |
|
234 | 238 | of a patch by Jorgen. |
|
235 | 239 | |
|
236 | 240 | 2005-11-12 Fernando Perez <Fernando.Perez@colorado.edu> |
|
237 | 241 | |
|
238 | 242 | * IPython/iplib.py (embed_mainloop): Fix handling of globals for |
|
239 | 243 | embedded instances. I added a user_global_ns attribute to the |
|
240 | 244 | InteractiveShell class to handle this. |
|
241 | 245 | |
|
242 | 246 | 2005-10-31 Fernando Perez <Fernando.Perez@colorado.edu> |
|
243 | 247 | |
|
244 | 248 | * IPython/Shell.py (IPShellGTK.mainloop): Change timeout_add to |
|
245 | 249 | idle_add, which fixes horrible keyboard lag problems under gtk 2.6 |
|
246 | 250 | (reported under win32, but may happen also in other platforms). |
|
247 | 251 | Bug report and fix courtesy of Sean Moore <smm-AT-logic.bm> |
|
248 | 252 | |
|
249 | 253 | 2005-10-15 Fernando Perez <Fernando.Perez@colorado.edu> |
|
250 | 254 | |
|
251 | 255 | * IPython/Magic.py (magic_psearch): new support for wildcard |
|
252 | 256 | patterns. Now, typing ?a*b will list all names which begin with a |
|
253 | 257 | and end in b, for example. The %psearch magic has full |
|
254 | 258 | docstrings. Many thanks to JΓΆrgen Stenarson |
|
255 | 259 | <jorgen.stenarson-AT-bostream.nu>, author of the patches |
|
256 | 260 | implementing this functionality. |
|
257 | 261 | |
|
258 | 262 | 2005-09-27 Fernando Perez <Fernando.Perez@colorado.edu> |
|
259 | 263 | |
|
260 | 264 | * Manual: fixed long-standing annoyance of double-dashes (as in |
|
261 | 265 | --prefix=~, for example) being stripped in the HTML version. This |
|
262 | 266 | is a latex2html bug, but a workaround was provided. Many thanks |
|
263 | 267 | to George K. Thiruvathukal <gthiruv-AT-luc.edu> for the detailed |
|
264 | 268 | help, and Michael Tobis <mtobis-AT-gmail.com> for getting the ball |
|
265 | 269 | rolling. This seemingly small issue had tripped a number of users |
|
266 | 270 | when first installing, so I'm glad to see it gone. |
|
267 | 271 | |
|
268 | 272 | 2005-09-27 Fernando Perez <Fernando.Perez@colorado.edu> |
|
269 | 273 | |
|
270 | 274 | * IPython/Extensions/numeric_formats.py: fix missing import, |
|
271 | 275 | reported by Stephen Walton. |
|
272 | 276 | |
|
273 | 277 | 2005-09-24 Fernando Perez <Fernando.Perez@colorado.edu> |
|
274 | 278 | |
|
275 | 279 | * IPython/demo.py: finish demo module, fully documented now. |
|
276 | 280 | |
|
277 | 281 | * IPython/genutils.py (file_read): simple little utility to read a |
|
278 | 282 | file and ensure it's closed afterwards. |
|
279 | 283 | |
|
280 | 284 | 2005-09-23 Fernando Perez <Fernando.Perez@colorado.edu> |
|
281 | 285 | |
|
282 | 286 | * IPython/demo.py (Demo.__init__): added support for individually |
|
283 | 287 | tagging blocks for automatic execution. |
|
284 | 288 | |
|
285 | 289 | * IPython/Magic.py (magic_pycat): new %pycat magic for showing |
|
286 | 290 | syntax-highlighted python sources, requested by John. |
|
287 | 291 | |
|
288 | 292 | 2005-09-22 Fernando Perez <Fernando.Perez@colorado.edu> |
|
289 | 293 | |
|
290 | 294 | * IPython/demo.py (Demo.again): fix bug where again() blocks after |
|
291 | 295 | finishing. |
|
292 | 296 | |
|
293 | 297 | * IPython/genutils.py (shlex_split): moved from Magic to here, |
|
294 | 298 | where all 2.2 compatibility stuff lives. I needed it for demo.py. |
|
295 | 299 | |
|
296 | 300 | * IPython/demo.py (Demo.__init__): added support for silent |
|
297 | 301 | blocks, improved marks as regexps, docstrings written. |
|
298 | 302 | (Demo.__init__): better docstring, added support for sys.argv. |
|
299 | 303 | |
|
300 | 304 | * IPython/genutils.py (marquee): little utility used by the demo |
|
301 | 305 | code, handy in general. |
|
302 | 306 | |
|
303 | 307 | * IPython/demo.py (Demo.__init__): new class for interactive |
|
304 | 308 | demos. Not documented yet, I just wrote it in a hurry for |
|
305 | 309 | scipy'05. Will docstring later. |
|
306 | 310 | |
|
307 | 311 | 2005-09-20 Fernando Perez <Fernando.Perez@colorado.edu> |
|
308 | 312 | |
|
309 | 313 | * IPython/Shell.py (sigint_handler): Drastic simplification which |
|
310 | 314 | also seems to make Ctrl-C work correctly across threads! This is |
|
311 | 315 | so simple, that I can't beleive I'd missed it before. Needs more |
|
312 | 316 | testing, though. |
|
313 | 317 | (KBINT): Never mind, revert changes. I'm sure I'd tried something |
|
314 | 318 | like this before... |
|
315 | 319 | |
|
316 | 320 | * IPython/genutils.py (get_home_dir): add protection against |
|
317 | 321 | non-dirs in win32 registry. |
|
318 | 322 | |
|
319 | 323 | * IPython/iplib.py (InteractiveShell.alias_table_validate): fix |
|
320 | 324 | bug where dict was mutated while iterating (pysh crash). |
|
321 | 325 | |
|
322 | 326 | 2005-09-06 Fernando Perez <Fernando.Perez@colorado.edu> |
|
323 | 327 | |
|
324 | 328 | * IPython/iplib.py (handle_auto): Fix inconsistency arising from |
|
325 | 329 | spurious newlines added by this routine. After a report by |
|
326 | 330 | F. Mantegazza. |
|
327 | 331 | |
|
328 | 332 | 2005-09-05 Fernando Perez <Fernando.Perez@colorado.edu> |
|
329 | 333 | |
|
330 | 334 | * IPython/Shell.py (hijack_gtk): remove pygtk.require("2.0") |
|
331 | 335 | calls. These were a leftover from the GTK 1.x days, and can cause |
|
332 | 336 | problems in certain cases (after a report by John Hunter). |
|
333 | 337 | |
|
334 | 338 | * IPython/iplib.py (InteractiveShell.__init__): Trap exception if |
|
335 | 339 | os.getcwd() fails at init time. Thanks to patch from David Remahl |
|
336 | 340 | <chmod007-AT-mac.com>. |
|
337 | 341 | (InteractiveShell.__init__): prevent certain special magics from |
|
338 | 342 | being shadowed by aliases. Closes |
|
339 | 343 | http://www.scipy.net/roundup/ipython/issue41. |
|
340 | 344 | |
|
341 | 345 | 2005-08-31 Fernando Perez <Fernando.Perez@colorado.edu> |
|
342 | 346 | |
|
343 | 347 | * IPython/iplib.py (InteractiveShell.complete): Added new |
|
344 | 348 | top-level completion method to expose the completion mechanism |
|
345 | 349 | beyond readline-based environments. |
|
346 | 350 | |
|
347 | 351 | 2005-08-19 Fernando Perez <Fernando.Perez@colorado.edu> |
|
348 | 352 | |
|
349 | 353 | * tools/ipsvnc (svnversion): fix svnversion capture. |
|
350 | 354 | |
|
351 | 355 | * IPython/iplib.py (InteractiveShell.__init__): Add has_readline |
|
352 | 356 | attribute to self, which was missing. Before, it was set by a |
|
353 | 357 | routine which in certain cases wasn't being called, so the |
|
354 | 358 | instance could end up missing the attribute. This caused a crash. |
|
355 | 359 | Closes http://www.scipy.net/roundup/ipython/issue40. |
|
356 | 360 | |
|
357 | 361 | 2005-08-16 Fernando Perez <fperez@colorado.edu> |
|
358 | 362 | |
|
359 | 363 | * IPython/ultraTB.py (VerboseTB.text): don't crash if object |
|
360 | 364 | contains non-string attribute. Closes |
|
361 | 365 | http://www.scipy.net/roundup/ipython/issue38. |
|
362 | 366 | |
|
363 | 367 | 2005-08-14 Fernando Perez <fperez@colorado.edu> |
|
364 | 368 | |
|
365 | 369 | * tools/ipsvnc: Minor improvements, to add changeset info. |
|
366 | 370 | |
|
367 | 371 | 2005-08-12 Fernando Perez <fperez@colorado.edu> |
|
368 | 372 | |
|
369 | 373 | * IPython/iplib.py (runsource): remove self.code_to_run_src |
|
370 | 374 | attribute. I realized this is nothing more than |
|
371 | 375 | '\n'.join(self.buffer), and having the same data in two different |
|
372 | 376 | places is just asking for synchronization bugs. This may impact |
|
373 | 377 | people who have custom exception handlers, so I need to warn |
|
374 | 378 | ipython-dev about it (F. Mantegazza may use them). |
|
375 | 379 | |
|
376 | 380 | 2005-07-29 Fernando Perez <Fernando.Perez@colorado.edu> |
|
377 | 381 | |
|
378 | 382 | * IPython/genutils.py: fix 2.2 compatibility (generators) |
|
379 | 383 | |
|
380 | 384 | 2005-07-18 Fernando Perez <fperez@colorado.edu> |
|
381 | 385 | |
|
382 | 386 | * IPython/genutils.py (get_home_dir): fix to help users with |
|
383 | 387 | invalid $HOME under win32. |
|
384 | 388 | |
|
385 | 389 | 2005-07-17 Fernando Perez <fperez@colorado.edu> |
|
386 | 390 | |
|
387 | 391 | * IPython/Prompts.py (str_safe): Make unicode-safe. Also remove |
|
388 | 392 | some old hacks and clean up a bit other routines; code should be |
|
389 | 393 | simpler and a bit faster. |
|
390 | 394 | |
|
391 | 395 | * IPython/iplib.py (interact): removed some last-resort attempts |
|
392 | 396 | to survive broken stdout/stderr. That code was only making it |
|
393 | 397 | harder to abstract out the i/o (necessary for gui integration), |
|
394 | 398 | and the crashes it could prevent were extremely rare in practice |
|
395 | 399 | (besides being fully user-induced in a pretty violent manner). |
|
396 | 400 | |
|
397 | 401 | * IPython/genutils.py (IOStream.__init__): Simplify the i/o stuff. |
|
398 | 402 | Nothing major yet, but the code is simpler to read; this should |
|
399 | 403 | make it easier to do more serious modifications in the future. |
|
400 | 404 | |
|
401 | 405 | * IPython/Extensions/InterpreterExec.py: Fix auto-quoting in pysh, |
|
402 | 406 | which broke in .15 (thanks to a report by Ville). |
|
403 | 407 | |
|
404 | 408 | * IPython/Itpl.py (Itpl.__init__): add unicode support (it may not |
|
405 | 409 | be quite correct, I know next to nothing about unicode). This |
|
406 | 410 | will allow unicode strings to be used in prompts, amongst other |
|
407 | 411 | cases. It also will prevent ipython from crashing when unicode |
|
408 | 412 | shows up unexpectedly in many places. If ascii encoding fails, we |
|
409 | 413 | assume utf_8. Currently the encoding is not a user-visible |
|
410 | 414 | setting, though it could be made so if there is demand for it. |
|
411 | 415 | |
|
412 | 416 | * IPython/ipmaker.py (make_IPython): remove old 2.1-specific hack. |
|
413 | 417 | |
|
414 | 418 | * IPython/Struct.py (Struct.merge): switch keys() to iterator. |
|
415 | 419 | |
|
416 | 420 | * IPython/background_jobs.py: moved 2.2 compatibility to genutils. |
|
417 | 421 | |
|
418 | 422 | * IPython/genutils.py: Add 2.2 compatibility here, so all other |
|
419 | 423 | code can work transparently for 2.2/2.3. |
|
420 | 424 | |
|
421 | 425 | 2005-07-16 Fernando Perez <fperez@colorado.edu> |
|
422 | 426 | |
|
423 | 427 | * IPython/ultraTB.py (ExceptionColors): Make a global variable |
|
424 | 428 | out of the color scheme table used for coloring exception |
|
425 | 429 | tracebacks. This allows user code to add new schemes at runtime. |
|
426 | 430 | This is a minimally modified version of the patch at |
|
427 | 431 | http://www.scipy.net/roundup/ipython/issue35, many thanks to pabw |
|
428 | 432 | for the contribution. |
|
429 | 433 | |
|
430 | 434 | * IPython/FlexCompleter.py (Completer.attr_matches): Add a |
|
431 | 435 | slightly modified version of the patch in |
|
432 | 436 | http://www.scipy.net/roundup/ipython/issue34, which also allows me |
|
433 | 437 | to remove the previous try/except solution (which was costlier). |
|
434 | 438 | Thanks to Gaetan Lehmann <gaetan.lehmann-AT-jouy.inra.fr> for the fix. |
|
435 | 439 | |
|
436 | 440 | 2005-06-08 Fernando Perez <fperez@colorado.edu> |
|
437 | 441 | |
|
438 | 442 | * IPython/iplib.py (write/write_err): Add methods to abstract all |
|
439 | 443 | I/O a bit more. |
|
440 | 444 | |
|
441 | 445 | * IPython/Shell.py (IPShellGTK.mainloop): Fix GTK deprecation |
|
442 | 446 | warning, reported by Aric Hagberg, fix by JD Hunter. |
|
443 | 447 | |
|
444 | 448 | 2005-06-02 *** Released version 0.6.15 |
|
445 | 449 | |
|
446 | 450 | 2005-06-01 Fernando Perez <fperez@colorado.edu> |
|
447 | 451 | |
|
448 | 452 | * IPython/iplib.py (MagicCompleter.file_matches): Fix |
|
449 | 453 | tab-completion of filenames within open-quoted strings. Note that |
|
450 | 454 | this requires that in ~/.ipython/ipythonrc, users change the |
|
451 | 455 | readline delimiters configuration to read: |
|
452 | 456 | |
|
453 | 457 | readline_remove_delims -/~ |
|
454 | 458 | |
|
455 | 459 | |
|
456 | 460 | 2005-05-31 *** Released version 0.6.14 |
|
457 | 461 | |
|
458 | 462 | 2005-05-29 Fernando Perez <fperez@colorado.edu> |
|
459 | 463 | |
|
460 | 464 | * IPython/ultraTB.py (VerboseTB.text): Fix crash for tracebacks |
|
461 | 465 | with files not on the filesystem. Reported by Eliyahu Sandler |
|
462 | 466 | <eli@gondolin.net> |
|
463 | 467 | |
|
464 | 468 | 2005-05-22 Fernando Perez <fperez@colorado.edu> |
|
465 | 469 | |
|
466 | 470 | * IPython/iplib.py: Fix a few crashes in the --upgrade option. |
|
467 | 471 | After an initial report by LUK ShunTim <shuntim.luk@polyu.edu.hk>. |
|
468 | 472 | |
|
469 | 473 | 2005-05-19 Fernando Perez <fperez@colorado.edu> |
|
470 | 474 | |
|
471 | 475 | * IPython/iplib.py (safe_execfile): close a file which could be |
|
472 | 476 | left open (causing problems in win32, which locks open files). |
|
473 | 477 | Thanks to a bug report by D Brown <dbrown2@yahoo.com>. |
|
474 | 478 | |
|
475 | 479 | 2005-05-18 Fernando Perez <fperez@colorado.edu> |
|
476 | 480 | |
|
477 | 481 | * IPython/Shell.py (MatplotlibShellBase.mplot_exec): pass all |
|
478 | 482 | keyword arguments correctly to safe_execfile(). |
|
479 | 483 | |
|
480 | 484 | 2005-05-13 Fernando Perez <fperez@colorado.edu> |
|
481 | 485 | |
|
482 | 486 | * ipython.1: Added info about Qt to manpage, and threads warning |
|
483 | 487 | to usage page (invoked with --help). |
|
484 | 488 | |
|
485 | 489 | * IPython/iplib.py (MagicCompleter.python_func_kw_matches): Added |
|
486 | 490 | new matcher (it goes at the end of the priority list) to do |
|
487 | 491 | tab-completion on named function arguments. Submitted by George |
|
488 | 492 | Sakkis <gsakkis-AT-eden.rutgers.edu>. See the thread at |
|
489 | 493 | http://www.scipy.net/pipermail/ipython-dev/2005-April/000436.html |
|
490 | 494 | for more details. |
|
491 | 495 | |
|
492 | 496 | * IPython/Magic.py (magic_run): Added new -e flag to ignore |
|
493 | 497 | SystemExit exceptions in the script being run. Thanks to a report |
|
494 | 498 | by danny shevitz <danny_shevitz-AT-yahoo.com>, about this |
|
495 | 499 | producing very annoying behavior when running unit tests. |
|
496 | 500 | |
|
497 | 501 | 2005-05-12 Fernando Perez <fperez@colorado.edu> |
|
498 | 502 | |
|
499 | 503 | * IPython/iplib.py (handle_auto): fixed auto-quoting and parens, |
|
500 | 504 | which I'd broken (again) due to a changed regexp. In the process, |
|
501 | 505 | added ';' as an escape to auto-quote the whole line without |
|
502 | 506 | splitting its arguments. Thanks to a report by Jerry McRae |
|
503 | 507 | <qrs0xyc02-AT-sneakemail.com>. |
|
504 | 508 | |
|
505 | 509 | * IPython/ultraTB.py (VerboseTB.text): protect against rare but |
|
506 | 510 | possible crashes caused by a TokenError. Reported by Ed Schofield |
|
507 | 511 | <schofield-AT-ftw.at>. |
|
508 | 512 | |
|
509 | 513 | 2005-05-06 Fernando Perez <fperez@colorado.edu> |
|
510 | 514 | |
|
511 | 515 | * IPython/Shell.py (hijack_wx): Fix to work with WX v.2.6. |
|
512 | 516 | |
|
513 | 517 | 2005-04-29 Fernando Perez <fperez@colorado.edu> |
|
514 | 518 | |
|
515 | 519 | * IPython/Shell.py (IPShellQt): Thanks to Denis Rivière |
|
516 | 520 | <nudz-AT-free.fr>, Yann Cointepas <yann-AT-sapetnioc.org> and Benjamin |
|
517 | 521 | Thyreau <Benji2-AT-decideur.info>, we now have a -qthread option |
|
518 | 522 | which provides support for Qt interactive usage (similar to the |
|
519 | 523 | existing one for WX and GTK). This had been often requested. |
|
520 | 524 | |
|
521 | 525 | 2005-04-14 *** Released version 0.6.13 |
|
522 | 526 | |
|
523 | 527 | 2005-04-08 Fernando Perez <fperez@colorado.edu> |
|
524 | 528 | |
|
525 | 529 | * IPython/Magic.py (Magic._ofind): remove docstring evaluation |
|
526 | 530 | from _ofind, which gets called on almost every input line. Now, |
|
527 | 531 | we only try to get docstrings if they are actually going to be |
|
528 | 532 | used (the overhead of fetching unnecessary docstrings can be |
|
529 | 533 | noticeable for certain objects, such as Pyro proxies). |
|
530 | 534 | |
|
531 | 535 | * IPython/iplib.py (MagicCompleter.python_matches): Change the API |
|
532 | 536 | for completers. For some reason I had been passing them the state |
|
533 | 537 | variable, which completers never actually need, and was in |
|
534 | 538 | conflict with the rlcompleter API. Custom completers ONLY need to |
|
535 | 539 | take the text parameter. |
|
536 | 540 | |
|
537 | 541 | * IPython/Extensions/InterpreterExec.py: Fix regexp so that magics |
|
538 | 542 | work correctly in pysh. I've also moved all the logic which used |
|
539 | 543 | to be in pysh.py here, which will prevent problems with future |
|
540 | 544 | upgrades. However, this time I must warn users to update their |
|
541 | 545 | pysh profile to include the line |
|
542 | 546 | |
|
543 | 547 | import_all IPython.Extensions.InterpreterExec |
|
544 | 548 | |
|
545 | 549 | because otherwise things won't work for them. They MUST also |
|
546 | 550 | delete pysh.py and the line |
|
547 | 551 | |
|
548 | 552 | execfile pysh.py |
|
549 | 553 | |
|
550 | 554 | from their ipythonrc-pysh. |
|
551 | 555 | |
|
552 | 556 | * IPython/FlexCompleter.py (Completer.attr_matches): Make more |
|
553 | 557 | robust in the face of objects whose dir() returns non-strings |
|
554 | 558 | (which it shouldn't, but some broken libs like ITK do). Thanks to |
|
555 | 559 | a patch by John Hunter (implemented differently, though). Also |
|
556 | 560 | minor improvements by using .extend instead of + on lists. |
|
557 | 561 | |
|
558 | 562 | * pysh.py: |
|
559 | 563 | |
|
560 | 564 | 2005-04-06 Fernando Perez <fperez@colorado.edu> |
|
561 | 565 | |
|
562 | 566 | * IPython/ipmaker.py (make_IPython): Make multi_line_specials on |
|
563 | 567 | by default, so that all users benefit from it. Those who don't |
|
564 | 568 | want it can still turn it off. |
|
565 | 569 | |
|
566 | 570 | * IPython/UserConfig/ipythonrc: Add multi_line_specials to the |
|
567 | 571 | config file, I'd forgotten about this, so users were getting it |
|
568 | 572 | off by default. |
|
569 | 573 | |
|
570 | 574 | * IPython/iplib.py (ipmagic): big overhaul of the magic system for |
|
571 | 575 | consistency. Now magics can be called in multiline statements, |
|
572 | 576 | and python variables can be expanded in magic calls via $var. |
|
573 | 577 | This makes the magic system behave just like aliases or !system |
|
574 | 578 | calls. |
|
575 | 579 | |
|
576 | 580 | 2005-03-28 Fernando Perez <fperez@colorado.edu> |
|
577 | 581 | |
|
578 | 582 | * IPython/iplib.py (handle_auto): cleanup to use %s instead of |
|
579 | 583 | expensive string additions for building command. Add support for |
|
580 | 584 | trailing ';' when autocall is used. |
|
581 | 585 | |
|
582 | 586 | 2005-03-26 Fernando Perez <fperez@colorado.edu> |
|
583 | 587 | |
|
584 | 588 | * ipython.el: Fix http://www.scipy.net/roundup/ipython/issue31. |
|
585 | 589 | Bugfix by A. Schmolck, the ipython.el maintainer. Also make |
|
586 | 590 | ipython.el robust against prompts with any number of spaces |
|
587 | 591 | (including 0) after the ':' character. |
|
588 | 592 | |
|
589 | 593 | * IPython/Prompts.py (Prompt2.set_p_str): Fix spurious space in |
|
590 | 594 | continuation prompt, which misled users to think the line was |
|
591 | 595 | already indented. Closes debian Bug#300847, reported to me by |
|
592 | 596 | Norbert Tretkowski <tretkowski-AT-inittab.de>. |
|
593 | 597 | |
|
594 | 598 | 2005-03-23 Fernando Perez <fperez@colorado.edu> |
|
595 | 599 | |
|
596 | 600 | * IPython/Prompts.py (Prompt1.__str__): Make sure that prompts are |
|
597 | 601 | properly aligned if they have embedded newlines. |
|
598 | 602 | |
|
599 | 603 | * IPython/iplib.py (runlines): Add a public method to expose |
|
600 | 604 | IPython's code execution machinery, so that users can run strings |
|
601 | 605 | as if they had been typed at the prompt interactively. |
|
602 | 606 | (InteractiveShell.__init__): Added getoutput() to the __IPYTHON__ |
|
603 | 607 | methods which can call the system shell, but with python variable |
|
604 | 608 | expansion. The three such methods are: __IPYTHON__.system, |
|
605 | 609 | .getoutput and .getoutputerror. These need to be documented in a |
|
606 | 610 | 'public API' section (to be written) of the manual. |
|
607 | 611 | |
|
608 | 612 | 2005-03-20 Fernando Perez <fperez@colorado.edu> |
|
609 | 613 | |
|
610 | 614 | * IPython/iplib.py (InteractiveShell.set_custom_exc): new system |
|
611 | 615 | for custom exception handling. This is quite powerful, and it |
|
612 | 616 | allows for user-installable exception handlers which can trap |
|
613 | 617 | custom exceptions at runtime and treat them separately from |
|
614 | 618 | IPython's default mechanisms. At the request of FrΓ©dΓ©ric |
|
615 | 619 | Mantegazza <mantegazza-AT-ill.fr>. |
|
616 | 620 | (InteractiveShell.set_custom_completer): public API function to |
|
617 | 621 | add new completers at runtime. |
|
618 | 622 | |
|
619 | 623 | 2005-03-19 Fernando Perez <fperez@colorado.edu> |
|
620 | 624 | |
|
621 | 625 | * IPython/OInspect.py (getdoc): Add a call to obj.getdoc(), to |
|
622 | 626 | allow objects which provide their docstrings via non-standard |
|
623 | 627 | mechanisms (like Pyro proxies) to still be inspected by ipython's |
|
624 | 628 | ? system. |
|
625 | 629 | |
|
626 | 630 | * IPython/iplib.py (InteractiveShell.__init__): back off the _o/_e |
|
627 | 631 | automatic capture system. I tried quite hard to make it work |
|
628 | 632 | reliably, and simply failed. I tried many combinations with the |
|
629 | 633 | subprocess module, but eventually nothing worked in all needed |
|
630 | 634 | cases (not blocking stdin for the child, duplicating stdout |
|
631 | 635 | without blocking, etc). The new %sc/%sx still do capture to these |
|
632 | 636 | magical list/string objects which make shell use much more |
|
633 | 637 | conveninent, so not all is lost. |
|
634 | 638 | |
|
635 | 639 | XXX - FIX MANUAL for the change above! |
|
636 | 640 | |
|
637 | 641 | (runsource): I copied code.py's runsource() into ipython to modify |
|
638 | 642 | it a bit. Now the code object and source to be executed are |
|
639 | 643 | stored in ipython. This makes this info accessible to third-party |
|
640 | 644 | tools, like custom exception handlers. After a request by FrΓ©dΓ©ric |
|
641 | 645 | Mantegazza <mantegazza-AT-ill.fr>. |
|
642 | 646 | |
|
643 | 647 | * IPython/UserConfig/ipythonrc: Add up/down arrow keys to |
|
644 | 648 | history-search via readline (like C-p/C-n). I'd wanted this for a |
|
645 | 649 | long time, but only recently found out how to do it. For users |
|
646 | 650 | who already have their ipythonrc files made and want this, just |
|
647 | 651 | add: |
|
648 | 652 | |
|
649 | 653 | readline_parse_and_bind "\e[A": history-search-backward |
|
650 | 654 | readline_parse_and_bind "\e[B": history-search-forward |
|
651 | 655 | |
|
652 | 656 | 2005-03-18 Fernando Perez <fperez@colorado.edu> |
|
653 | 657 | |
|
654 | 658 | * IPython/Magic.py (magic_sc): %sc and %sx now use the fancy |
|
655 | 659 | LSString and SList classes which allow transparent conversions |
|
656 | 660 | between list mode and whitespace-separated string. |
|
657 | 661 | (magic_r): Fix recursion problem in %r. |
|
658 | 662 | |
|
659 | 663 | * IPython/genutils.py (LSString): New class to be used for |
|
660 | 664 | automatic storage of the results of all alias/system calls in _o |
|
661 | 665 | and _e (stdout/err). These provide a .l/.list attribute which |
|
662 | 666 | does automatic splitting on newlines. This means that for most |
|
663 | 667 | uses, you'll never need to do capturing of output with %sc/%sx |
|
664 | 668 | anymore, since ipython keeps this always done for you. Note that |
|
665 | 669 | only the LAST results are stored, the _o/e variables are |
|
666 | 670 | overwritten on each call. If you need to save their contents |
|
667 | 671 | further, simply bind them to any other name. |
|
668 | 672 | |
|
669 | 673 | 2005-03-17 Fernando Perez <fperez@colorado.edu> |
|
670 | 674 | |
|
671 | 675 | * IPython/Prompts.py (BasePrompt.cwd_filt): a few more fixes for |
|
672 | 676 | prompt namespace handling. |
|
673 | 677 | |
|
674 | 678 | 2005-03-16 Fernando Perez <fperez@colorado.edu> |
|
675 | 679 | |
|
676 | 680 | * IPython/Prompts.py (CachedOutput.__init__): Fix default and |
|
677 | 681 | classic prompts to be '>>> ' (final space was missing, and it |
|
678 | 682 | trips the emacs python mode). |
|
679 | 683 | (BasePrompt.__str__): Added safe support for dynamic prompt |
|
680 | 684 | strings. Now you can set your prompt string to be '$x', and the |
|
681 | 685 | value of x will be printed from your interactive namespace. The |
|
682 | 686 | interpolation syntax includes the full Itpl support, so |
|
683 | 687 | ${foo()+x+bar()} is a valid prompt string now, and the function |
|
684 | 688 | calls will be made at runtime. |
|
685 | 689 | |
|
686 | 690 | 2005-03-15 Fernando Perez <fperez@colorado.edu> |
|
687 | 691 | |
|
688 | 692 | * IPython/Magic.py (magic_history): renamed %hist to %history, to |
|
689 | 693 | avoid name clashes in pylab. %hist still works, it just forwards |
|
690 | 694 | the call to %history. |
|
691 | 695 | |
|
692 | 696 | 2005-03-02 *** Released version 0.6.12 |
|
693 | 697 | |
|
694 | 698 | 2005-03-02 Fernando Perez <fperez@colorado.edu> |
|
695 | 699 | |
|
696 | 700 | * IPython/iplib.py (handle_magic): log magic calls properly as |
|
697 | 701 | ipmagic() function calls. |
|
698 | 702 | |
|
699 | 703 | * IPython/Magic.py (magic_time): Improved %time to support |
|
700 | 704 | statements and provide wall-clock as well as CPU time. |
|
701 | 705 | |
|
702 | 706 | 2005-02-27 Fernando Perez <fperez@colorado.edu> |
|
703 | 707 | |
|
704 | 708 | * IPython/hooks.py: New hooks module, to expose user-modifiable |
|
705 | 709 | IPython functionality in a clean manner. For now only the editor |
|
706 | 710 | hook is actually written, and other thigns which I intend to turn |
|
707 | 711 | into proper hooks aren't yet there. The display and prefilter |
|
708 | 712 | stuff, for example, should be hooks. But at least now the |
|
709 | 713 | framework is in place, and the rest can be moved here with more |
|
710 | 714 | time later. IPython had had a .hooks variable for a long time for |
|
711 | 715 | this purpose, but I'd never actually used it for anything. |
|
712 | 716 | |
|
713 | 717 | 2005-02-26 Fernando Perez <fperez@colorado.edu> |
|
714 | 718 | |
|
715 | 719 | * IPython/ipmaker.py (make_IPython): make the default ipython |
|
716 | 720 | directory be called _ipython under win32, to follow more the |
|
717 | 721 | naming peculiarities of that platform (where buggy software like |
|
718 | 722 | Visual Sourcesafe breaks with .named directories). Reported by |
|
719 | 723 | Ville Vainio. |
|
720 | 724 | |
|
721 | 725 | 2005-02-23 Fernando Perez <fperez@colorado.edu> |
|
722 | 726 | |
|
723 | 727 | * IPython/iplib.py (InteractiveShell.__init__): removed a few |
|
724 | 728 | auto_aliases for win32 which were causing problems. Users can |
|
725 | 729 | define the ones they personally like. |
|
726 | 730 | |
|
727 | 731 | 2005-02-21 Fernando Perez <fperez@colorado.edu> |
|
728 | 732 | |
|
729 | 733 | * IPython/Magic.py (magic_time): new magic to time execution of |
|
730 | 734 | expressions. After a request by Charles Moad <cmoad-AT-indiana.edu>. |
|
731 | 735 | |
|
732 | 736 | 2005-02-19 Fernando Perez <fperez@colorado.edu> |
|
733 | 737 | |
|
734 | 738 | * IPython/ConfigLoader.py (ConfigLoader.load): Allow empty strings |
|
735 | 739 | into keys (for prompts, for example). |
|
736 | 740 | |
|
737 | 741 | * IPython/Prompts.py (BasePrompt.set_p_str): Fix to allow empty |
|
738 | 742 | prompts in case users want them. This introduces a small behavior |
|
739 | 743 | change: ipython does not automatically add a space to all prompts |
|
740 | 744 | anymore. To get the old prompts with a space, users should add it |
|
741 | 745 | manually to their ipythonrc file, so for example prompt_in1 should |
|
742 | 746 | now read 'In [\#]: ' instead of 'In [\#]:'. |
|
743 | 747 | (BasePrompt.__init__): New option prompts_pad_left (only in rc |
|
744 | 748 | file) to control left-padding of secondary prompts. |
|
745 | 749 | |
|
746 | 750 | * IPython/Magic.py (Magic.profile_missing_notice): Don't crash if |
|
747 | 751 | the profiler can't be imported. Fix for Debian, which removed |
|
748 | 752 | profile.py because of License issues. I applied a slightly |
|
749 | 753 | modified version of the original Debian patch at |
|
750 | 754 | http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=294500. |
|
751 | 755 | |
|
752 | 756 | 2005-02-17 Fernando Perez <fperez@colorado.edu> |
|
753 | 757 | |
|
754 | 758 | * IPython/genutils.py (native_line_ends): Fix bug which would |
|
755 | 759 | cause improper line-ends under win32 b/c I was not opening files |
|
756 | 760 | in binary mode. Bug report and fix thanks to Ville. |
|
757 | 761 | |
|
758 | 762 | * IPython/iplib.py (handle_auto): Fix bug which I introduced when |
|
759 | 763 | trying to catch spurious foo[1] autocalls. My fix actually broke |
|
760 | 764 | ',/' autoquote/call with explicit escape (bad regexp). |
|
761 | 765 | |
|
762 | 766 | 2005-02-15 *** Released version 0.6.11 |
|
763 | 767 | |
|
764 | 768 | 2005-02-14 Fernando Perez <fperez@colorado.edu> |
|
765 | 769 | |
|
766 | 770 | * IPython/background_jobs.py: New background job management |
|
767 | 771 | subsystem. This is implemented via a new set of classes, and |
|
768 | 772 | IPython now provides a builtin 'jobs' object for background job |
|
769 | 773 | execution. A convenience %bg magic serves as a lightweight |
|
770 | 774 | frontend for starting the more common type of calls. This was |
|
771 | 775 | inspired by discussions with B. Granger and the BackgroundCommand |
|
772 | 776 | class described in the book Python Scripting for Computational |
|
773 | 777 | Science, by H. P. Langtangen: http://folk.uio.no/hpl/scripting |
|
774 | 778 | (although ultimately no code from this text was used, as IPython's |
|
775 | 779 | system is a separate implementation). |
|
776 | 780 | |
|
777 | 781 | * IPython/iplib.py (MagicCompleter.python_matches): add new option |
|
778 | 782 | to control the completion of single/double underscore names |
|
779 | 783 | separately. As documented in the example ipytonrc file, the |
|
780 | 784 | readline_omit__names variable can now be set to 2, to omit even |
|
781 | 785 | single underscore names. Thanks to a patch by Brian Wong |
|
782 | 786 | <BrianWong-AT-AirgoNetworks.Com>. |
|
783 | 787 | (InteractiveShell.__init__): Fix bug which would cause foo[1] to |
|
784 | 788 | be autocalled as foo([1]) if foo were callable. A problem for |
|
785 | 789 | things which are both callable and implement __getitem__. |
|
786 | 790 | (init_readline): Fix autoindentation for win32. Thanks to a patch |
|
787 | 791 | by Vivian De Smedt <vivian-AT-vdesmedt.com>. |
|
788 | 792 | |
|
789 | 793 | 2005-02-12 Fernando Perez <fperez@colorado.edu> |
|
790 | 794 | |
|
791 | 795 | * IPython/ipmaker.py (make_IPython): Disabled the stout traps |
|
792 | 796 | which I had written long ago to sort out user error messages which |
|
793 | 797 | may occur during startup. This seemed like a good idea initially, |
|
794 | 798 | but it has proven a disaster in retrospect. I don't want to |
|
795 | 799 | change much code for now, so my fix is to set the internal 'debug' |
|
796 | 800 | flag to true everywhere, whose only job was precisely to control |
|
797 | 801 | this subsystem. This closes issue 28 (as well as avoiding all |
|
798 | 802 | sorts of strange hangups which occur from time to time). |
|
799 | 803 | |
|
800 | 804 | 2005-02-07 Fernando Perez <fperez@colorado.edu> |
|
801 | 805 | |
|
802 | 806 | * IPython/Magic.py (magic_edit): Fix 'ed -p' not working when the |
|
803 | 807 | previous call produced a syntax error. |
|
804 | 808 | |
|
805 | 809 | * IPython/OInspect.py (Inspector.pinfo): Fix crash when inspecting |
|
806 | 810 | classes without constructor. |
|
807 | 811 | |
|
808 | 812 | 2005-02-06 Fernando Perez <fperez@colorado.edu> |
|
809 | 813 | |
|
810 | 814 | * IPython/iplib.py (MagicCompleter.complete): Extend the list of |
|
811 | 815 | completions with the results of each matcher, so we return results |
|
812 | 816 | to the user from all namespaces. This breaks with ipython |
|
813 | 817 | tradition, but I think it's a nicer behavior. Now you get all |
|
814 | 818 | possible completions listed, from all possible namespaces (python, |
|
815 | 819 | filesystem, magics...) After a request by John Hunter |
|
816 | 820 | <jdhunter-AT-nitace.bsd.uchicago.edu>. |
|
817 | 821 | |
|
818 | 822 | 2005-02-05 Fernando Perez <fperez@colorado.edu> |
|
819 | 823 | |
|
820 | 824 | * IPython/Magic.py (magic_prun): Fix bug where prun would fail if |
|
821 | 825 | the call had quote characters in it (the quotes were stripped). |
|
822 | 826 | |
|
823 | 827 | 2005-01-31 Fernando Perez <fperez@colorado.edu> |
|
824 | 828 | |
|
825 | 829 | * IPython/iplib.py (InteractiveShell.__init__): reduce reliance on |
|
826 | 830 | Itpl.itpl() to make the code more robust against psyco |
|
827 | 831 | optimizations. |
|
828 | 832 | |
|
829 | 833 | * IPython/Itpl.py (Itpl.__str__): Use a _getframe() call instead |
|
830 | 834 | of causing an exception. Quicker, cleaner. |
|
831 | 835 | |
|
832 | 836 | 2005-01-28 Fernando Perez <fperez@colorado.edu> |
|
833 | 837 | |
|
834 | 838 | * scripts/ipython_win_post_install.py (install): hardcode |
|
835 | 839 | sys.prefix+'python.exe' as the executable path. It turns out that |
|
836 | 840 | during the post-installation run, sys.executable resolves to the |
|
837 | 841 | name of the binary installer! I should report this as a distutils |
|
838 | 842 | bug, I think. I updated the .10 release with this tiny fix, to |
|
839 | 843 | avoid annoying the lists further. |
|
840 | 844 | |
|
841 | 845 | 2005-01-27 *** Released version 0.6.10 |
|
842 | 846 | |
|
843 | 847 | 2005-01-27 Fernando Perez <fperez@colorado.edu> |
|
844 | 848 | |
|
845 | 849 | * IPython/numutils.py (norm): Added 'inf' as optional name for |
|
846 | 850 | L-infinity norm, included references to mathworld.com for vector |
|
847 | 851 | norm definitions. |
|
848 | 852 | (amin/amax): added amin/amax for array min/max. Similar to what |
|
849 | 853 | pylab ships with after the recent reorganization of names. |
|
850 | 854 | (spike/spike_odd): removed deprecated spike/spike_odd functions. |
|
851 | 855 | |
|
852 | 856 | * ipython.el: committed Alex's recent fixes and improvements. |
|
853 | 857 | Tested with python-mode from CVS, and it looks excellent. Since |
|
854 | 858 | python-mode hasn't released anything in a while, I'm temporarily |
|
855 | 859 | putting a copy of today's CVS (v 4.70) of python-mode in: |
|
856 | 860 | http://ipython.scipy.org/tmp/python-mode.el |
|
857 | 861 | |
|
858 | 862 | * scripts/ipython_win_post_install.py (install): Win32 fix to use |
|
859 | 863 | sys.executable for the executable name, instead of assuming it's |
|
860 | 864 | called 'python.exe' (the post-installer would have produced broken |
|
861 | 865 | setups on systems with a differently named python binary). |
|
862 | 866 | |
|
863 | 867 | * IPython/PyColorize.py (Parser.__call__): change explicit '\n' |
|
864 | 868 | references to os.linesep, to make the code more |
|
865 | 869 | platform-independent. This is also part of the win32 coloring |
|
866 | 870 | fixes. |
|
867 | 871 | |
|
868 | 872 | * IPython/genutils.py (page_dumb): Remove attempts to chop long |
|
869 | 873 | lines, which actually cause coloring bugs because the length of |
|
870 | 874 | the line is very difficult to correctly compute with embedded |
|
871 | 875 | escapes. This was the source of all the coloring problems under |
|
872 | 876 | Win32. I think that _finally_, Win32 users have a properly |
|
873 | 877 | working ipython in all respects. This would never have happened |
|
874 | 878 | if not for Gary Bishop and Viktor Ransmayr's great help and work. |
|
875 | 879 | |
|
876 | 880 | 2005-01-26 *** Released version 0.6.9 |
|
877 | 881 | |
|
878 | 882 | 2005-01-25 Fernando Perez <fperez@colorado.edu> |
|
879 | 883 | |
|
880 | 884 | * setup.py: finally, we have a true Windows installer, thanks to |
|
881 | 885 | the excellent work of Viktor Ransmayr |
|
882 | 886 | <viktor.ransmayr-AT-t-online.de>. The docs have been updated for |
|
883 | 887 | Windows users. The setup routine is quite a bit cleaner thanks to |
|
884 | 888 | this, and the post-install script uses the proper functions to |
|
885 | 889 | allow a clean de-installation using the standard Windows Control |
|
886 | 890 | Panel. |
|
887 | 891 | |
|
888 | 892 | * IPython/genutils.py (get_home_dir): changed to use the $HOME |
|
889 | 893 | environment variable under all OSes (including win32) if |
|
890 | 894 | available. This will give consistency to win32 users who have set |
|
891 | 895 | this variable for any reason. If os.environ['HOME'] fails, the |
|
892 | 896 | previous policy of using HOMEDRIVE\HOMEPATH kicks in. |
|
893 | 897 | |
|
894 | 898 | 2005-01-24 Fernando Perez <fperez@colorado.edu> |
|
895 | 899 | |
|
896 | 900 | * IPython/numutils.py (empty_like): add empty_like(), similar to |
|
897 | 901 | zeros_like() but taking advantage of the new empty() Numeric routine. |
|
898 | 902 | |
|
899 | 903 | 2005-01-23 *** Released version 0.6.8 |
|
900 | 904 | |
|
901 | 905 | 2005-01-22 Fernando Perez <fperez@colorado.edu> |
|
902 | 906 | |
|
903 | 907 | * IPython/Shell.py (MatplotlibShellBase.mplot_exec): I removed the |
|
904 | 908 | automatic show() calls. After discussing things with JDH, it |
|
905 | 909 | turns out there are too many corner cases where this can go wrong. |
|
906 | 910 | It's best not to try to be 'too smart', and simply have ipython |
|
907 | 911 | reproduce as much as possible the default behavior of a normal |
|
908 | 912 | python shell. |
|
909 | 913 | |
|
910 | 914 | * IPython/iplib.py (InteractiveShell.__init__): Modified the |
|
911 | 915 | line-splitting regexp and _prefilter() to avoid calling getattr() |
|
912 | 916 | on assignments. This closes |
|
913 | 917 | http://www.scipy.net/roundup/ipython/issue24. Note that Python's |
|
914 | 918 | readline uses getattr(), so a simple <TAB> keypress is still |
|
915 | 919 | enough to trigger getattr() calls on an object. |
|
916 | 920 | |
|
917 | 921 | 2005-01-21 Fernando Perez <fperez@colorado.edu> |
|
918 | 922 | |
|
919 | 923 | * IPython/Shell.py (MatplotlibShellBase.magic_run): Fix the %run |
|
920 | 924 | docstring under pylab so it doesn't mask the original. |
|
921 | 925 | |
|
922 | 926 | 2005-01-21 *** Released version 0.6.7 |
|
923 | 927 | |
|
924 | 928 | 2005-01-21 Fernando Perez <fperez@colorado.edu> |
|
925 | 929 | |
|
926 | 930 | * IPython/Shell.py (MTInteractiveShell.runcode): Trap a crash with |
|
927 | 931 | signal handling for win32 users in multithreaded mode. |
|
928 | 932 | |
|
929 | 933 | 2005-01-17 Fernando Perez <fperez@colorado.edu> |
|
930 | 934 | |
|
931 | 935 | * IPython/OInspect.py (Inspector.pinfo): Fix crash when inspecting |
|
932 | 936 | instances with no __init__. After a crash report by Norbert Nemec |
|
933 | 937 | <Norbert-AT-nemec-online.de>. |
|
934 | 938 | |
|
935 | 939 | 2005-01-14 Fernando Perez <fperez@colorado.edu> |
|
936 | 940 | |
|
937 | 941 | * IPython/ultraTB.py (VerboseTB.text): Fix bug in reporting of |
|
938 | 942 | names for verbose exceptions, when multiple dotted names and the |
|
939 | 943 | 'parent' object were present on the same line. |
|
940 | 944 | |
|
941 | 945 | 2005-01-11 Fernando Perez <fperez@colorado.edu> |
|
942 | 946 | |
|
943 | 947 | * IPython/genutils.py (flag_calls): new utility to trap and flag |
|
944 | 948 | calls in functions. I need it to clean up matplotlib support. |
|
945 | 949 | Also removed some deprecated code in genutils. |
|
946 | 950 | |
|
947 | 951 | * IPython/Shell.py (MatplotlibShellBase.mplot_exec): small fix so |
|
948 | 952 | that matplotlib scripts called with %run, which don't call show() |
|
949 | 953 | themselves, still have their plotting windows open. |
|
950 | 954 | |
|
951 | 955 | 2005-01-05 Fernando Perez <fperez@colorado.edu> |
|
952 | 956 | |
|
953 | 957 | * IPython/Shell.py (IPShellGTK.__init__): Patch by Andrew Straw |
|
954 | 958 | <astraw-AT-caltech.edu>, to fix gtk deprecation warnings. |
|
955 | 959 | |
|
956 | 960 | 2004-12-19 Fernando Perez <fperez@colorado.edu> |
|
957 | 961 | |
|
958 | 962 | * IPython/Shell.py (MTInteractiveShell.runcode): Get rid of |
|
959 | 963 | parent_runcode, which was an eyesore. The same result can be |
|
960 | 964 | obtained with Python's regular superclass mechanisms. |
|
961 | 965 | |
|
962 | 966 | 2004-12-17 Fernando Perez <fperez@colorado.edu> |
|
963 | 967 | |
|
964 | 968 | * IPython/Magic.py (Magic.magic_sc): Fix quote stripping problem |
|
965 | 969 | reported by Prabhu. |
|
966 | 970 | (Magic.magic_sx): direct all errors to Term.cerr (defaults to |
|
967 | 971 | sys.stderr) instead of explicitly calling sys.stderr. This helps |
|
968 | 972 | maintain our I/O abstractions clean, for future GUI embeddings. |
|
969 | 973 | |
|
970 | 974 | * IPython/genutils.py (info): added new utility for sys.stderr |
|
971 | 975 | unified info message handling (thin wrapper around warn()). |
|
972 | 976 | |
|
973 | 977 | * IPython/ultraTB.py (VerboseTB.text): Fix misreported global |
|
974 | 978 | composite (dotted) names on verbose exceptions. |
|
975 | 979 | (VerboseTB.nullrepr): harden against another kind of errors which |
|
976 | 980 | Python's inspect module can trigger, and which were crashing |
|
977 | 981 | IPython. Thanks to a report by Marco Lombardi |
|
978 | 982 | <mlombard-AT-ma010192.hq.eso.org>. |
|
979 | 983 | |
|
980 | 984 | 2004-12-13 *** Released version 0.6.6 |
|
981 | 985 | |
|
982 | 986 | 2004-12-12 Fernando Perez <fperez@colorado.edu> |
|
983 | 987 | |
|
984 | 988 | * IPython/Shell.py (IPShellGTK.mainloop): catch RuntimeErrors |
|
985 | 989 | generated by pygtk upon initialization if it was built without |
|
986 | 990 | threads (for matplotlib users). After a crash reported by |
|
987 | 991 | Leguijt, Jaap J SIEP-EPT-RES <Jaap.Leguijt-AT-shell.com>. |
|
988 | 992 | |
|
989 | 993 | * IPython/ipmaker.py (make_IPython): fix small bug in the |
|
990 | 994 | import_some parameter for multiple imports. |
|
991 | 995 | |
|
992 | 996 | * IPython/iplib.py (ipmagic): simplified the interface of |
|
993 | 997 | ipmagic() to take a single string argument, just as it would be |
|
994 | 998 | typed at the IPython cmd line. |
|
995 | 999 | (ipalias): Added new ipalias() with an interface identical to |
|
996 | 1000 | ipmagic(). This completes exposing a pure python interface to the |
|
997 | 1001 | alias and magic system, which can be used in loops or more complex |
|
998 | 1002 | code where IPython's automatic line mangling is not active. |
|
999 | 1003 | |
|
1000 | 1004 | * IPython/genutils.py (timing): changed interface of timing to |
|
1001 | 1005 | simply run code once, which is the most common case. timings() |
|
1002 | 1006 | remains unchanged, for the cases where you want multiple runs. |
|
1003 | 1007 | |
|
1004 | 1008 | * IPython/Shell.py (MatplotlibShellBase._matplotlib_config): Fix a |
|
1005 | 1009 | bug where Python2.2 crashes with exec'ing code which does not end |
|
1006 | 1010 | in a single newline. Python 2.3 is OK, so I hadn't noticed this |
|
1007 | 1011 | before. |
|
1008 | 1012 | |
|
1009 | 1013 | 2004-12-10 Fernando Perez <fperez@colorado.edu> |
|
1010 | 1014 | |
|
1011 | 1015 | * IPython/Magic.py (Magic.magic_prun): changed name of option from |
|
1012 | 1016 | -t to -T, to accomodate the new -t flag in %run (the %run and |
|
1013 | 1017 | %prun options are kind of intermixed, and it's not easy to change |
|
1014 | 1018 | this with the limitations of python's getopt). |
|
1015 | 1019 | |
|
1016 | 1020 | * IPython/Magic.py (Magic.magic_run): Added new -t option to time |
|
1017 | 1021 | the execution of scripts. It's not as fine-tuned as timeit.py, |
|
1018 | 1022 | but it works from inside ipython (and under 2.2, which lacks |
|
1019 | 1023 | timeit.py). Optionally a number of runs > 1 can be given for |
|
1020 | 1024 | timing very short-running code. |
|
1021 | 1025 | |
|
1022 | 1026 | * IPython/genutils.py (uniq_stable): new routine which returns a |
|
1023 | 1027 | list of unique elements in any iterable, but in stable order of |
|
1024 | 1028 | appearance. I needed this for the ultraTB fixes, and it's a handy |
|
1025 | 1029 | utility. |
|
1026 | 1030 | |
|
1027 | 1031 | * IPython/ultraTB.py (VerboseTB.text): Fix proper reporting of |
|
1028 | 1032 | dotted names in Verbose exceptions. This had been broken since |
|
1029 | 1033 | the very start, now x.y will properly be printed in a Verbose |
|
1030 | 1034 | traceback, instead of x being shown and y appearing always as an |
|
1031 | 1035 | 'undefined global'. Getting this to work was a bit tricky, |
|
1032 | 1036 | because by default python tokenizers are stateless. Saved by |
|
1033 | 1037 | python's ability to easily add a bit of state to an arbitrary |
|
1034 | 1038 | function (without needing to build a full-blown callable object). |
|
1035 | 1039 | |
|
1036 | 1040 | Also big cleanup of this code, which had horrendous runtime |
|
1037 | 1041 | lookups of zillions of attributes for colorization. Moved all |
|
1038 | 1042 | this code into a few templates, which make it cleaner and quicker. |
|
1039 | 1043 | |
|
1040 | 1044 | Printout quality was also improved for Verbose exceptions: one |
|
1041 | 1045 | variable per line, and memory addresses are printed (this can be |
|
1042 | 1046 | quite handy in nasty debugging situations, which is what Verbose |
|
1043 | 1047 | is for). |
|
1044 | 1048 | |
|
1045 | 1049 | * IPython/ipmaker.py (make_IPython): Do NOT execute files named in |
|
1046 | 1050 | the command line as scripts to be loaded by embedded instances. |
|
1047 | 1051 | Doing so has the potential for an infinite recursion if there are |
|
1048 | 1052 | exceptions thrown in the process. This fixes a strange crash |
|
1049 | 1053 | reported by Philippe MULLER <muller-AT-irit.fr>. |
|
1050 | 1054 | |
|
1051 | 1055 | 2004-12-09 Fernando Perez <fperez@colorado.edu> |
|
1052 | 1056 | |
|
1053 | 1057 | * IPython/Shell.py (MatplotlibShellBase.use): Change pylab support |
|
1054 | 1058 | to reflect new names in matplotlib, which now expose the |
|
1055 | 1059 | matlab-compatible interface via a pylab module instead of the |
|
1056 | 1060 | 'matlab' name. The new code is backwards compatible, so users of |
|
1057 | 1061 | all matplotlib versions are OK. Patch by J. Hunter. |
|
1058 | 1062 | |
|
1059 | 1063 | * IPython/OInspect.py (Inspector.pinfo): Add to object? printing |
|
1060 | 1064 | of __init__ docstrings for instances (class docstrings are already |
|
1061 | 1065 | automatically printed). Instances with customized docstrings |
|
1062 | 1066 | (indep. of the class) are also recognized and all 3 separate |
|
1063 | 1067 | docstrings are printed (instance, class, constructor). After some |
|
1064 | 1068 | comments/suggestions by J. Hunter. |
|
1065 | 1069 | |
|
1066 | 1070 | 2004-12-05 Fernando Perez <fperez@colorado.edu> |
|
1067 | 1071 | |
|
1068 | 1072 | * IPython/iplib.py (MagicCompleter.complete): Remove annoying |
|
1069 | 1073 | warnings when tab-completion fails and triggers an exception. |
|
1070 | 1074 | |
|
1071 | 1075 | 2004-12-03 Fernando Perez <fperez@colorado.edu> |
|
1072 | 1076 | |
|
1073 | 1077 | * IPython/Magic.py (magic_prun): Fix bug where an exception would |
|
1074 | 1078 | be triggered when using 'run -p'. An incorrect option flag was |
|
1075 | 1079 | being set ('d' instead of 'D'). |
|
1076 | 1080 | (manpage): fix missing escaped \- sign. |
|
1077 | 1081 | |
|
1078 | 1082 | 2004-11-30 *** Released version 0.6.5 |
|
1079 | 1083 | |
|
1080 | 1084 | 2004-11-30 Fernando Perez <fperez@colorado.edu> |
|
1081 | 1085 | |
|
1082 | 1086 | * IPython/Magic.py (Magic.magic_run): Fix bug in breakpoint |
|
1083 | 1087 | setting with -d option. |
|
1084 | 1088 | |
|
1085 | 1089 | * setup.py (docfiles): Fix problem where the doc glob I was using |
|
1086 | 1090 | was COMPLETELY BROKEN. It was giving the right files by pure |
|
1087 | 1091 | accident, but failed once I tried to include ipython.el. Note: |
|
1088 | 1092 | glob() does NOT allow you to do exclusion on multiple endings! |
|
1089 | 1093 | |
|
1090 | 1094 | 2004-11-29 Fernando Perez <fperez@colorado.edu> |
|
1091 | 1095 | |
|
1092 | 1096 | * IPython/usage.py (__doc__): cleaned up usage docstring, by using |
|
1093 | 1097 | the manpage as the source. Better formatting & consistency. |
|
1094 | 1098 | |
|
1095 | 1099 | * IPython/Magic.py (magic_run): Added new -d option, to run |
|
1096 | 1100 | scripts under the control of the python pdb debugger. Note that |
|
1097 | 1101 | this required changing the %prun option -d to -D, to avoid a clash |
|
1098 | 1102 | (since %run must pass options to %prun, and getopt is too dumb to |
|
1099 | 1103 | handle options with string values with embedded spaces). Thanks |
|
1100 | 1104 | to a suggestion by Matthew Arnison <maffew-AT-cat.org.au>. |
|
1101 | 1105 | (magic_who_ls): added type matching to %who and %whos, so that one |
|
1102 | 1106 | can filter their output to only include variables of certain |
|
1103 | 1107 | types. Another suggestion by Matthew. |
|
1104 | 1108 | (magic_whos): Added memory summaries in kb and Mb for arrays. |
|
1105 | 1109 | (magic_who): Improve formatting (break lines every 9 vars). |
|
1106 | 1110 | |
|
1107 | 1111 | 2004-11-28 Fernando Perez <fperez@colorado.edu> |
|
1108 | 1112 | |
|
1109 | 1113 | * IPython/Logger.py (Logger.log): Fix bug in syncing the input |
|
1110 | 1114 | cache when empty lines were present. |
|
1111 | 1115 | |
|
1112 | 1116 | 2004-11-24 Fernando Perez <fperez@colorado.edu> |
|
1113 | 1117 | |
|
1114 | 1118 | * IPython/usage.py (__doc__): document the re-activated threading |
|
1115 | 1119 | options for WX and GTK. |
|
1116 | 1120 | |
|
1117 | 1121 | 2004-11-23 Fernando Perez <fperez@colorado.edu> |
|
1118 | 1122 | |
|
1119 | 1123 | * IPython/Shell.py (start): Added Prabhu's big patch to reactivate |
|
1120 | 1124 | the -wthread and -gthread options, along with a new -tk one to try |
|
1121 | 1125 | and coordinate Tk threading with wx/gtk. The tk support is very |
|
1122 | 1126 | platform dependent, since it seems to require Tcl and Tk to be |
|
1123 | 1127 | built with threads (Fedora1/2 appears NOT to have it, but in |
|
1124 | 1128 | Prabhu's Debian boxes it works OK). But even with some Tk |
|
1125 | 1129 | limitations, this is a great improvement. |
|
1126 | 1130 | |
|
1127 | 1131 | * IPython/Prompts.py (prompt_specials_color): Added \t for time |
|
1128 | 1132 | info in user prompts. Patch by Prabhu. |
|
1129 | 1133 | |
|
1130 | 1134 | 2004-11-18 Fernando Perez <fperez@colorado.edu> |
|
1131 | 1135 | |
|
1132 | 1136 | * IPython/genutils.py (ask_yes_no): Add check for a max of 20 |
|
1133 | 1137 | EOFErrors and bail, to avoid infinite loops if a non-terminating |
|
1134 | 1138 | file is fed into ipython. Patch submitted in issue 19 by user, |
|
1135 | 1139 | many thanks. |
|
1136 | 1140 | |
|
1137 | 1141 | * IPython/iplib.py (InteractiveShell.handle_auto): do NOT trigger |
|
1138 | 1142 | autoquote/parens in continuation prompts, which can cause lots of |
|
1139 | 1143 | problems. Closes roundup issue 20. |
|
1140 | 1144 | |
|
1141 | 1145 | 2004-11-17 Fernando Perez <fperez@colorado.edu> |
|
1142 | 1146 | |
|
1143 | 1147 | * debian/control (Build-Depends-Indep): Fix dpatch dependency, |
|
1144 | 1148 | reported as debian bug #280505. I'm not sure my local changelog |
|
1145 | 1149 | entry has the proper debian format (Jack?). |
|
1146 | 1150 | |
|
1147 | 1151 | 2004-11-08 *** Released version 0.6.4 |
|
1148 | 1152 | |
|
1149 | 1153 | 2004-11-08 Fernando Perez <fperez@colorado.edu> |
|
1150 | 1154 | |
|
1151 | 1155 | * IPython/iplib.py (init_readline): Fix exit message for Windows |
|
1152 | 1156 | when readline is active. Thanks to a report by Eric Jones |
|
1153 | 1157 | <eric-AT-enthought.com>. |
|
1154 | 1158 | |
|
1155 | 1159 | 2004-11-07 Fernando Perez <fperez@colorado.edu> |
|
1156 | 1160 | |
|
1157 | 1161 | * IPython/genutils.py (page): Add a trap for OSError exceptions, |
|
1158 | 1162 | sometimes seen by win2k/cygwin users. |
|
1159 | 1163 | |
|
1160 | 1164 | 2004-11-06 Fernando Perez <fperez@colorado.edu> |
|
1161 | 1165 | |
|
1162 | 1166 | * IPython/iplib.py (interact): Change the handling of %Exit from |
|
1163 | 1167 | trying to propagate a SystemExit to an internal ipython flag. |
|
1164 | 1168 | This is less elegant than using Python's exception mechanism, but |
|
1165 | 1169 | I can't get that to work reliably with threads, so under -pylab |
|
1166 | 1170 | %Exit was hanging IPython. Cross-thread exception handling is |
|
1167 | 1171 | really a bitch. Thaks to a bug report by Stephen Walton |
|
1168 | 1172 | <stephen.walton-AT-csun.edu>. |
|
1169 | 1173 | |
|
1170 | 1174 | 2004-11-04 Fernando Perez <fperez@colorado.edu> |
|
1171 | 1175 | |
|
1172 | 1176 | * IPython/iplib.py (raw_input_original): store a pointer to the |
|
1173 | 1177 | true raw_input to harden against code which can modify it |
|
1174 | 1178 | (wx.py.PyShell does this and would otherwise crash ipython). |
|
1175 | 1179 | Thanks to a bug report by Jim Flowers <james.flowers-AT-lgx.com>. |
|
1176 | 1180 | |
|
1177 | 1181 | * IPython/Shell.py (MTInteractiveShell.runsource): Cleaner fix for |
|
1178 | 1182 | Ctrl-C problem, which does not mess up the input line. |
|
1179 | 1183 | |
|
1180 | 1184 | 2004-11-03 Fernando Perez <fperez@colorado.edu> |
|
1181 | 1185 | |
|
1182 | 1186 | * IPython/Release.py: Changed licensing to BSD, in all files. |
|
1183 | 1187 | (name): lowercase name for tarball/RPM release. |
|
1184 | 1188 | |
|
1185 | 1189 | * IPython/OInspect.py (getdoc): wrap inspect.getdoc() safely for |
|
1186 | 1190 | use throughout ipython. |
|
1187 | 1191 | |
|
1188 | 1192 | * IPython/Magic.py (Magic._ofind): Switch to using the new |
|
1189 | 1193 | OInspect.getdoc() function. |
|
1190 | 1194 | |
|
1191 | 1195 | * IPython/Shell.py (sigint_handler): Hack to ignore the execution |
|
1192 | 1196 | of the line currently being canceled via Ctrl-C. It's extremely |
|
1193 | 1197 | ugly, but I don't know how to do it better (the problem is one of |
|
1194 | 1198 | handling cross-thread exceptions). |
|
1195 | 1199 | |
|
1196 | 1200 | 2004-10-28 Fernando Perez <fperez@colorado.edu> |
|
1197 | 1201 | |
|
1198 | 1202 | * IPython/Shell.py (signal_handler): add signal handlers to trap |
|
1199 | 1203 | SIGINT and SIGSEGV in threaded code properly. Thanks to a bug |
|
1200 | 1204 | report by Francesc Alted. |
|
1201 | 1205 | |
|
1202 | 1206 | 2004-10-21 Fernando Perez <fperez@colorado.edu> |
|
1203 | 1207 | |
|
1204 | 1208 | * IPython/Extensions/InterpreterExec.py (prefilter_shell): Fix @ |
|
1205 | 1209 | to % for pysh syntax extensions. |
|
1206 | 1210 | |
|
1207 | 1211 | 2004-10-09 Fernando Perez <fperez@colorado.edu> |
|
1208 | 1212 | |
|
1209 | 1213 | * IPython/Magic.py (Magic.magic_whos): modify output of Numeric |
|
1210 | 1214 | arrays to print a more useful summary, without calling str(arr). |
|
1211 | 1215 | This avoids the problem of extremely lengthy computations which |
|
1212 | 1216 | occur if arr is large, and appear to the user as a system lockup |
|
1213 | 1217 | with 100% cpu activity. After a suggestion by Kristian Sandberg |
|
1214 | 1218 | <Kristian.Sandberg@colorado.edu>. |
|
1215 | 1219 | (Magic.__init__): fix bug in global magic escapes not being |
|
1216 | 1220 | correctly set. |
|
1217 | 1221 | |
|
1218 | 1222 | 2004-10-08 Fernando Perez <fperez@colorado.edu> |
|
1219 | 1223 | |
|
1220 | 1224 | * IPython/Magic.py (__license__): change to absolute imports of |
|
1221 | 1225 | ipython's own internal packages, to start adapting to the absolute |
|
1222 | 1226 | import requirement of PEP-328. |
|
1223 | 1227 | |
|
1224 | 1228 | * IPython/genutils.py (__author__): Fix coding to utf-8 on all |
|
1225 | 1229 | files, and standardize author/license marks through the Release |
|
1226 | 1230 | module instead of having per/file stuff (except for files with |
|
1227 | 1231 | particular licenses, like the MIT/PSF-licensed codes). |
|
1228 | 1232 | |
|
1229 | 1233 | * IPython/Debugger.py: remove dead code for python 2.1 |
|
1230 | 1234 | |
|
1231 | 1235 | 2004-10-04 Fernando Perez <fperez@colorado.edu> |
|
1232 | 1236 | |
|
1233 | 1237 | * IPython/iplib.py (ipmagic): New function for accessing magics |
|
1234 | 1238 | via a normal python function call. |
|
1235 | 1239 | |
|
1236 | 1240 | * IPython/Magic.py (Magic.magic_magic): Change the magic escape |
|
1237 | 1241 | from '@' to '%', to accomodate the new @decorator syntax of python |
|
1238 | 1242 | 2.4. |
|
1239 | 1243 | |
|
1240 | 1244 | 2004-09-29 Fernando Perez <fperez@colorado.edu> |
|
1241 | 1245 | |
|
1242 | 1246 | * IPython/Shell.py (MatplotlibShellBase.use): Added a wrapper to |
|
1243 | 1247 | matplotlib.use to prevent running scripts which try to switch |
|
1244 | 1248 | interactive backends from within ipython. This will just crash |
|
1245 | 1249 | the python interpreter, so we can't allow it (but a detailed error |
|
1246 | 1250 | is given to the user). |
|
1247 | 1251 | |
|
1248 | 1252 | 2004-09-28 Fernando Perez <fperez@colorado.edu> |
|
1249 | 1253 | |
|
1250 | 1254 | * IPython/Shell.py (MatplotlibShellBase.mplot_exec): |
|
1251 | 1255 | matplotlib-related fixes so that using @run with non-matplotlib |
|
1252 | 1256 | scripts doesn't pop up spurious plot windows. This requires |
|
1253 | 1257 | matplotlib >= 0.63, where I had to make some changes as well. |
|
1254 | 1258 | |
|
1255 | 1259 | * IPython/ipmaker.py (make_IPython): update version requirement to |
|
1256 | 1260 | python 2.2. |
|
1257 | 1261 | |
|
1258 | 1262 | * IPython/iplib.py (InteractiveShell.mainloop): Add an optional |
|
1259 | 1263 | banner arg for embedded customization. |
|
1260 | 1264 | |
|
1261 | 1265 | * IPython/Magic.py (Magic.__init__): big cleanup to remove all |
|
1262 | 1266 | explicit uses of __IP as the IPython's instance name. Now things |
|
1263 | 1267 | are properly handled via the shell.name value. The actual code |
|
1264 | 1268 | is a bit ugly b/c I'm doing it via a global in Magic.py, but this |
|
1265 | 1269 | is much better than before. I'll clean things completely when the |
|
1266 | 1270 | magic stuff gets a real overhaul. |
|
1267 | 1271 | |
|
1268 | 1272 | * ipython.1: small fixes, sent in by Jack Moffit. He also sent in |
|
1269 | 1273 | minor changes to debian dir. |
|
1270 | 1274 | |
|
1271 | 1275 | * IPython/iplib.py (InteractiveShell.__init__): Fix adding a |
|
1272 | 1276 | pointer to the shell itself in the interactive namespace even when |
|
1273 | 1277 | a user-supplied dict is provided. This is needed for embedding |
|
1274 | 1278 | purposes (found by tests with Michel Sanner). |
|
1275 | 1279 | |
|
1276 | 1280 | 2004-09-27 Fernando Perez <fperez@colorado.edu> |
|
1277 | 1281 | |
|
1278 | 1282 | * IPython/UserConfig/ipythonrc: remove []{} from |
|
1279 | 1283 | readline_remove_delims, so that things like [modname.<TAB> do |
|
1280 | 1284 | proper completion. This disables [].TAB, but that's a less common |
|
1281 | 1285 | case than module names in list comprehensions, for example. |
|
1282 | 1286 | Thanks to a report by Andrea Riciputi. |
|
1283 | 1287 | |
|
1284 | 1288 | 2004-09-09 Fernando Perez <fperez@colorado.edu> |
|
1285 | 1289 | |
|
1286 | 1290 | * IPython/Shell.py (IPShellGTK.mainloop): reorder to avoid |
|
1287 | 1291 | blocking problems in win32 and osx. Fix by John. |
|
1288 | 1292 | |
|
1289 | 1293 | 2004-09-08 Fernando Perez <fperez@colorado.edu> |
|
1290 | 1294 | |
|
1291 | 1295 | * IPython/Shell.py (IPShellWX.OnInit): Fix output redirection bug |
|
1292 | 1296 | for Win32 and OSX. Fix by John Hunter. |
|
1293 | 1297 | |
|
1294 | 1298 | 2004-08-30 *** Released version 0.6.3 |
|
1295 | 1299 | |
|
1296 | 1300 | 2004-08-30 Fernando Perez <fperez@colorado.edu> |
|
1297 | 1301 | |
|
1298 | 1302 | * setup.py (isfile): Add manpages to list of dependent files to be |
|
1299 | 1303 | updated. |
|
1300 | 1304 | |
|
1301 | 1305 | 2004-08-27 Fernando Perez <fperez@colorado.edu> |
|
1302 | 1306 | |
|
1303 | 1307 | * IPython/Shell.py (start): I've disabled -wthread and -gthread |
|
1304 | 1308 | for now. They don't really work with standalone WX/GTK code |
|
1305 | 1309 | (though matplotlib IS working fine with both of those backends). |
|
1306 | 1310 | This will neeed much more testing. I disabled most things with |
|
1307 | 1311 | comments, so turning it back on later should be pretty easy. |
|
1308 | 1312 | |
|
1309 | 1313 | * IPython/iplib.py (InteractiveShell.__init__): Fix accidental |
|
1310 | 1314 | autocalling of expressions like r'foo', by modifying the line |
|
1311 | 1315 | split regexp. Closes |
|
1312 | 1316 | http://www.scipy.net/roundup/ipython/issue18, reported by Nicholas |
|
1313 | 1317 | Riley <ipythonbugs-AT-sabi.net>. |
|
1314 | 1318 | (InteractiveShell.mainloop): honor --nobanner with banner |
|
1315 | 1319 | extensions. |
|
1316 | 1320 | |
|
1317 | 1321 | * IPython/Shell.py: Significant refactoring of all classes, so |
|
1318 | 1322 | that we can really support ALL matplotlib backends and threading |
|
1319 | 1323 | models (John spotted a bug with Tk which required this). Now we |
|
1320 | 1324 | should support single-threaded, WX-threads and GTK-threads, both |
|
1321 | 1325 | for generic code and for matplotlib. |
|
1322 | 1326 | |
|
1323 | 1327 | * IPython/ipmaker.py (__call__): Changed -mpthread option to |
|
1324 | 1328 | -pylab, to simplify things for users. Will also remove the pylab |
|
1325 | 1329 | profile, since now all of matplotlib configuration is directly |
|
1326 | 1330 | handled here. This also reduces startup time. |
|
1327 | 1331 | |
|
1328 | 1332 | * IPython/Shell.py (IPShellGTK.run): Fixed bug where mainloop() of |
|
1329 | 1333 | shell wasn't being correctly called. Also in IPShellWX. |
|
1330 | 1334 | |
|
1331 | 1335 | * IPython/iplib.py (InteractiveShell.__init__): Added option to |
|
1332 | 1336 | fine-tune banner. |
|
1333 | 1337 | |
|
1334 | 1338 | * IPython/numutils.py (spike): Deprecate these spike functions, |
|
1335 | 1339 | delete (long deprecated) gnuplot_exec handler. |
|
1336 | 1340 | |
|
1337 | 1341 | 2004-08-26 Fernando Perez <fperez@colorado.edu> |
|
1338 | 1342 | |
|
1339 | 1343 | * ipython.1: Update for threading options, plus some others which |
|
1340 | 1344 | were missing. |
|
1341 | 1345 | |
|
1342 | 1346 | * IPython/ipmaker.py (__call__): Added -wthread option for |
|
1343 | 1347 | wxpython thread handling. Make sure threading options are only |
|
1344 | 1348 | valid at the command line. |
|
1345 | 1349 | |
|
1346 | 1350 | * scripts/ipython: moved shell selection into a factory function |
|
1347 | 1351 | in Shell.py, to keep the starter script to a minimum. |
|
1348 | 1352 | |
|
1349 | 1353 | 2004-08-25 Fernando Perez <fperez@colorado.edu> |
|
1350 | 1354 | |
|
1351 | 1355 | * IPython/Shell.py (IPShellWX.wxexit): fixes to WX threading, by |
|
1352 | 1356 | John. Along with some recent changes he made to matplotlib, the |
|
1353 | 1357 | next versions of both systems should work very well together. |
|
1354 | 1358 | |
|
1355 | 1359 | 2004-08-24 Fernando Perez <fperez@colorado.edu> |
|
1356 | 1360 | |
|
1357 | 1361 | * IPython/Magic.py (Magic.magic_prun): cleanup some dead code. I |
|
1358 | 1362 | tried to switch the profiling to using hotshot, but I'm getting |
|
1359 | 1363 | strange errors from prof.runctx() there. I may be misreading the |
|
1360 | 1364 | docs, but it looks weird. For now the profiling code will |
|
1361 | 1365 | continue to use the standard profiler. |
|
1362 | 1366 | |
|
1363 | 1367 | 2004-08-23 Fernando Perez <fperez@colorado.edu> |
|
1364 | 1368 | |
|
1365 | 1369 | * IPython/Shell.py (IPShellWX.__init__): Improvements to the WX |
|
1366 | 1370 | threaded shell, by John Hunter. It's not quite ready yet, but |
|
1367 | 1371 | close. |
|
1368 | 1372 | |
|
1369 | 1373 | 2004-08-22 Fernando Perez <fperez@colorado.edu> |
|
1370 | 1374 | |
|
1371 | 1375 | * IPython/iplib.py (InteractiveShell.interact): tab cleanups, also |
|
1372 | 1376 | in Magic and ultraTB. |
|
1373 | 1377 | |
|
1374 | 1378 | * ipython.1: document threading options in manpage. |
|
1375 | 1379 | |
|
1376 | 1380 | * scripts/ipython: Changed name of -thread option to -gthread, |
|
1377 | 1381 | since this is GTK specific. I want to leave the door open for a |
|
1378 | 1382 | -wthread option for WX, which will most likely be necessary. This |
|
1379 | 1383 | change affects usage and ipmaker as well. |
|
1380 | 1384 | |
|
1381 | 1385 | * IPython/Shell.py (matplotlib_shell): Add a factory function to |
|
1382 | 1386 | handle the matplotlib shell issues. Code by John Hunter |
|
1383 | 1387 | <jdhunter-AT-nitace.bsd.uchicago.edu>. |
|
1384 | 1388 | (IPShellMatplotlibWX.__init__): Rudimentary WX support. It's |
|
1385 | 1389 | broken (and disabled for end users) for now, but it puts the |
|
1386 | 1390 | infrastructure in place. |
|
1387 | 1391 | |
|
1388 | 1392 | 2004-08-21 Fernando Perez <fperez@colorado.edu> |
|
1389 | 1393 | |
|
1390 | 1394 | * ipythonrc-pylab: Add matplotlib support. |
|
1391 | 1395 | |
|
1392 | 1396 | * matplotlib_config.py: new files for matplotlib support, part of |
|
1393 | 1397 | the pylab profile. |
|
1394 | 1398 | |
|
1395 | 1399 | * IPython/usage.py (__doc__): documented the threading options. |
|
1396 | 1400 | |
|
1397 | 1401 | 2004-08-20 Fernando Perez <fperez@colorado.edu> |
|
1398 | 1402 | |
|
1399 | 1403 | * ipython: Modified the main calling routine to handle the -thread |
|
1400 | 1404 | and -mpthread options. This needs to be done as a top-level hack, |
|
1401 | 1405 | because it determines which class to instantiate for IPython |
|
1402 | 1406 | itself. |
|
1403 | 1407 | |
|
1404 | 1408 | * IPython/Shell.py (MTInteractiveShell.__init__): New set of |
|
1405 | 1409 | classes to support multithreaded GTK operation without blocking, |
|
1406 | 1410 | and matplotlib with all backends. This is a lot of still very |
|
1407 | 1411 | experimental code, and threads are tricky. So it may still have a |
|
1408 | 1412 | few rough edges... This code owes a lot to |
|
1409 | 1413 | http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/65109, by |
|
1410 | 1414 | Brian # McErlean and John Finlay, to Antoon Pardon for fixes, and |
|
1411 | 1415 | to John Hunter for all the matplotlib work. |
|
1412 | 1416 | |
|
1413 | 1417 | * IPython/ipmaker.py (__call__): Added -thread and -mpthread |
|
1414 | 1418 | options for gtk thread and matplotlib support. |
|
1415 | 1419 | |
|
1416 | 1420 | 2004-08-16 Fernando Perez <fperez@colorado.edu> |
|
1417 | 1421 | |
|
1418 | 1422 | * IPython/iplib.py (InteractiveShell.__init__): don't trigger |
|
1419 | 1423 | autocall for things like p*q,p/q,p+q,p-q, when p is callable. Bug |
|
1420 | 1424 | reported by Stephen Walton <stephen.walton-AT-csun.edu>. |
|
1421 | 1425 | |
|
1422 | 1426 | 2004-08-11 Fernando Perez <fperez@colorado.edu> |
|
1423 | 1427 | |
|
1424 | 1428 | * setup.py (isfile): Fix build so documentation gets updated for |
|
1425 | 1429 | rpms (it was only done for .tgz builds). |
|
1426 | 1430 | |
|
1427 | 1431 | 2004-08-10 Fernando Perez <fperez@colorado.edu> |
|
1428 | 1432 | |
|
1429 | 1433 | * genutils.py (Term): Fix misspell of stdin stream (sin->cin). |
|
1430 | 1434 | |
|
1431 | 1435 | * iplib.py : Silence syntax error exceptions in tab-completion. |
|
1432 | 1436 | |
|
1433 | 1437 | 2004-08-05 Fernando Perez <fperez@colorado.edu> |
|
1434 | 1438 | |
|
1435 | 1439 | * IPython/Prompts.py (Prompt2.set_colors): Fix incorrectly set |
|
1436 | 1440 | 'color off' mark for continuation prompts. This was causing long |
|
1437 | 1441 | continuation lines to mis-wrap. |
|
1438 | 1442 | |
|
1439 | 1443 | 2004-08-01 Fernando Perez <fperez@colorado.edu> |
|
1440 | 1444 | |
|
1441 | 1445 | * IPython/ipmaker.py (make_IPython): Allow the shell class used |
|
1442 | 1446 | for building ipython to be a parameter. All this is necessary |
|
1443 | 1447 | right now to have a multithreaded version, but this insane |
|
1444 | 1448 | non-design will be cleaned up soon. For now, it's a hack that |
|
1445 | 1449 | works. |
|
1446 | 1450 | |
|
1447 | 1451 | * IPython/Shell.py (IPShell.__init__): Stop using mutable default |
|
1448 | 1452 | args in various places. No bugs so far, but it's a dangerous |
|
1449 | 1453 | practice. |
|
1450 | 1454 | |
|
1451 | 1455 | 2004-07-31 Fernando Perez <fperez@colorado.edu> |
|
1452 | 1456 | |
|
1453 | 1457 | * IPython/iplib.py (complete): ignore SyntaxError exceptions to |
|
1454 | 1458 | fix completion of files with dots in their names under most |
|
1455 | 1459 | profiles (pysh was OK because the completion order is different). |
|
1456 | 1460 | |
|
1457 | 1461 | 2004-07-27 Fernando Perez <fperez@colorado.edu> |
|
1458 | 1462 | |
|
1459 | 1463 | * IPython/iplib.py (InteractiveShell.__init__): build dict of |
|
1460 | 1464 | keywords manually, b/c the one in keyword.py was removed in python |
|
1461 | 1465 | 2.4. Patch by Anakim Border <aborder-AT-users.sourceforge.net>. |
|
1462 | 1466 | This is NOT a bug under python 2.3 and earlier. |
|
1463 | 1467 | |
|
1464 | 1468 | 2004-07-26 Fernando Perez <fperez@colorado.edu> |
|
1465 | 1469 | |
|
1466 | 1470 | * IPython/ultraTB.py (VerboseTB.text): Add another |
|
1467 | 1471 | linecache.checkcache() call to try to prevent inspect.py from |
|
1468 | 1472 | crashing under python 2.3. I think this fixes |
|
1469 | 1473 | http://www.scipy.net/roundup/ipython/issue17. |
|
1470 | 1474 | |
|
1471 | 1475 | 2004-07-26 *** Released version 0.6.2 |
|
1472 | 1476 | |
|
1473 | 1477 | 2004-07-26 Fernando Perez <fperez@colorado.edu> |
|
1474 | 1478 | |
|
1475 | 1479 | * IPython/Magic.py (Magic.magic_cd): Fix bug where 'cd -N' would |
|
1476 | 1480 | fail for any number. |
|
1477 | 1481 | (Magic.magic_bookmark): Fix bug where 'bookmark -l' would fail for |
|
1478 | 1482 | empty bookmarks. |
|
1479 | 1483 | |
|
1480 | 1484 | 2004-07-26 *** Released version 0.6.1 |
|
1481 | 1485 | |
|
1482 | 1486 | 2004-07-26 Fernando Perez <fperez@colorado.edu> |
|
1483 | 1487 | |
|
1484 | 1488 | * ipython_win_post_install.py (run): Added pysh shortcut for Windows. |
|
1485 | 1489 | |
|
1486 | 1490 | * IPython/iplib.py (protect_filename): Applied Ville's patch for |
|
1487 | 1491 | escaping '()[]{}' in filenames. |
|
1488 | 1492 | |
|
1489 | 1493 | * IPython/Magic.py (shlex_split): Fix handling of '*' and '?' for |
|
1490 | 1494 | Python 2.2 users who lack a proper shlex.split. |
|
1491 | 1495 | |
|
1492 | 1496 | 2004-07-19 Fernando Perez <fperez@colorado.edu> |
|
1493 | 1497 | |
|
1494 | 1498 | * IPython/iplib.py (InteractiveShell.init_readline): Add support |
|
1495 | 1499 | for reading readline's init file. I follow the normal chain: |
|
1496 | 1500 | $INPUTRC is honored, otherwise ~/.inputrc is used. Thanks to a |
|
1497 | 1501 | report by Mike Heeter. This closes |
|
1498 | 1502 | http://www.scipy.net/roundup/ipython/issue16. |
|
1499 | 1503 | |
|
1500 | 1504 | 2004-07-18 Fernando Perez <fperez@colorado.edu> |
|
1501 | 1505 | |
|
1502 | 1506 | * IPython/iplib.py (__init__): Add better handling of '\' under |
|
1503 | 1507 | Win32 for filenames. After a patch by Ville. |
|
1504 | 1508 | |
|
1505 | 1509 | 2004-07-17 Fernando Perez <fperez@colorado.edu> |
|
1506 | 1510 | |
|
1507 | 1511 | * IPython/iplib.py (InteractiveShell._prefilter): fix bug where |
|
1508 | 1512 | autocalling would be triggered for 'foo is bar' if foo is |
|
1509 | 1513 | callable. I also cleaned up the autocall detection code to use a |
|
1510 | 1514 | regexp, which is faster. Bug reported by Alexander Schmolck. |
|
1511 | 1515 | |
|
1512 | 1516 | * IPython/Magic.py (Magic.magic_pinfo): Fix bug where strings with |
|
1513 | 1517 | '?' in them would confuse the help system. Reported by Alex |
|
1514 | 1518 | Schmolck. |
|
1515 | 1519 | |
|
1516 | 1520 | 2004-07-16 Fernando Perez <fperez@colorado.edu> |
|
1517 | 1521 | |
|
1518 | 1522 | * IPython/GnuplotInteractive.py (__all__): added plot2. |
|
1519 | 1523 | |
|
1520 | 1524 | * IPython/Gnuplot2.py (Gnuplot.plot2): added new function for |
|
1521 | 1525 | plotting dictionaries, lists or tuples of 1d arrays. |
|
1522 | 1526 | |
|
1523 | 1527 | * IPython/Magic.py (Magic.magic_hist): small clenaups and |
|
1524 | 1528 | optimizations. |
|
1525 | 1529 | |
|
1526 | 1530 | * IPython/iplib.py:Remove old Changelog info for cleanup. This is |
|
1527 | 1531 | the information which was there from Janko's original IPP code: |
|
1528 | 1532 | |
|
1529 | 1533 | 03.05.99 20:53 porto.ifm.uni-kiel.de |
|
1530 | 1534 | --Started changelog. |
|
1531 | 1535 | --make clear do what it say it does |
|
1532 | 1536 | --added pretty output of lines from inputcache |
|
1533 | 1537 | --Made Logger a mixin class, simplifies handling of switches |
|
1534 | 1538 | --Added own completer class. .string<TAB> expands to last history |
|
1535 | 1539 | line which starts with string. The new expansion is also present |
|
1536 | 1540 | with Ctrl-r from the readline library. But this shows, who this |
|
1537 | 1541 | can be done for other cases. |
|
1538 | 1542 | --Added convention that all shell functions should accept a |
|
1539 | 1543 | parameter_string This opens the door for different behaviour for |
|
1540 | 1544 | each function. @cd is a good example of this. |
|
1541 | 1545 | |
|
1542 | 1546 | 04.05.99 12:12 porto.ifm.uni-kiel.de |
|
1543 | 1547 | --added logfile rotation |
|
1544 | 1548 | --added new mainloop method which freezes first the namespace |
|
1545 | 1549 | |
|
1546 | 1550 | 07.05.99 21:24 porto.ifm.uni-kiel.de |
|
1547 | 1551 | --added the docreader classes. Now there is a help system. |
|
1548 | 1552 | -This is only a first try. Currently it's not easy to put new |
|
1549 | 1553 | stuff in the indices. But this is the way to go. Info would be |
|
1550 | 1554 | better, but HTML is every where and not everybody has an info |
|
1551 | 1555 | system installed and it's not so easy to change html-docs to info. |
|
1552 | 1556 | --added global logfile option |
|
1553 | 1557 | --there is now a hook for object inspection method pinfo needs to |
|
1554 | 1558 | be provided for this. Can be reached by two '??'. |
|
1555 | 1559 | |
|
1556 | 1560 | 08.05.99 20:51 porto.ifm.uni-kiel.de |
|
1557 | 1561 | --added a README |
|
1558 | 1562 | --bug in rc file. Something has changed so functions in the rc |
|
1559 | 1563 | file need to reference the shell and not self. Not clear if it's a |
|
1560 | 1564 | bug or feature. |
|
1561 | 1565 | --changed rc file for new behavior |
|
1562 | 1566 | |
|
1563 | 1567 | 2004-07-15 Fernando Perez <fperez@colorado.edu> |
|
1564 | 1568 | |
|
1565 | 1569 | * IPython/Logger.py (Logger.log): fixed recent bug where the input |
|
1566 | 1570 | cache was falling out of sync in bizarre manners when multi-line |
|
1567 | 1571 | input was present. Minor optimizations and cleanup. |
|
1568 | 1572 | |
|
1569 | 1573 | (Logger): Remove old Changelog info for cleanup. This is the |
|
1570 | 1574 | information which was there from Janko's original code: |
|
1571 | 1575 | |
|
1572 | 1576 | Changes to Logger: - made the default log filename a parameter |
|
1573 | 1577 | |
|
1574 | 1578 | - put a check for lines beginning with !@? in log(). Needed |
|
1575 | 1579 | (even if the handlers properly log their lines) for mid-session |
|
1576 | 1580 | logging activation to work properly. Without this, lines logged |
|
1577 | 1581 | in mid session, which get read from the cache, would end up |
|
1578 | 1582 | 'bare' (with !@? in the open) in the log. Now they are caught |
|
1579 | 1583 | and prepended with a #. |
|
1580 | 1584 | |
|
1581 | 1585 | * IPython/iplib.py (InteractiveShell.init_readline): added check |
|
1582 | 1586 | in case MagicCompleter fails to be defined, so we don't crash. |
|
1583 | 1587 | |
|
1584 | 1588 | 2004-07-13 Fernando Perez <fperez@colorado.edu> |
|
1585 | 1589 | |
|
1586 | 1590 | * IPython/Gnuplot2.py (Gnuplot.hardcopy): add automatic generation |
|
1587 | 1591 | of EPS if the requested filename ends in '.eps'. |
|
1588 | 1592 | |
|
1589 | 1593 | 2004-07-04 Fernando Perez <fperez@colorado.edu> |
|
1590 | 1594 | |
|
1591 | 1595 | * IPython/iplib.py (InteractiveShell.handle_shell_escape): Fix |
|
1592 | 1596 | escaping of quotes when calling the shell. |
|
1593 | 1597 | |
|
1594 | 1598 | 2004-07-02 Fernando Perez <fperez@colorado.edu> |
|
1595 | 1599 | |
|
1596 | 1600 | * IPython/Prompts.py (CachedOutput.update): Fix problem with |
|
1597 | 1601 | gettext not working because we were clobbering '_'. Fixes |
|
1598 | 1602 | http://www.scipy.net/roundup/ipython/issue6. |
|
1599 | 1603 | |
|
1600 | 1604 | 2004-07-01 Fernando Perez <fperez@colorado.edu> |
|
1601 | 1605 | |
|
1602 | 1606 | * IPython/Magic.py (Magic.magic_cd): integrated bookmark handling |
|
1603 | 1607 | into @cd. Patch by Ville. |
|
1604 | 1608 | |
|
1605 | 1609 | * IPython/iplib.py (InteractiveShell.post_config_initialization): |
|
1606 | 1610 | new function to store things after ipmaker runs. Patch by Ville. |
|
1607 | 1611 | Eventually this will go away once ipmaker is removed and the class |
|
1608 | 1612 | gets cleaned up, but for now it's ok. Key functionality here is |
|
1609 | 1613 | the addition of the persistent storage mechanism, a dict for |
|
1610 | 1614 | keeping data across sessions (for now just bookmarks, but more can |
|
1611 | 1615 | be implemented later). |
|
1612 | 1616 | |
|
1613 | 1617 | * IPython/Magic.py (Magic.magic_bookmark): New bookmark system, |
|
1614 | 1618 | persistent across sections. Patch by Ville, I modified it |
|
1615 | 1619 | soemwhat to allow bookmarking arbitrary dirs other than CWD. Also |
|
1616 | 1620 | added a '-l' option to list all bookmarks. |
|
1617 | 1621 | |
|
1618 | 1622 | * IPython/iplib.py (InteractiveShell.atexit_operations): new |
|
1619 | 1623 | center for cleanup. Registered with atexit.register(). I moved |
|
1620 | 1624 | here the old exit_cleanup(). After a patch by Ville. |
|
1621 | 1625 | |
|
1622 | 1626 | * IPython/Magic.py (get_py_filename): added '~' to the accepted |
|
1623 | 1627 | characters in the hacked shlex_split for python 2.2. |
|
1624 | 1628 | |
|
1625 | 1629 | * IPython/iplib.py (file_matches): more fixes to filenames with |
|
1626 | 1630 | whitespace in them. It's not perfect, but limitations in python's |
|
1627 | 1631 | readline make it impossible to go further. |
|
1628 | 1632 | |
|
1629 | 1633 | 2004-06-29 Fernando Perez <fperez@colorado.edu> |
|
1630 | 1634 | |
|
1631 | 1635 | * IPython/iplib.py (file_matches): escape whitespace correctly in |
|
1632 | 1636 | filename completions. Bug reported by Ville. |
|
1633 | 1637 | |
|
1634 | 1638 | 2004-06-28 Fernando Perez <fperez@colorado.edu> |
|
1635 | 1639 | |
|
1636 | 1640 | * IPython/ipmaker.py (__call__): Added per-profile histories. Now |
|
1637 | 1641 | the history file will be called 'history-PROFNAME' (or just |
|
1638 | 1642 | 'history' if no profile is loaded). I was getting annoyed at |
|
1639 | 1643 | getting my Numerical work history clobbered by pysh sessions. |
|
1640 | 1644 | |
|
1641 | 1645 | * IPython/iplib.py (InteractiveShell.__init__): Internal |
|
1642 | 1646 | getoutputerror() function so that we can honor the system_verbose |
|
1643 | 1647 | flag for _all_ system calls. I also added escaping of # |
|
1644 | 1648 | characters here to avoid confusing Itpl. |
|
1645 | 1649 | |
|
1646 | 1650 | * IPython/Magic.py (shlex_split): removed call to shell in |
|
1647 | 1651 | parse_options and replaced it with shlex.split(). The annoying |
|
1648 | 1652 | part was that in Python 2.2, shlex.split() doesn't exist, so I had |
|
1649 | 1653 | to backport it from 2.3, with several frail hacks (the shlex |
|
1650 | 1654 | module is rather limited in 2.2). Thanks to a suggestion by Ville |
|
1651 | 1655 | Vainio <vivainio@kolumbus.fi>. For Python 2.3 there should be no |
|
1652 | 1656 | problem. |
|
1653 | 1657 | |
|
1654 | 1658 | (Magic.magic_system_verbose): new toggle to print the actual |
|
1655 | 1659 | system calls made by ipython. Mainly for debugging purposes. |
|
1656 | 1660 | |
|
1657 | 1661 | * IPython/GnuplotRuntime.py (gnu_out): fix bug for cygwin, which |
|
1658 | 1662 | doesn't support persistence. Reported (and fix suggested) by |
|
1659 | 1663 | Travis Caldwell <travis_caldwell2000@yahoo.com>. |
|
1660 | 1664 | |
|
1661 | 1665 | 2004-06-26 Fernando Perez <fperez@colorado.edu> |
|
1662 | 1666 | |
|
1663 | 1667 | * IPython/Logger.py (Logger.log): fix to handle correctly empty |
|
1664 | 1668 | continue prompts. |
|
1665 | 1669 | |
|
1666 | 1670 | * IPython/Extensions/InterpreterExec.py (pysh): moved the pysh() |
|
1667 | 1671 | function (basically a big docstring) and a few more things here to |
|
1668 | 1672 | speedup startup. pysh.py is now very lightweight. We want because |
|
1669 | 1673 | it gets execfile'd, while InterpreterExec gets imported, so |
|
1670 | 1674 | byte-compilation saves time. |
|
1671 | 1675 | |
|
1672 | 1676 | 2004-06-25 Fernando Perez <fperez@colorado.edu> |
|
1673 | 1677 | |
|
1674 | 1678 | * IPython/Magic.py (Magic.magic_cd): Fixed to restore usage of 'cd |
|
1675 | 1679 | -NUM', which was recently broken. |
|
1676 | 1680 | |
|
1677 | 1681 | * IPython/iplib.py (InteractiveShell.handle_shell_escape): allow ! |
|
1678 | 1682 | in multi-line input (but not !!, which doesn't make sense there). |
|
1679 | 1683 | |
|
1680 | 1684 | * IPython/UserConfig/ipythonrc: made autoindent on by default. |
|
1681 | 1685 | It's just too useful, and people can turn it off in the less |
|
1682 | 1686 | common cases where it's a problem. |
|
1683 | 1687 | |
|
1684 | 1688 | 2004-06-24 Fernando Perez <fperez@colorado.edu> |
|
1685 | 1689 | |
|
1686 | 1690 | * IPython/iplib.py (InteractiveShell._prefilter): big change - |
|
1687 | 1691 | special syntaxes (like alias calling) is now allied in multi-line |
|
1688 | 1692 | input. This is still _very_ experimental, but it's necessary for |
|
1689 | 1693 | efficient shell usage combining python looping syntax with system |
|
1690 | 1694 | calls. For now it's restricted to aliases, I don't think it |
|
1691 | 1695 | really even makes sense to have this for magics. |
|
1692 | 1696 | |
|
1693 | 1697 | 2004-06-23 Fernando Perez <fperez@colorado.edu> |
|
1694 | 1698 | |
|
1695 | 1699 | * IPython/Extensions/InterpreterExec.py (prefilter_shell): Added |
|
1696 | 1700 | $var=cmd <=> @sc var=cmd and $$var=cmd <=> @sc -l var=cmd. |
|
1697 | 1701 | |
|
1698 | 1702 | * IPython/Magic.py (Magic.magic_rehashx): modified to handle |
|
1699 | 1703 | extensions under Windows (after code sent by Gary Bishop). The |
|
1700 | 1704 | extensions considered 'executable' are stored in IPython's rc |
|
1701 | 1705 | structure as win_exec_ext. |
|
1702 | 1706 | |
|
1703 | 1707 | * IPython/genutils.py (shell): new function, like system() but |
|
1704 | 1708 | without return value. Very useful for interactive shell work. |
|
1705 | 1709 | |
|
1706 | 1710 | * IPython/Magic.py (Magic.magic_unalias): New @unalias function to |
|
1707 | 1711 | delete aliases. |
|
1708 | 1712 | |
|
1709 | 1713 | * IPython/iplib.py (InteractiveShell.alias_table_update): make |
|
1710 | 1714 | sure that the alias table doesn't contain python keywords. |
|
1711 | 1715 | |
|
1712 | 1716 | 2004-06-21 Fernando Perez <fperez@colorado.edu> |
|
1713 | 1717 | |
|
1714 | 1718 | * IPython/Magic.py (Magic.magic_rehash): Fix crash when |
|
1715 | 1719 | non-existent items are found in $PATH. Reported by Thorsten. |
|
1716 | 1720 | |
|
1717 | 1721 | 2004-06-20 Fernando Perez <fperez@colorado.edu> |
|
1718 | 1722 | |
|
1719 | 1723 | * IPython/iplib.py (complete): modified the completer so that the |
|
1720 | 1724 | order of priorities can be easily changed at runtime. |
|
1721 | 1725 | |
|
1722 | 1726 | * IPython/Extensions/InterpreterExec.py (prefilter_shell): |
|
1723 | 1727 | Modified to auto-execute all lines beginning with '~', '/' or '.'. |
|
1724 | 1728 | |
|
1725 | 1729 | * IPython/Magic.py (Magic.magic_sx): modified @sc and @sx to |
|
1726 | 1730 | expand Python variables prepended with $ in all system calls. The |
|
1727 | 1731 | same was done to InteractiveShell.handle_shell_escape. Now all |
|
1728 | 1732 | system access mechanisms (!, !!, @sc, @sx and aliases) allow the |
|
1729 | 1733 | expansion of python variables and expressions according to the |
|
1730 | 1734 | syntax of PEP-215 - http://www.python.org/peps/pep-0215.html. |
|
1731 | 1735 | |
|
1732 | 1736 | Though PEP-215 has been rejected, a similar (but simpler) one |
|
1733 | 1737 | seems like it will go into Python 2.4, PEP-292 - |
|
1734 | 1738 | http://www.python.org/peps/pep-0292.html. |
|
1735 | 1739 | |
|
1736 | 1740 | I'll keep the full syntax of PEP-215, since IPython has since the |
|
1737 | 1741 | start used Ka-Ping Yee's reference implementation discussed there |
|
1738 | 1742 | (Itpl), and I actually like the powerful semantics it offers. |
|
1739 | 1743 | |
|
1740 | 1744 | In order to access normal shell variables, the $ has to be escaped |
|
1741 | 1745 | via an extra $. For example: |
|
1742 | 1746 | |
|
1743 | 1747 | In [7]: PATH='a python variable' |
|
1744 | 1748 | |
|
1745 | 1749 | In [8]: !echo $PATH |
|
1746 | 1750 | a python variable |
|
1747 | 1751 | |
|
1748 | 1752 | In [9]: !echo $$PATH |
|
1749 | 1753 | /usr/local/lf9560/bin:/usr/local/intel/compiler70/ia32/bin:... |
|
1750 | 1754 | |
|
1751 | 1755 | (Magic.parse_options): escape $ so the shell doesn't evaluate |
|
1752 | 1756 | things prematurely. |
|
1753 | 1757 | |
|
1754 | 1758 | * IPython/iplib.py (InteractiveShell.call_alias): added the |
|
1755 | 1759 | ability for aliases to expand python variables via $. |
|
1756 | 1760 | |
|
1757 | 1761 | * IPython/Magic.py (Magic.magic_rehash): based on the new alias |
|
1758 | 1762 | system, now there's a @rehash/@rehashx pair of magics. These work |
|
1759 | 1763 | like the csh rehash command, and can be invoked at any time. They |
|
1760 | 1764 | build a table of aliases to everything in the user's $PATH |
|
1761 | 1765 | (@rehash uses everything, @rehashx is slower but only adds |
|
1762 | 1766 | executable files). With this, the pysh.py-based shell profile can |
|
1763 | 1767 | now simply call rehash upon startup, and full access to all |
|
1764 | 1768 | programs in the user's path is obtained. |
|
1765 | 1769 | |
|
1766 | 1770 | * IPython/iplib.py (InteractiveShell.call_alias): The new alias |
|
1767 | 1771 | functionality is now fully in place. I removed the old dynamic |
|
1768 | 1772 | code generation based approach, in favor of a much lighter one |
|
1769 | 1773 | based on a simple dict. The advantage is that this allows me to |
|
1770 | 1774 | now have thousands of aliases with negligible cost (unthinkable |
|
1771 | 1775 | with the old system). |
|
1772 | 1776 | |
|
1773 | 1777 | 2004-06-19 Fernando Perez <fperez@colorado.edu> |
|
1774 | 1778 | |
|
1775 | 1779 | * IPython/iplib.py (__init__): extended MagicCompleter class to |
|
1776 | 1780 | also complete (last in priority) on user aliases. |
|
1777 | 1781 | |
|
1778 | 1782 | * IPython/Itpl.py (Itpl.__str__): fixed order of globals/locals in |
|
1779 | 1783 | call to eval. |
|
1780 | 1784 | (ItplNS.__init__): Added a new class which functions like Itpl, |
|
1781 | 1785 | but allows configuring the namespace for the evaluation to occur |
|
1782 | 1786 | in. |
|
1783 | 1787 | |
|
1784 | 1788 | 2004-06-18 Fernando Perez <fperez@colorado.edu> |
|
1785 | 1789 | |
|
1786 | 1790 | * IPython/iplib.py (InteractiveShell.runcode): modify to print a |
|
1787 | 1791 | better message when 'exit' or 'quit' are typed (a common newbie |
|
1788 | 1792 | confusion). |
|
1789 | 1793 | |
|
1790 | 1794 | * IPython/Magic.py (Magic.magic_colors): Added the runtime color |
|
1791 | 1795 | check for Windows users. |
|
1792 | 1796 | |
|
1793 | 1797 | * IPython/iplib.py (InteractiveShell.user_setup): removed |
|
1794 | 1798 | disabling of colors for Windows. I'll test at runtime and issue a |
|
1795 | 1799 | warning if Gary's readline isn't found, as to nudge users to |
|
1796 | 1800 | download it. |
|
1797 | 1801 | |
|
1798 | 1802 | 2004-06-16 Fernando Perez <fperez@colorado.edu> |
|
1799 | 1803 | |
|
1800 | 1804 | * IPython/genutils.py (Stream.__init__): changed to print errors |
|
1801 | 1805 | to sys.stderr. I had a circular dependency here. Now it's |
|
1802 | 1806 | possible to run ipython as IDLE's shell (consider this pre-alpha, |
|
1803 | 1807 | since true stdout things end up in the starting terminal instead |
|
1804 | 1808 | of IDLE's out). |
|
1805 | 1809 | |
|
1806 | 1810 | * IPython/Prompts.py (Prompt2.set_colors): prevent crashes for |
|
1807 | 1811 | users who haven't # updated their prompt_in2 definitions. Remove |
|
1808 | 1812 | eventually. |
|
1809 | 1813 | (multiple_replace): added credit to original ASPN recipe. |
|
1810 | 1814 | |
|
1811 | 1815 | 2004-06-15 Fernando Perez <fperez@colorado.edu> |
|
1812 | 1816 | |
|
1813 | 1817 | * IPython/iplib.py (InteractiveShell.__init__): add 'cp' to the |
|
1814 | 1818 | list of auto-defined aliases. |
|
1815 | 1819 | |
|
1816 | 1820 | 2004-06-13 Fernando Perez <fperez@colorado.edu> |
|
1817 | 1821 | |
|
1818 | 1822 | * setup.py (scriptfiles): Don't trigger win_post_install unless an |
|
1819 | 1823 | install was really requested (so setup.py can be used for other |
|
1820 | 1824 | things under Windows). |
|
1821 | 1825 | |
|
1822 | 1826 | 2004-06-10 Fernando Perez <fperez@colorado.edu> |
|
1823 | 1827 | |
|
1824 | 1828 | * IPython/Logger.py (Logger.create_log): Manually remove any old |
|
1825 | 1829 | backup, since os.remove may fail under Windows. Fixes bug |
|
1826 | 1830 | reported by Thorsten. |
|
1827 | 1831 | |
|
1828 | 1832 | 2004-06-09 Fernando Perez <fperez@colorado.edu> |
|
1829 | 1833 | |
|
1830 | 1834 | * examples/example-embed.py: fixed all references to %n (replaced |
|
1831 | 1835 | with \\# for ps1/out prompts and with \\D for ps2 prompts). Done |
|
1832 | 1836 | for all examples and the manual as well. |
|
1833 | 1837 | |
|
1834 | 1838 | 2004-06-08 Fernando Perez <fperez@colorado.edu> |
|
1835 | 1839 | |
|
1836 | 1840 | * IPython/Prompts.py (Prompt2.set_p_str): fixed all prompt |
|
1837 | 1841 | alignment and color management. All 3 prompt subsystems now |
|
1838 | 1842 | inherit from BasePrompt. |
|
1839 | 1843 | |
|
1840 | 1844 | * tools/release: updates for windows installer build and tag rpms |
|
1841 | 1845 | with python version (since paths are fixed). |
|
1842 | 1846 | |
|
1843 | 1847 | * IPython/UserConfig/ipythonrc: modified to use \# instead of %n, |
|
1844 | 1848 | which will become eventually obsolete. Also fixed the default |
|
1845 | 1849 | prompt_in2 to use \D, so at least new users start with the correct |
|
1846 | 1850 | defaults. |
|
1847 | 1851 | WARNING: Users with existing ipythonrc files will need to apply |
|
1848 | 1852 | this fix manually! |
|
1849 | 1853 | |
|
1850 | 1854 | * setup.py: make windows installer (.exe). This is finally the |
|
1851 | 1855 | integration of an old patch by Cory Dodt <dodt-AT-fcoe.k12.ca.us>, |
|
1852 | 1856 | which I hadn't included because it required Python 2.3 (or recent |
|
1853 | 1857 | distutils). |
|
1854 | 1858 | |
|
1855 | 1859 | * IPython/usage.py (__doc__): update docs (and manpage) to reflect |
|
1856 | 1860 | usage of new '\D' escape. |
|
1857 | 1861 | |
|
1858 | 1862 | * IPython/Prompts.py (ROOT_SYMBOL): Small fix for Windows (which |
|
1859 | 1863 | lacks os.getuid()) |
|
1860 | 1864 | (CachedOutput.set_colors): Added the ability to turn coloring |
|
1861 | 1865 | on/off with @colors even for manually defined prompt colors. It |
|
1862 | 1866 | uses a nasty global, but it works safely and via the generic color |
|
1863 | 1867 | handling mechanism. |
|
1864 | 1868 | (Prompt2.__init__): Introduced new escape '\D' for continuation |
|
1865 | 1869 | prompts. It represents the counter ('\#') as dots. |
|
1866 | 1870 | *** NOTE *** THIS IS A BACKWARDS-INCOMPATIBLE CHANGE. Users will |
|
1867 | 1871 | need to update their ipythonrc files and replace '%n' with '\D' in |
|
1868 | 1872 | their prompt_in2 settings everywhere. Sorry, but there's |
|
1869 | 1873 | otherwise no clean way to get all prompts to properly align. The |
|
1870 | 1874 | ipythonrc shipped with IPython has been updated. |
|
1871 | 1875 | |
|
1872 | 1876 | 2004-06-07 Fernando Perez <fperez@colorado.edu> |
|
1873 | 1877 | |
|
1874 | 1878 | * setup.py (isfile): Pass local_icons option to latex2html, so the |
|
1875 | 1879 | resulting HTML file is self-contained. Thanks to |
|
1876 | 1880 | dryice-AT-liu.com.cn for the tip. |
|
1877 | 1881 | |
|
1878 | 1882 | * pysh.py: I created a new profile 'shell', which implements a |
|
1879 | 1883 | _rudimentary_ IPython-based shell. This is in NO WAY a realy |
|
1880 | 1884 | system shell, nor will it become one anytime soon. It's mainly |
|
1881 | 1885 | meant to illustrate the use of the new flexible bash-like prompts. |
|
1882 | 1886 | I guess it could be used by hardy souls for true shell management, |
|
1883 | 1887 | but it's no tcsh/bash... pysh.py is loaded by the 'shell' |
|
1884 | 1888 | profile. This uses the InterpreterExec extension provided by |
|
1885 | 1889 | W.J. van der Laan <gnufnork-AT-hetdigitalegat.nl> |
|
1886 | 1890 | |
|
1887 | 1891 | * IPython/Prompts.py (PromptOut.__str__): now it will correctly |
|
1888 | 1892 | auto-align itself with the length of the previous input prompt |
|
1889 | 1893 | (taking into account the invisible color escapes). |
|
1890 | 1894 | (CachedOutput.__init__): Large restructuring of this class. Now |
|
1891 | 1895 | all three prompts (primary1, primary2, output) are proper objects, |
|
1892 | 1896 | managed by the 'parent' CachedOutput class. The code is still a |
|
1893 | 1897 | bit hackish (all prompts share state via a pointer to the cache), |
|
1894 | 1898 | but it's overall far cleaner than before. |
|
1895 | 1899 | |
|
1896 | 1900 | * IPython/genutils.py (getoutputerror): modified to add verbose, |
|
1897 | 1901 | debug and header options. This makes the interface of all getout* |
|
1898 | 1902 | functions uniform. |
|
1899 | 1903 | (SystemExec.getoutputerror): added getoutputerror to SystemExec. |
|
1900 | 1904 | |
|
1901 | 1905 | * IPython/Magic.py (Magic.default_option): added a function to |
|
1902 | 1906 | allow registering default options for any magic command. This |
|
1903 | 1907 | makes it easy to have profiles which customize the magics globally |
|
1904 | 1908 | for a certain use. The values set through this function are |
|
1905 | 1909 | picked up by the parse_options() method, which all magics should |
|
1906 | 1910 | use to parse their options. |
|
1907 | 1911 | |
|
1908 | 1912 | * IPython/genutils.py (warn): modified the warnings framework to |
|
1909 | 1913 | use the Term I/O class. I'm trying to slowly unify all of |
|
1910 | 1914 | IPython's I/O operations to pass through Term. |
|
1911 | 1915 | |
|
1912 | 1916 | * IPython/Prompts.py (Prompt2._str_other): Added functionality in |
|
1913 | 1917 | the secondary prompt to correctly match the length of the primary |
|
1914 | 1918 | one for any prompt. Now multi-line code will properly line up |
|
1915 | 1919 | even for path dependent prompts, such as the new ones available |
|
1916 | 1920 | via the prompt_specials. |
|
1917 | 1921 | |
|
1918 | 1922 | 2004-06-06 Fernando Perez <fperez@colorado.edu> |
|
1919 | 1923 | |
|
1920 | 1924 | * IPython/Prompts.py (prompt_specials): Added the ability to have |
|
1921 | 1925 | bash-like special sequences in the prompts, which get |
|
1922 | 1926 | automatically expanded. Things like hostname, current working |
|
1923 | 1927 | directory and username are implemented already, but it's easy to |
|
1924 | 1928 | add more in the future. Thanks to a patch by W.J. van der Laan |
|
1925 | 1929 | <gnufnork-AT-hetdigitalegat.nl> |
|
1926 | 1930 | (prompt_specials): Added color support for prompt strings, so |
|
1927 | 1931 | users can define arbitrary color setups for their prompts. |
|
1928 | 1932 | |
|
1929 | 1933 | 2004-06-05 Fernando Perez <fperez@colorado.edu> |
|
1930 | 1934 | |
|
1931 | 1935 | * IPython/genutils.py (Term.reopen_all): Added Windows-specific |
|
1932 | 1936 | code to load Gary Bishop's readline and configure it |
|
1933 | 1937 | automatically. Thanks to Gary for help on this. |
|
1934 | 1938 | |
|
1935 | 1939 | 2004-06-01 Fernando Perez <fperez@colorado.edu> |
|
1936 | 1940 | |
|
1937 | 1941 | * IPython/Logger.py (Logger.create_log): fix bug for logging |
|
1938 | 1942 | with no filename (previous fix was incomplete). |
|
1939 | 1943 | |
|
1940 | 1944 | 2004-05-25 Fernando Perez <fperez@colorado.edu> |
|
1941 | 1945 | |
|
1942 | 1946 | * IPython/Magic.py (Magic.parse_options): fix bug where naked |
|
1943 | 1947 | parens would get passed to the shell. |
|
1944 | 1948 | |
|
1945 | 1949 | 2004-05-20 Fernando Perez <fperez@colorado.edu> |
|
1946 | 1950 | |
|
1947 | 1951 | * IPython/Magic.py (Magic.magic_prun): changed default profile |
|
1948 | 1952 | sort order to 'time' (the more common profiling need). |
|
1949 | 1953 | |
|
1950 | 1954 | * IPython/OInspect.py (Inspector.pinfo): flush the inspect cache |
|
1951 | 1955 | so that source code shown is guaranteed in sync with the file on |
|
1952 | 1956 | disk (also changed in psource). Similar fix to the one for |
|
1953 | 1957 | ultraTB on 2004-05-06. Thanks to a bug report by Yann Le Du |
|
1954 | 1958 | <yann.ledu-AT-noos.fr>. |
|
1955 | 1959 | |
|
1956 | 1960 | * IPython/Magic.py (Magic.parse_options): Fixed bug where commands |
|
1957 | 1961 | with a single option would not be correctly parsed. Closes |
|
1958 | 1962 | http://www.scipy.net/roundup/ipython/issue14. This bug had been |
|
1959 | 1963 | introduced in 0.6.0 (on 2004-05-06). |
|
1960 | 1964 | |
|
1961 | 1965 | 2004-05-13 *** Released version 0.6.0 |
|
1962 | 1966 | |
|
1963 | 1967 | 2004-05-13 Fernando Perez <fperez@colorado.edu> |
|
1964 | 1968 | |
|
1965 | 1969 | * debian/: Added debian/ directory to CVS, so that debian support |
|
1966 | 1970 | is publicly accessible. The debian package is maintained by Jack |
|
1967 | 1971 | Moffit <jack-AT-xiph.org>. |
|
1968 | 1972 | |
|
1969 | 1973 | * Documentation: included the notes about an ipython-based system |
|
1970 | 1974 | shell (the hypothetical 'pysh') into the new_design.pdf document, |
|
1971 | 1975 | so that these ideas get distributed to users along with the |
|
1972 | 1976 | official documentation. |
|
1973 | 1977 | |
|
1974 | 1978 | 2004-05-10 Fernando Perez <fperez@colorado.edu> |
|
1975 | 1979 | |
|
1976 | 1980 | * IPython/Logger.py (Logger.create_log): fix recently introduced |
|
1977 | 1981 | bug (misindented line) where logstart would fail when not given an |
|
1978 | 1982 | explicit filename. |
|
1979 | 1983 | |
|
1980 | 1984 | 2004-05-09 Fernando Perez <fperez@colorado.edu> |
|
1981 | 1985 | |
|
1982 | 1986 | * IPython/Magic.py (Magic.parse_options): skip system call when |
|
1983 | 1987 | there are no options to look for. Faster, cleaner for the common |
|
1984 | 1988 | case. |
|
1985 | 1989 | |
|
1986 | 1990 | * Documentation: many updates to the manual: describing Windows |
|
1987 | 1991 | support better, Gnuplot updates, credits, misc small stuff. Also |
|
1988 | 1992 | updated the new_design doc a bit. |
|
1989 | 1993 | |
|
1990 | 1994 | 2004-05-06 *** Released version 0.6.0.rc1 |
|
1991 | 1995 | |
|
1992 | 1996 | 2004-05-06 Fernando Perez <fperez@colorado.edu> |
|
1993 | 1997 | |
|
1994 | 1998 | * IPython/ultraTB.py (ListTB.text): modified a ton of string += |
|
1995 | 1999 | operations to use the vastly more efficient list/''.join() method. |
|
1996 | 2000 | (FormattedTB.text): Fix |
|
1997 | 2001 | http://www.scipy.net/roundup/ipython/issue12 - exception source |
|
1998 | 2002 | extract not updated after reload. Thanks to Mike Salib |
|
1999 | 2003 | <msalib-AT-mit.edu> for pinning the source of the problem. |
|
2000 | 2004 | Fortunately, the solution works inside ipython and doesn't require |
|
2001 | 2005 | any changes to python proper. |
|
2002 | 2006 | |
|
2003 | 2007 | * IPython/Magic.py (Magic.parse_options): Improved to process the |
|
2004 | 2008 | argument list as a true shell would (by actually using the |
|
2005 | 2009 | underlying system shell). This way, all @magics automatically get |
|
2006 | 2010 | shell expansion for variables. Thanks to a comment by Alex |
|
2007 | 2011 | Schmolck. |
|
2008 | 2012 | |
|
2009 | 2013 | 2004-04-04 Fernando Perez <fperez@colorado.edu> |
|
2010 | 2014 | |
|
2011 | 2015 | * IPython/iplib.py (InteractiveShell.interact): Added a special |
|
2012 | 2016 | trap for a debugger quit exception, which is basically impossible |
|
2013 | 2017 | to handle by normal mechanisms, given what pdb does to the stack. |
|
2014 | 2018 | This fixes a crash reported by <fgibbons-AT-llama.med.harvard.edu>. |
|
2015 | 2019 | |
|
2016 | 2020 | 2004-04-03 Fernando Perez <fperez@colorado.edu> |
|
2017 | 2021 | |
|
2018 | 2022 | * IPython/genutils.py (Term): Standardized the names of the Term |
|
2019 | 2023 | class streams to cin/cout/cerr, following C++ naming conventions |
|
2020 | 2024 | (I can't use in/out/err because 'in' is not a valid attribute |
|
2021 | 2025 | name). |
|
2022 | 2026 | |
|
2023 | 2027 | * IPython/iplib.py (InteractiveShell.interact): don't increment |
|
2024 | 2028 | the prompt if there's no user input. By Daniel 'Dang' Griffith |
|
2025 | 2029 | <pythondev-dang-AT-lazytwinacres.net>, after a suggestion from |
|
2026 | 2030 | Francois Pinard. |
|
2027 | 2031 | |
|
2028 | 2032 | 2004-04-02 Fernando Perez <fperez@colorado.edu> |
|
2029 | 2033 | |
|
2030 | 2034 | * IPython/genutils.py (Stream.__init__): Modified to survive at |
|
2031 | 2035 | least importing in contexts where stdin/out/err aren't true file |
|
2032 | 2036 | objects, such as PyCrust (they lack fileno() and mode). However, |
|
2033 | 2037 | the recovery facilities which rely on these things existing will |
|
2034 | 2038 | not work. |
|
2035 | 2039 | |
|
2036 | 2040 | 2004-04-01 Fernando Perez <fperez@colorado.edu> |
|
2037 | 2041 | |
|
2038 | 2042 | * IPython/Magic.py (Magic.magic_sx): modified (as well as @sc) to |
|
2039 | 2043 | use the new getoutputerror() function, so it properly |
|
2040 | 2044 | distinguishes stdout/err. |
|
2041 | 2045 | |
|
2042 | 2046 | * IPython/genutils.py (getoutputerror): added a function to |
|
2043 | 2047 | capture separately the standard output and error of a command. |
|
2044 | 2048 | After a comment from dang on the mailing lists. This code is |
|
2045 | 2049 | basically a modified version of commands.getstatusoutput(), from |
|
2046 | 2050 | the standard library. |
|
2047 | 2051 | |
|
2048 | 2052 | * IPython/iplib.py (InteractiveShell.handle_shell_escape): added |
|
2049 | 2053 | '!!' as a special syntax (shorthand) to access @sx. |
|
2050 | 2054 | |
|
2051 | 2055 | * IPython/Magic.py (Magic.magic_sx): new magic, to execute a shell |
|
2052 | 2056 | command and return its output as a list split on '\n'. |
|
2053 | 2057 | |
|
2054 | 2058 | 2004-03-31 Fernando Perez <fperez@colorado.edu> |
|
2055 | 2059 | |
|
2056 | 2060 | * IPython/FakeModule.py (FakeModule.__init__): added __nonzero__ |
|
2057 | 2061 | method to dictionaries used as FakeModule instances if they lack |
|
2058 | 2062 | it. At least pydoc in python2.3 breaks for runtime-defined |
|
2059 | 2063 | functions without this hack. At some point I need to _really_ |
|
2060 | 2064 | understand what FakeModule is doing, because it's a gross hack. |
|
2061 | 2065 | But it solves Arnd's problem for now... |
|
2062 | 2066 | |
|
2063 | 2067 | 2004-02-27 Fernando Perez <fperez@colorado.edu> |
|
2064 | 2068 | |
|
2065 | 2069 | * IPython/Logger.py (Logger.create_log): Fix bug where 'rotate' |
|
2066 | 2070 | mode would behave erratically. Also increased the number of |
|
2067 | 2071 | possible logs in rotate mod to 999. Thanks to Rod Holland |
|
2068 | 2072 | <rhh@StructureLABS.com> for the report and fixes. |
|
2069 | 2073 | |
|
2070 | 2074 | 2004-02-26 Fernando Perez <fperez@colorado.edu> |
|
2071 | 2075 | |
|
2072 | 2076 | * IPython/genutils.py (page): Check that the curses module really |
|
2073 | 2077 | has the initscr attribute before trying to use it. For some |
|
2074 | 2078 | reason, the Solaris curses module is missing this. I think this |
|
2075 | 2079 | should be considered a Solaris python bug, but I'm not sure. |
|
2076 | 2080 | |
|
2077 | 2081 | 2004-01-17 Fernando Perez <fperez@colorado.edu> |
|
2078 | 2082 | |
|
2079 | 2083 | * IPython/genutils.py (Stream.__init__): Changes to try to make |
|
2080 | 2084 | ipython robust against stdin/out/err being closed by the user. |
|
2081 | 2085 | This is 'user error' (and blocks a normal python session, at least |
|
2082 | 2086 | the stdout case). However, Ipython should be able to survive such |
|
2083 | 2087 | instances of abuse as gracefully as possible. To simplify the |
|
2084 | 2088 | coding and maintain compatibility with Gary Bishop's Term |
|
2085 | 2089 | contributions, I've made use of classmethods for this. I think |
|
2086 | 2090 | this introduces a dependency on python 2.2. |
|
2087 | 2091 | |
|
2088 | 2092 | 2004-01-13 Fernando Perez <fperez@colorado.edu> |
|
2089 | 2093 | |
|
2090 | 2094 | * IPython/numutils.py (exp_safe): simplified the code a bit and |
|
2091 | 2095 | removed the need for importing the kinds module altogether. |
|
2092 | 2096 | |
|
2093 | 2097 | 2004-01-06 Fernando Perez <fperez@colorado.edu> |
|
2094 | 2098 | |
|
2095 | 2099 | * IPython/Magic.py (Magic.magic_sc): Made the shell capture system |
|
2096 | 2100 | a magic function instead, after some community feedback. No |
|
2097 | 2101 | special syntax will exist for it, but its name is deliberately |
|
2098 | 2102 | very short. |
|
2099 | 2103 | |
|
2100 | 2104 | 2003-12-20 Fernando Perez <fperez@colorado.edu> |
|
2101 | 2105 | |
|
2102 | 2106 | * IPython/iplib.py (InteractiveShell.handle_shell_assign): Added |
|
2103 | 2107 | new functionality, to automagically assign the result of a shell |
|
2104 | 2108 | command to a variable. I'll solicit some community feedback on |
|
2105 | 2109 | this before making it permanent. |
|
2106 | 2110 | |
|
2107 | 2111 | * IPython/OInspect.py (Inspector.pinfo): Fix crash when info was |
|
2108 | 2112 | requested about callables for which inspect couldn't obtain a |
|
2109 | 2113 | proper argspec. Thanks to a crash report sent by Etienne |
|
2110 | 2114 | Posthumus <etienne-AT-apple01.cs.vu.nl>. |
|
2111 | 2115 | |
|
2112 | 2116 | 2003-12-09 Fernando Perez <fperez@colorado.edu> |
|
2113 | 2117 | |
|
2114 | 2118 | * IPython/genutils.py (page): patch for the pager to work across |
|
2115 | 2119 | various versions of Windows. By Gary Bishop. |
|
2116 | 2120 | |
|
2117 | 2121 | 2003-12-04 Fernando Perez <fperez@colorado.edu> |
|
2118 | 2122 | |
|
2119 | 2123 | * IPython/Gnuplot2.py (PlotItems): Fixes for working with |
|
2120 | 2124 | Gnuplot.py version 1.7, whose internal names changed quite a bit. |
|
2121 | 2125 | While I tested this and it looks ok, there may still be corner |
|
2122 | 2126 | cases I've missed. |
|
2123 | 2127 | |
|
2124 | 2128 | 2003-12-01 Fernando Perez <fperez@colorado.edu> |
|
2125 | 2129 | |
|
2126 | 2130 | * IPython/iplib.py (InteractiveShell._prefilter): Fixed a bug |
|
2127 | 2131 | where a line like 'p,q=1,2' would fail because the automagic |
|
2128 | 2132 | system would be triggered for @p. |
|
2129 | 2133 | |
|
2130 | 2134 | * IPython/DPyGetOpt.py (DPyGetOpt.processArguments): Tab-related |
|
2131 | 2135 | cleanups, code unmodified. |
|
2132 | 2136 | |
|
2133 | 2137 | * IPython/genutils.py (Term): added a class for IPython to handle |
|
2134 | 2138 | output. In most cases it will just be a proxy for stdout/err, but |
|
2135 | 2139 | having this allows modifications to be made for some platforms, |
|
2136 | 2140 | such as handling color escapes under Windows. All of this code |
|
2137 | 2141 | was contributed by Gary Bishop, with minor modifications by me. |
|
2138 | 2142 | The actual changes affect many files. |
|
2139 | 2143 | |
|
2140 | 2144 | 2003-11-30 Fernando Perez <fperez@colorado.edu> |
|
2141 | 2145 | |
|
2142 | 2146 | * IPython/iplib.py (file_matches): new completion code, courtesy |
|
2143 | 2147 | of Jeff Collins. This enables filename completion again under |
|
2144 | 2148 | python 2.3, which disabled it at the C level. |
|
2145 | 2149 | |
|
2146 | 2150 | 2003-11-11 Fernando Perez <fperez@colorado.edu> |
|
2147 | 2151 | |
|
2148 | 2152 | * IPython/numutils.py (amap): Added amap() fn. Simple shorthand |
|
2149 | 2153 | for Numeric.array(map(...)), but often convenient. |
|
2150 | 2154 | |
|
2151 | 2155 | 2003-11-05 Fernando Perez <fperez@colorado.edu> |
|
2152 | 2156 | |
|
2153 | 2157 | * IPython/numutils.py (frange): Changed a call from int() to |
|
2154 | 2158 | int(round()) to prevent a problem reported with arange() in the |
|
2155 | 2159 | numpy list. |
|
2156 | 2160 | |
|
2157 | 2161 | 2003-10-06 Fernando Perez <fperez@colorado.edu> |
|
2158 | 2162 | |
|
2159 | 2163 | * IPython/DPyGetOpt.py (DPyGetOpt.processArguments): changed to |
|
2160 | 2164 | prevent crashes if sys lacks an argv attribute (it happens with |
|
2161 | 2165 | embedded interpreters which build a bare-bones sys module). |
|
2162 | 2166 | Thanks to a report/bugfix by Adam Hupp <hupp-AT-cs.wisc.edu>. |
|
2163 | 2167 | |
|
2164 | 2168 | 2003-09-24 Fernando Perez <fperez@colorado.edu> |
|
2165 | 2169 | |
|
2166 | 2170 | * IPython/Magic.py (Magic._ofind): blanket except around getattr() |
|
2167 | 2171 | to protect against poorly written user objects where __getattr__ |
|
2168 | 2172 | raises exceptions other than AttributeError. Thanks to a bug |
|
2169 | 2173 | report by Oliver Sander <osander-AT-gmx.de>. |
|
2170 | 2174 | |
|
2171 | 2175 | * IPython/FakeModule.py (FakeModule.__repr__): this method was |
|
2172 | 2176 | missing. Thanks to bug report by Ralf Schmitt <ralf-AT-brainbot.com>. |
|
2173 | 2177 | |
|
2174 | 2178 | 2003-09-09 Fernando Perez <fperez@colorado.edu> |
|
2175 | 2179 | |
|
2176 | 2180 | * IPython/iplib.py (InteractiveShell._prefilter): fix bug where |
|
2177 | 2181 | unpacking a list whith a callable as first element would |
|
2178 | 2182 | mistakenly trigger autocalling. Thanks to a bug report by Jeffery |
|
2179 | 2183 | Collins. |
|
2180 | 2184 | |
|
2181 | 2185 | 2003-08-25 *** Released version 0.5.0 |
|
2182 | 2186 | |
|
2183 | 2187 | 2003-08-22 Fernando Perez <fperez@colorado.edu> |
|
2184 | 2188 | |
|
2185 | 2189 | * IPython/ultraTB.py (VerboseTB.linereader): Improved handling of |
|
2186 | 2190 | improperly defined user exceptions. Thanks to feedback from Mark |
|
2187 | 2191 | Russell <mrussell-AT-verio.net>. |
|
2188 | 2192 | |
|
2189 | 2193 | 2003-08-20 Fernando Perez <fperez@colorado.edu> |
|
2190 | 2194 | |
|
2191 | 2195 | * IPython/OInspect.py (Inspector.pinfo): changed String Form |
|
2192 | 2196 | printing so that it would print multi-line string forms starting |
|
2193 | 2197 | with a new line. This way the formatting is better respected for |
|
2194 | 2198 | objects which work hard to make nice string forms. |
|
2195 | 2199 | |
|
2196 | 2200 | * IPython/iplib.py (InteractiveShell.handle_auto): Fix bug where |
|
2197 | 2201 | autocall would overtake data access for objects with both |
|
2198 | 2202 | __getitem__ and __call__. |
|
2199 | 2203 | |
|
2200 | 2204 | 2003-08-19 *** Released version 0.5.0-rc1 |
|
2201 | 2205 | |
|
2202 | 2206 | 2003-08-19 Fernando Perez <fperez@colorado.edu> |
|
2203 | 2207 | |
|
2204 | 2208 | * IPython/deep_reload.py (load_tail): single tiny change here |
|
2205 | 2209 | seems to fix the long-standing bug of dreload() failing to work |
|
2206 | 2210 | for dotted names. But this module is pretty tricky, so I may have |
|
2207 | 2211 | missed some subtlety. Needs more testing!. |
|
2208 | 2212 | |
|
2209 | 2213 | * IPython/ultraTB.py (VerboseTB.linereader): harden against user |
|
2210 | 2214 | exceptions which have badly implemented __str__ methods. |
|
2211 | 2215 | (VerboseTB.text): harden against inspect.getinnerframes crashing, |
|
2212 | 2216 | which I've been getting reports about from Python 2.3 users. I |
|
2213 | 2217 | wish I had a simple test case to reproduce the problem, so I could |
|
2214 | 2218 | either write a cleaner workaround or file a bug report if |
|
2215 | 2219 | necessary. |
|
2216 | 2220 | |
|
2217 | 2221 | * IPython/Magic.py (Magic.magic_edit): fixed bug where after |
|
2218 | 2222 | making a class 'foo', file 'foo.py' couldn't be edited. Thanks to |
|
2219 | 2223 | a bug report by Tjabo Kloppenburg. |
|
2220 | 2224 | |
|
2221 | 2225 | * IPython/ultraTB.py (VerboseTB.debugger): hardened against pdb |
|
2222 | 2226 | crashes. Wrapped the pdb call in a blanket try/except, since pdb |
|
2223 | 2227 | seems rather unstable. Thanks to a bug report by Tjabo |
|
2224 | 2228 | Kloppenburg <tjabo.kloppenburg-AT-unix-ag.uni-siegen.de>. |
|
2225 | 2229 | |
|
2226 | 2230 | * IPython/Release.py (version): release 0.5.0-rc1. I want to put |
|
2227 | 2231 | this out soon because of the critical fixes in the inner loop for |
|
2228 | 2232 | generators. |
|
2229 | 2233 | |
|
2230 | 2234 | * IPython/Magic.py (Magic.getargspec): removed. This (and |
|
2231 | 2235 | _get_def) have been obsoleted by OInspect for a long time, I |
|
2232 | 2236 | hadn't noticed that they were dead code. |
|
2233 | 2237 | (Magic._ofind): restored _ofind functionality for a few literals |
|
2234 | 2238 | (those in ["''",'""','[]','{}','()']). But it won't work anymore |
|
2235 | 2239 | for things like "hello".capitalize?, since that would require a |
|
2236 | 2240 | potentially dangerous eval() again. |
|
2237 | 2241 | |
|
2238 | 2242 | * IPython/iplib.py (InteractiveShell._prefilter): reorganized the |
|
2239 | 2243 | logic a bit more to clean up the escapes handling and minimize the |
|
2240 | 2244 | use of _ofind to only necessary cases. The interactive 'feel' of |
|
2241 | 2245 | IPython should have improved quite a bit with the changes in |
|
2242 | 2246 | _prefilter and _ofind (besides being far safer than before). |
|
2243 | 2247 | |
|
2244 | 2248 | * IPython/Magic.py (Magic.magic_edit): Fixed old bug (but rather |
|
2245 | 2249 | obscure, never reported). Edit would fail to find the object to |
|
2246 | 2250 | edit under some circumstances. |
|
2247 | 2251 | (Magic._ofind): CRITICAL FIX. Finally removed the eval() calls |
|
2248 | 2252 | which were causing double-calling of generators. Those eval calls |
|
2249 | 2253 | were _very_ dangerous, since code with side effects could be |
|
2250 | 2254 | triggered. As they say, 'eval is evil'... These were the |
|
2251 | 2255 | nastiest evals in IPython. Besides, _ofind is now far simpler, |
|
2252 | 2256 | and it should also be quite a bit faster. Its use of inspect is |
|
2253 | 2257 | also safer, so perhaps some of the inspect-related crashes I've |
|
2254 | 2258 | seen lately with Python 2.3 might be taken care of. That will |
|
2255 | 2259 | need more testing. |
|
2256 | 2260 | |
|
2257 | 2261 | 2003-08-17 Fernando Perez <fperez@colorado.edu> |
|
2258 | 2262 | |
|
2259 | 2263 | * IPython/iplib.py (InteractiveShell._prefilter): significant |
|
2260 | 2264 | simplifications to the logic for handling user escapes. Faster |
|
2261 | 2265 | and simpler code. |
|
2262 | 2266 | |
|
2263 | 2267 | 2003-08-14 Fernando Perez <fperez@colorado.edu> |
|
2264 | 2268 | |
|
2265 | 2269 | * IPython/numutils.py (sum_flat): rewrote to be non-recursive. |
|
2266 | 2270 | Now it requires O(N) storage (N=size(a)) for non-contiguous input, |
|
2267 | 2271 | but it should be quite a bit faster. And the recursive version |
|
2268 | 2272 | generated O(log N) intermediate storage for all rank>1 arrays, |
|
2269 | 2273 | even if they were contiguous. |
|
2270 | 2274 | (l1norm): Added this function. |
|
2271 | 2275 | (norm): Added this function for arbitrary norms (including |
|
2272 | 2276 | l-infinity). l1 and l2 are still special cases for convenience |
|
2273 | 2277 | and speed. |
|
2274 | 2278 | |
|
2275 | 2279 | 2003-08-03 Fernando Perez <fperez@colorado.edu> |
|
2276 | 2280 | |
|
2277 | 2281 | * IPython/Magic.py (Magic.magic_edit): Removed all remaining string |
|
2278 | 2282 | exceptions, which now raise PendingDeprecationWarnings in Python |
|
2279 | 2283 | 2.3. There were some in Magic and some in Gnuplot2. |
|
2280 | 2284 | |
|
2281 | 2285 | 2003-06-30 Fernando Perez <fperez@colorado.edu> |
|
2282 | 2286 | |
|
2283 | 2287 | * IPython/genutils.py (page): modified to call curses only for |
|
2284 | 2288 | terminals where TERM=='xterm'. After problems under many other |
|
2285 | 2289 | terminals were reported by Keith Beattie <KSBeattie-AT-lbl.gov>. |
|
2286 | 2290 | |
|
2287 | 2291 | * IPython/iplib.py (complete): removed spurious 'print "IE"' which |
|
2288 | 2292 | would be triggered when readline was absent. This was just an old |
|
2289 | 2293 | debugging statement I'd forgotten to take out. |
|
2290 | 2294 | |
|
2291 | 2295 | 2003-06-20 Fernando Perez <fperez@colorado.edu> |
|
2292 | 2296 | |
|
2293 | 2297 | * IPython/genutils.py (clock): modified to return only user time |
|
2294 | 2298 | (not counting system time), after a discussion on scipy. While |
|
2295 | 2299 | system time may be a useful quantity occasionally, it may much |
|
2296 | 2300 | more easily be skewed by occasional swapping or other similar |
|
2297 | 2301 | activity. |
|
2298 | 2302 | |
|
2299 | 2303 | 2003-06-05 Fernando Perez <fperez@colorado.edu> |
|
2300 | 2304 | |
|
2301 | 2305 | * IPython/numutils.py (identity): new function, for building |
|
2302 | 2306 | arbitrary rank Kronecker deltas (mostly backwards compatible with |
|
2303 | 2307 | Numeric.identity) |
|
2304 | 2308 | |
|
2305 | 2309 | 2003-06-03 Fernando Perez <fperez@colorado.edu> |
|
2306 | 2310 | |
|
2307 | 2311 | * IPython/iplib.py (InteractiveShell.handle_magic): protect |
|
2308 | 2312 | arguments passed to magics with spaces, to allow trailing '\' to |
|
2309 | 2313 | work normally (mainly for Windows users). |
|
2310 | 2314 | |
|
2311 | 2315 | 2003-05-29 Fernando Perez <fperez@colorado.edu> |
|
2312 | 2316 | |
|
2313 | 2317 | * IPython/ipmaker.py (make_IPython): Load site._Helper() as help |
|
2314 | 2318 | instead of pydoc.help. This fixes a bizarre behavior where |
|
2315 | 2319 | printing '%s' % locals() would trigger the help system. Now |
|
2316 | 2320 | ipython behaves like normal python does. |
|
2317 | 2321 | |
|
2318 | 2322 | Note that if one does 'from pydoc import help', the bizarre |
|
2319 | 2323 | behavior returns, but this will also happen in normal python, so |
|
2320 | 2324 | it's not an ipython bug anymore (it has to do with how pydoc.help |
|
2321 | 2325 | is implemented). |
|
2322 | 2326 | |
|
2323 | 2327 | 2003-05-22 Fernando Perez <fperez@colorado.edu> |
|
2324 | 2328 | |
|
2325 | 2329 | * IPython/FlexCompleter.py (Completer.attr_matches): fixed to |
|
2326 | 2330 | return [] instead of None when nothing matches, also match to end |
|
2327 | 2331 | of line. Patch by Gary Bishop. |
|
2328 | 2332 | |
|
2329 | 2333 | * IPython/ipmaker.py (make_IPython): Added same sys.excepthook |
|
2330 | 2334 | protection as before, for files passed on the command line. This |
|
2331 | 2335 | prevents the CrashHandler from kicking in if user files call into |
|
2332 | 2336 | sys.excepthook (such as PyQt and WxWindows have a nasty habit of |
|
2333 | 2337 | doing). After a report by Kasper Souren <Kasper.Souren-AT-ircam.fr> |
|
2334 | 2338 | |
|
2335 | 2339 | 2003-05-20 *** Released version 0.4.0 |
|
2336 | 2340 | |
|
2337 | 2341 | 2003-05-20 Fernando Perez <fperez@colorado.edu> |
|
2338 | 2342 | |
|
2339 | 2343 | * setup.py: added support for manpages. It's a bit hackish b/c of |
|
2340 | 2344 | a bug in the way the bdist_rpm distutils target handles gzipped |
|
2341 | 2345 | manpages, but it works. After a patch by Jack. |
|
2342 | 2346 | |
|
2343 | 2347 | 2003-05-19 Fernando Perez <fperez@colorado.edu> |
|
2344 | 2348 | |
|
2345 | 2349 | * IPython/numutils.py: added a mockup of the kinds module, since |
|
2346 | 2350 | it was recently removed from Numeric. This way, numutils will |
|
2347 | 2351 | work for all users even if they are missing kinds. |
|
2348 | 2352 | |
|
2349 | 2353 | * IPython/Magic.py (Magic._ofind): Harden against an inspect |
|
2350 | 2354 | failure, which can occur with SWIG-wrapped extensions. After a |
|
2351 | 2355 | crash report from Prabhu. |
|
2352 | 2356 | |
|
2353 | 2357 | 2003-05-16 Fernando Perez <fperez@colorado.edu> |
|
2354 | 2358 | |
|
2355 | 2359 | * IPython/iplib.py (InteractiveShell.excepthook): New method to |
|
2356 | 2360 | protect ipython from user code which may call directly |
|
2357 | 2361 | sys.excepthook (this looks like an ipython crash to the user, even |
|
2358 | 2362 | when it isn't). After a patch by Gary Bishop <gb-AT-cs.unc.edu>. |
|
2359 | 2363 | This is especially important to help users of WxWindows, but may |
|
2360 | 2364 | also be useful in other cases. |
|
2361 | 2365 | |
|
2362 | 2366 | * IPython/ultraTB.py (AutoFormattedTB.__call__): Changed to allow |
|
2363 | 2367 | an optional tb_offset to be specified, and to preserve exception |
|
2364 | 2368 | info if given. After a patch by Gary Bishop <gb-AT-cs.unc.edu>. |
|
2365 | 2369 | |
|
2366 | 2370 | * ipython.1 (Default): Thanks to Jack's work, we now have manpages! |
|
2367 | 2371 | |
|
2368 | 2372 | 2003-05-15 Fernando Perez <fperez@colorado.edu> |
|
2369 | 2373 | |
|
2370 | 2374 | * IPython/iplib.py (InteractiveShell.user_setup): Fix crash when |
|
2371 | 2375 | installing for a new user under Windows. |
|
2372 | 2376 | |
|
2373 | 2377 | 2003-05-12 Fernando Perez <fperez@colorado.edu> |
|
2374 | 2378 | |
|
2375 | 2379 | * IPython/iplib.py (InteractiveShell.handle_emacs): New line |
|
2376 | 2380 | handler for Emacs comint-based lines. Currently it doesn't do |
|
2377 | 2381 | much (but importantly, it doesn't update the history cache). In |
|
2378 | 2382 | the future it may be expanded if Alex needs more functionality |
|
2379 | 2383 | there. |
|
2380 | 2384 | |
|
2381 | 2385 | * IPython/CrashHandler.py (CrashHandler.__call__): Added platform |
|
2382 | 2386 | info to crash reports. |
|
2383 | 2387 | |
|
2384 | 2388 | * IPython/iplib.py (InteractiveShell.mainloop): Added -c option, |
|
2385 | 2389 | just like Python's -c. Also fixed crash with invalid -color |
|
2386 | 2390 | option value at startup. Thanks to Will French |
|
2387 | 2391 | <wfrench-AT-bestweb.net> for the bug report. |
|
2388 | 2392 | |
|
2389 | 2393 | 2003-05-09 Fernando Perez <fperez@colorado.edu> |
|
2390 | 2394 | |
|
2391 | 2395 | * IPython/genutils.py (EvalDict.__getitem__): Renamed EvalString |
|
2392 | 2396 | to EvalDict (it's a mapping, after all) and simplified its code |
|
2393 | 2397 | quite a bit, after a nice discussion on c.l.py where Gustavo |
|
2394 | 2398 | CΓ³rdova <gcordova-AT-sismex.com> suggested the new version. |
|
2395 | 2399 | |
|
2396 | 2400 | 2003-04-30 Fernando Perez <fperez@colorado.edu> |
|
2397 | 2401 | |
|
2398 | 2402 | * IPython/genutils.py (timings_out): modified it to reduce its |
|
2399 | 2403 | overhead in the common reps==1 case. |
|
2400 | 2404 | |
|
2401 | 2405 | 2003-04-29 Fernando Perez <fperez@colorado.edu> |
|
2402 | 2406 | |
|
2403 | 2407 | * IPython/genutils.py (timings_out): Modified to use the resource |
|
2404 | 2408 | module, which avoids the wraparound problems of time.clock(). |
|
2405 | 2409 | |
|
2406 | 2410 | 2003-04-17 *** Released version 0.2.15pre4 |
|
2407 | 2411 | |
|
2408 | 2412 | 2003-04-17 Fernando Perez <fperez@colorado.edu> |
|
2409 | 2413 | |
|
2410 | 2414 | * setup.py (scriptfiles): Split windows-specific stuff over to a |
|
2411 | 2415 | separate file, in an attempt to have a Windows GUI installer. |
|
2412 | 2416 | That didn't work, but part of the groundwork is done. |
|
2413 | 2417 | |
|
2414 | 2418 | * IPython/UserConfig/ipythonrc: Added M-i, M-o and M-I for |
|
2415 | 2419 | indent/unindent with 4 spaces. Particularly useful in combination |
|
2416 | 2420 | with the new auto-indent option. |
|
2417 | 2421 | |
|
2418 | 2422 | 2003-04-16 Fernando Perez <fperez@colorado.edu> |
|
2419 | 2423 | |
|
2420 | 2424 | * IPython/Magic.py: various replacements of self.rc for |
|
2421 | 2425 | self.shell.rc. A lot more remains to be done to fully disentangle |
|
2422 | 2426 | this class from the main Shell class. |
|
2423 | 2427 | |
|
2424 | 2428 | * IPython/GnuplotRuntime.py: added checks for mouse support so |
|
2425 | 2429 | that we don't try to enable it if the current gnuplot doesn't |
|
2426 | 2430 | really support it. Also added checks so that we don't try to |
|
2427 | 2431 | enable persist under Windows (where Gnuplot doesn't recognize the |
|
2428 | 2432 | option). |
|
2429 | 2433 | |
|
2430 | 2434 | * IPython/iplib.py (InteractiveShell.interact): Added optional |
|
2431 | 2435 | auto-indenting code, after a patch by King C. Shu |
|
2432 | 2436 | <kingshu-AT-myrealbox.com>. It's off by default because it doesn't |
|
2433 | 2437 | get along well with pasting indented code. If I ever figure out |
|
2434 | 2438 | how to make that part go well, it will become on by default. |
|
2435 | 2439 | |
|
2436 | 2440 | * IPython/Prompts.py (Prompt1.auto_rewrite): Fixed bug which would |
|
2437 | 2441 | crash ipython if there was an unmatched '%' in the user's prompt |
|
2438 | 2442 | string. Reported by Thorsten Kampe <thorsten-AT-thorstenkampe.de>. |
|
2439 | 2443 | |
|
2440 | 2444 | * IPython/iplib.py (InteractiveShell.interact): removed the |
|
2441 | 2445 | ability to ask the user whether he wants to crash or not at the |
|
2442 | 2446 | 'last line' exception handler. Calling functions at that point |
|
2443 | 2447 | changes the stack, and the error reports would have incorrect |
|
2444 | 2448 | tracebacks. |
|
2445 | 2449 | |
|
2446 | 2450 | * IPython/Magic.py (Magic.magic_page): Added new @page magic, to |
|
2447 | 2451 | pass through a peger a pretty-printed form of any object. After a |
|
2448 | 2452 | contribution by Olivier Aubert <oaubert-AT-bat710.univ-lyon1.fr> |
|
2449 | 2453 | |
|
2450 | 2454 | 2003-04-14 Fernando Perez <fperez@colorado.edu> |
|
2451 | 2455 | |
|
2452 | 2456 | * IPython/iplib.py (InteractiveShell.user_setup): Fixed bug where |
|
2453 | 2457 | all files in ~ would be modified at first install (instead of |
|
2454 | 2458 | ~/.ipython). This could be potentially disastrous, as the |
|
2455 | 2459 | modification (make line-endings native) could damage binary files. |
|
2456 | 2460 | |
|
2457 | 2461 | 2003-04-10 Fernando Perez <fperez@colorado.edu> |
|
2458 | 2462 | |
|
2459 | 2463 | * IPython/iplib.py (InteractiveShell.handle_help): Modified to |
|
2460 | 2464 | handle only lines which are invalid python. This now means that |
|
2461 | 2465 | lines like 'x=1 #?' execute properly. Thanks to Jeffery Collins |
|
2462 | 2466 | for the bug report. |
|
2463 | 2467 | |
|
2464 | 2468 | 2003-04-01 Fernando Perez <fperez@colorado.edu> |
|
2465 | 2469 | |
|
2466 | 2470 | * IPython/iplib.py (InteractiveShell.showtraceback): Fixed bug |
|
2467 | 2471 | where failing to set sys.last_traceback would crash pdb.pm(). |
|
2468 | 2472 | Thanks to Jeffery D. Collins <Jeff.Collins-AT-vexcel.com> for the bug |
|
2469 | 2473 | report. |
|
2470 | 2474 | |
|
2471 | 2475 | 2003-03-25 Fernando Perez <fperez@colorado.edu> |
|
2472 | 2476 | |
|
2473 | 2477 | * IPython/Magic.py (Magic.magic_prun): rstrip() output of profiler |
|
2474 | 2478 | before printing it (it had a lot of spurious blank lines at the |
|
2475 | 2479 | end). |
|
2476 | 2480 | |
|
2477 | 2481 | * IPython/Gnuplot2.py (Gnuplot.hardcopy): fixed bug where lpr |
|
2478 | 2482 | output would be sent 21 times! Obviously people don't use this |
|
2479 | 2483 | too often, or I would have heard about it. |
|
2480 | 2484 | |
|
2481 | 2485 | 2003-03-24 Fernando Perez <fperez@colorado.edu> |
|
2482 | 2486 | |
|
2483 | 2487 | * setup.py (scriptfiles): renamed the data_files parameter from |
|
2484 | 2488 | 'base' to 'data' to fix rpm build issues. Thanks to Ralf Ahlbrink |
|
2485 | 2489 | for the patch. |
|
2486 | 2490 | |
|
2487 | 2491 | 2003-03-20 Fernando Perez <fperez@colorado.edu> |
|
2488 | 2492 | |
|
2489 | 2493 | * IPython/genutils.py (error): added error() and fatal() |
|
2490 | 2494 | functions. |
|
2491 | 2495 | |
|
2492 | 2496 | 2003-03-18 *** Released version 0.2.15pre3 |
|
2493 | 2497 | |
|
2494 | 2498 | 2003-03-18 Fernando Perez <fperez@colorado.edu> |
|
2495 | 2499 | |
|
2496 | 2500 | * setupext/install_data_ext.py |
|
2497 | 2501 | (install_data_ext.initialize_options): Class contributed by Jack |
|
2498 | 2502 | Moffit for fixing the old distutils hack. He is sending this to |
|
2499 | 2503 | the distutils folks so in the future we may not need it as a |
|
2500 | 2504 | private fix. |
|
2501 | 2505 | |
|
2502 | 2506 | * MANIFEST.in: Extensive reorganization, based on Jack Moffit's |
|
2503 | 2507 | changes for Debian packaging. See his patch for full details. |
|
2504 | 2508 | The old distutils hack of making the ipythonrc* files carry a |
|
2505 | 2509 | bogus .py extension is gone, at last. Examples were moved to a |
|
2506 | 2510 | separate subdir under doc/, and the separate executable scripts |
|
2507 | 2511 | now live in their own directory. Overall a great cleanup. The |
|
2508 | 2512 | manual was updated to use the new files, and setup.py has been |
|
2509 | 2513 | fixed for this setup. |
|
2510 | 2514 | |
|
2511 | 2515 | * IPython/PyColorize.py (Parser.usage): made non-executable and |
|
2512 | 2516 | created a pycolor wrapper around it to be included as a script. |
|
2513 | 2517 | |
|
2514 | 2518 | 2003-03-12 *** Released version 0.2.15pre2 |
|
2515 | 2519 | |
|
2516 | 2520 | 2003-03-12 Fernando Perez <fperez@colorado.edu> |
|
2517 | 2521 | |
|
2518 | 2522 | * IPython/ColorANSI.py (make_color_table): Finally fixed the |
|
2519 | 2523 | long-standing problem with garbage characters in some terminals. |
|
2520 | 2524 | The issue was really that the \001 and \002 escapes must _only_ be |
|
2521 | 2525 | passed to input prompts (which call readline), but _never_ to |
|
2522 | 2526 | normal text to be printed on screen. I changed ColorANSI to have |
|
2523 | 2527 | two classes: TermColors and InputTermColors, each with the |
|
2524 | 2528 | appropriate escapes for input prompts or normal text. The code in |
|
2525 | 2529 | Prompts.py got slightly more complicated, but this very old and |
|
2526 | 2530 | annoying bug is finally fixed. |
|
2527 | 2531 | |
|
2528 | 2532 | All the credit for nailing down the real origin of this problem |
|
2529 | 2533 | and the correct solution goes to Jack Moffit <jack-AT-xiph.org>. |
|
2530 | 2534 | *Many* thanks to him for spending quite a bit of effort on this. |
|
2531 | 2535 | |
|
2532 | 2536 | 2003-03-05 *** Released version 0.2.15pre1 |
|
2533 | 2537 | |
|
2534 | 2538 | 2003-03-03 Fernando Perez <fperez@colorado.edu> |
|
2535 | 2539 | |
|
2536 | 2540 | * IPython/FakeModule.py: Moved the former _FakeModule to a |
|
2537 | 2541 | separate file, because it's also needed by Magic (to fix a similar |
|
2538 | 2542 | pickle-related issue in @run). |
|
2539 | 2543 | |
|
2540 | 2544 | 2003-03-02 Fernando Perez <fperez@colorado.edu> |
|
2541 | 2545 | |
|
2542 | 2546 | * IPython/Magic.py (Magic.magic_autocall): new magic to control |
|
2543 | 2547 | the autocall option at runtime. |
|
2544 | 2548 | (Magic.magic_dhist): changed self.user_ns to self.shell.user_ns |
|
2545 | 2549 | across Magic.py to start separating Magic from InteractiveShell. |
|
2546 | 2550 | (Magic._ofind): Fixed to return proper namespace for dotted |
|
2547 | 2551 | names. Before, a dotted name would always return 'not currently |
|
2548 | 2552 | defined', because it would find the 'parent'. s.x would be found, |
|
2549 | 2553 | but since 'x' isn't defined by itself, it would get confused. |
|
2550 | 2554 | (Magic.magic_run): Fixed pickling problems reported by Ralf |
|
2551 | 2555 | Ahlbrink <RAhlbrink-AT-RosenInspection.net>. The fix was similar to |
|
2552 | 2556 | that I'd used when Mike Heeter reported similar issues at the |
|
2553 | 2557 | top-level, but now for @run. It boils down to injecting the |
|
2554 | 2558 | namespace where code is being executed with something that looks |
|
2555 | 2559 | enough like a module to fool pickle.dump(). Since a pickle stores |
|
2556 | 2560 | a named reference to the importing module, we need this for |
|
2557 | 2561 | pickles to save something sensible. |
|
2558 | 2562 | |
|
2559 | 2563 | * IPython/ipmaker.py (make_IPython): added an autocall option. |
|
2560 | 2564 | |
|
2561 | 2565 | * IPython/iplib.py (InteractiveShell._prefilter): reordered all of |
|
2562 | 2566 | the auto-eval code. Now autocalling is an option, and the code is |
|
2563 | 2567 | also vastly safer. There is no more eval() involved at all. |
|
2564 | 2568 | |
|
2565 | 2569 | 2003-03-01 Fernando Perez <fperez@colorado.edu> |
|
2566 | 2570 | |
|
2567 | 2571 | * IPython/Magic.py (Magic._ofind): Changed interface to return a |
|
2568 | 2572 | dict with named keys instead of a tuple. |
|
2569 | 2573 | |
|
2570 | 2574 | * IPython: Started using CVS for IPython as of 0.2.15pre1. |
|
2571 | 2575 | |
|
2572 | 2576 | * setup.py (make_shortcut): Fixed message about directories |
|
2573 | 2577 | created during Windows installation (the directories were ok, just |
|
2574 | 2578 | the printed message was misleading). Thanks to Chris Liechti |
|
2575 | 2579 | <cliechti-AT-gmx.net> for the heads up. |
|
2576 | 2580 | |
|
2577 | 2581 | 2003-02-21 Fernando Perez <fperez@colorado.edu> |
|
2578 | 2582 | |
|
2579 | 2583 | * IPython/iplib.py (InteractiveShell._prefilter): Fixed catching |
|
2580 | 2584 | of ValueError exception when checking for auto-execution. This |
|
2581 | 2585 | one is raised by things like Numeric arrays arr.flat when the |
|
2582 | 2586 | array is non-contiguous. |
|
2583 | 2587 | |
|
2584 | 2588 | 2003-01-31 Fernando Perez <fperez@colorado.edu> |
|
2585 | 2589 | |
|
2586 | 2590 | * IPython/genutils.py (SystemExec.bq): Fixed bug where bq would |
|
2587 | 2591 | not return any value at all (even though the command would get |
|
2588 | 2592 | executed). |
|
2589 | 2593 | (xsys): Flush stdout right after printing the command to ensure |
|
2590 | 2594 | proper ordering of commands and command output in the total |
|
2591 | 2595 | output. |
|
2592 | 2596 | (SystemExec/xsys/bq): Switched the names of xsys/bq and |
|
2593 | 2597 | system/getoutput as defaults. The old ones are kept for |
|
2594 | 2598 | compatibility reasons, so no code which uses this library needs |
|
2595 | 2599 | changing. |
|
2596 | 2600 | |
|
2597 | 2601 | 2003-01-27 *** Released version 0.2.14 |
|
2598 | 2602 | |
|
2599 | 2603 | 2003-01-25 Fernando Perez <fperez@colorado.edu> |
|
2600 | 2604 | |
|
2601 | 2605 | * IPython/Magic.py (Magic.magic_edit): Fixed problem where |
|
2602 | 2606 | functions defined in previous edit sessions could not be re-edited |
|
2603 | 2607 | (because the temp files were immediately removed). Now temp files |
|
2604 | 2608 | are removed only at IPython's exit. |
|
2605 | 2609 | (Magic.magic_run): Improved @run to perform shell-like expansions |
|
2606 | 2610 | on its arguments (~users and $VARS). With this, @run becomes more |
|
2607 | 2611 | like a normal command-line. |
|
2608 | 2612 | |
|
2609 | 2613 | * IPython/Shell.py (IPShellEmbed.__call__): Fixed a bunch of small |
|
2610 | 2614 | bugs related to embedding and cleaned up that code. A fairly |
|
2611 | 2615 | important one was the impossibility to access the global namespace |
|
2612 | 2616 | through the embedded IPython (only local variables were visible). |
|
2613 | 2617 | |
|
2614 | 2618 | 2003-01-14 Fernando Perez <fperez@colorado.edu> |
|
2615 | 2619 | |
|
2616 | 2620 | * IPython/iplib.py (InteractiveShell._prefilter): Fixed |
|
2617 | 2621 | auto-calling to be a bit more conservative. Now it doesn't get |
|
2618 | 2622 | triggered if any of '!=()<>' are in the rest of the input line, to |
|
2619 | 2623 | allow comparing callables. Thanks to Alex for the heads up. |
|
2620 | 2624 | |
|
2621 | 2625 | 2003-01-07 Fernando Perez <fperez@colorado.edu> |
|
2622 | 2626 | |
|
2623 | 2627 | * IPython/genutils.py (page): fixed estimation of the number of |
|
2624 | 2628 | lines in a string to be paged to simply count newlines. This |
|
2625 | 2629 | prevents over-guessing due to embedded escape sequences. A better |
|
2626 | 2630 | long-term solution would involve stripping out the control chars |
|
2627 | 2631 | for the count, but it's potentially so expensive I just don't |
|
2628 | 2632 | think it's worth doing. |
|
2629 | 2633 | |
|
2630 | 2634 | 2002-12-19 *** Released version 0.2.14pre50 |
|
2631 | 2635 | |
|
2632 | 2636 | 2002-12-19 Fernando Perez <fperez@colorado.edu> |
|
2633 | 2637 | |
|
2634 | 2638 | * tools/release (version): Changed release scripts to inform |
|
2635 | 2639 | Andrea and build a NEWS file with a list of recent changes. |
|
2636 | 2640 | |
|
2637 | 2641 | * IPython/ColorANSI.py (__all__): changed terminal detection |
|
2638 | 2642 | code. Seems to work better for xterms without breaking |
|
2639 | 2643 | konsole. Will need more testing to determine if WinXP and Mac OSX |
|
2640 | 2644 | also work ok. |
|
2641 | 2645 | |
|
2642 | 2646 | 2002-12-18 *** Released version 0.2.14pre49 |
|
2643 | 2647 | |
|
2644 | 2648 | 2002-12-18 Fernando Perez <fperez@colorado.edu> |
|
2645 | 2649 | |
|
2646 | 2650 | * Docs: added new info about Mac OSX, from Andrea. |
|
2647 | 2651 | |
|
2648 | 2652 | * IPython/Gnuplot2.py (String): Added a String PlotItem class to |
|
2649 | 2653 | allow direct plotting of python strings whose format is the same |
|
2650 | 2654 | of gnuplot data files. |
|
2651 | 2655 | |
|
2652 | 2656 | 2002-12-16 Fernando Perez <fperez@colorado.edu> |
|
2653 | 2657 | |
|
2654 | 2658 | * IPython/iplib.py (InteractiveShell.interact): fixed default (y) |
|
2655 | 2659 | value of exit question to be acknowledged. |
|
2656 | 2660 | |
|
2657 | 2661 | 2002-12-03 Fernando Perez <fperez@colorado.edu> |
|
2658 | 2662 | |
|
2659 | 2663 | * IPython/ipmaker.py: removed generators, which had been added |
|
2660 | 2664 | by mistake in an earlier debugging run. This was causing trouble |
|
2661 | 2665 | to users of python 2.1.x. Thanks to Abel Daniel <abli-AT-freemail.hu> |
|
2662 | 2666 | for pointing this out. |
|
2663 | 2667 | |
|
2664 | 2668 | 2002-11-17 Fernando Perez <fperez@colorado.edu> |
|
2665 | 2669 | |
|
2666 | 2670 | * Manual: updated the Gnuplot section. |
|
2667 | 2671 | |
|
2668 | 2672 | * IPython/GnuplotRuntime.py: refactored a lot all this code, with |
|
2669 | 2673 | a much better split of what goes in Runtime and what goes in |
|
2670 | 2674 | Interactive. |
|
2671 | 2675 | |
|
2672 | 2676 | * IPython/ipmaker.py: fixed bug where import_fail_info wasn't |
|
2673 | 2677 | being imported from iplib. |
|
2674 | 2678 | |
|
2675 | 2679 | * IPython/GnuplotInteractive.py (magic_gpc): renamed @gp to @gpc |
|
2676 | 2680 | for command-passing. Now the global Gnuplot instance is called |
|
2677 | 2681 | 'gp' instead of 'g', which was really a far too fragile and |
|
2678 | 2682 | common name. |
|
2679 | 2683 | |
|
2680 | 2684 | * IPython/Gnuplot2.py (eps_fix_bbox): added this to fix broken |
|
2681 | 2685 | bounding boxes generated by Gnuplot for square plots. |
|
2682 | 2686 | |
|
2683 | 2687 | * IPython/genutils.py (popkey): new function added. I should |
|
2684 | 2688 | suggest this on c.l.py as a dict method, it seems useful. |
|
2685 | 2689 | |
|
2686 | 2690 | * IPython/Gnuplot2.py (Gnuplot.plot): Overhauled plot and replot |
|
2687 | 2691 | to transparently handle PostScript generation. MUCH better than |
|
2688 | 2692 | the previous plot_eps/replot_eps (which I removed now). The code |
|
2689 | 2693 | is also fairly clean and well documented now (including |
|
2690 | 2694 | docstrings). |
|
2691 | 2695 | |
|
2692 | 2696 | 2002-11-13 Fernando Perez <fperez@colorado.edu> |
|
2693 | 2697 | |
|
2694 | 2698 | * IPython/Magic.py (Magic.magic_edit): fixed docstring |
|
2695 | 2699 | (inconsistent with options). |
|
2696 | 2700 | |
|
2697 | 2701 | * IPython/Gnuplot2.py (Gnuplot.hardcopy): hardcopy had been |
|
2698 | 2702 | manually disabled, I don't know why. Fixed it. |
|
2699 | 2703 | (Gnuplot._plot_eps): added new plot_eps/replot_eps to get directly |
|
2700 | 2704 | eps output. |
|
2701 | 2705 | |
|
2702 | 2706 | 2002-11-12 Fernando Perez <fperez@colorado.edu> |
|
2703 | 2707 | |
|
2704 | 2708 | * IPython/genutils.py (ask_yes_no): trap EOF and ^C so that they |
|
2705 | 2709 | don't propagate up to caller. Fixes crash reported by François |
|
2706 | 2710 | Pinard. |
|
2707 | 2711 | |
|
2708 | 2712 | 2002-11-09 Fernando Perez <fperez@colorado.edu> |
|
2709 | 2713 | |
|
2710 | 2714 | * IPython/ipmaker.py (make_IPython): fixed problem with writing |
|
2711 | 2715 | history file for new users. |
|
2712 | 2716 | (make_IPython): fixed bug where initial install would leave the |
|
2713 | 2717 | user running in the .ipython dir. |
|
2714 | 2718 | (make_IPython): fixed bug where config dir .ipython would be |
|
2715 | 2719 | created regardless of the given -ipythondir option. Thanks to Cory |
|
2716 | 2720 | Dodt <cdodt-AT-fcoe.k12.ca.us> for the bug report. |
|
2717 | 2721 | |
|
2718 | 2722 | * IPython/genutils.py (ask_yes_no): new function for asking yes/no |
|
2719 | 2723 | type confirmations. Will need to use it in all of IPython's code |
|
2720 | 2724 | consistently. |
|
2721 | 2725 | |
|
2722 | 2726 | * IPython/CrashHandler.py (CrashHandler.__call__): changed the |
|
2723 | 2727 | context to print 31 lines instead of the default 5. This will make |
|
2724 | 2728 | the crash reports extremely detailed in case the problem is in |
|
2725 | 2729 | libraries I don't have access to. |
|
2726 | 2730 | |
|
2727 | 2731 | * IPython/iplib.py (InteractiveShell.interact): changed the 'last |
|
2728 | 2732 | line of defense' code to still crash, but giving users fair |
|
2729 | 2733 | warning. I don't want internal errors to go unreported: if there's |
|
2730 | 2734 | an internal problem, IPython should crash and generate a full |
|
2731 | 2735 | report. |
|
2732 | 2736 | |
|
2733 | 2737 | 2002-11-08 Fernando Perez <fperez@colorado.edu> |
|
2734 | 2738 | |
|
2735 | 2739 | * IPython/iplib.py (InteractiveShell.interact): added code to trap |
|
2736 | 2740 | otherwise uncaught exceptions which can appear if people set |
|
2737 | 2741 | sys.stdout to something badly broken. Thanks to a crash report |
|
2738 | 2742 | from henni-AT-mail.brainbot.com. |
|
2739 | 2743 | |
|
2740 | 2744 | 2002-11-04 Fernando Perez <fperez@colorado.edu> |
|
2741 | 2745 | |
|
2742 | 2746 | * IPython/iplib.py (InteractiveShell.interact): added |
|
2743 | 2747 | __IPYTHON__active to the builtins. It's a flag which goes on when |
|
2744 | 2748 | the interaction starts and goes off again when it stops. This |
|
2745 | 2749 | allows embedding code to detect being inside IPython. Before this |
|
2746 | 2750 | was done via __IPYTHON__, but that only shows that an IPython |
|
2747 | 2751 | instance has been created. |
|
2748 | 2752 | |
|
2749 | 2753 | * IPython/Magic.py (Magic.magic_env): I realized that in a |
|
2750 | 2754 | UserDict, instance.data holds the data as a normal dict. So I |
|
2751 | 2755 | modified @env to return os.environ.data instead of rebuilding a |
|
2752 | 2756 | dict by hand. |
|
2753 | 2757 | |
|
2754 | 2758 | 2002-11-02 Fernando Perez <fperez@colorado.edu> |
|
2755 | 2759 | |
|
2756 | 2760 | * IPython/genutils.py (warn): changed so that level 1 prints no |
|
2757 | 2761 | header. Level 2 is now the default (with 'WARNING' header, as |
|
2758 | 2762 | before). I think I tracked all places where changes were needed in |
|
2759 | 2763 | IPython, but outside code using the old level numbering may have |
|
2760 | 2764 | broken. |
|
2761 | 2765 | |
|
2762 | 2766 | * IPython/iplib.py (InteractiveShell.runcode): added this to |
|
2763 | 2767 | handle the tracebacks in SystemExit traps correctly. The previous |
|
2764 | 2768 | code (through interact) was printing more of the stack than |
|
2765 | 2769 | necessary, showing IPython internal code to the user. |
|
2766 | 2770 | |
|
2767 | 2771 | * IPython/UserConfig/ipythonrc.py: Made confirm_exit 1 by |
|
2768 | 2772 | default. Now that the default at the confirmation prompt is yes, |
|
2769 | 2773 | it's not so intrusive. François' argument that ipython sessions |
|
2770 | 2774 | tend to be complex enough not to lose them from an accidental C-d, |
|
2771 | 2775 | is a valid one. |
|
2772 | 2776 | |
|
2773 | 2777 | * IPython/iplib.py (InteractiveShell.interact): added a |
|
2774 | 2778 | showtraceback() call to the SystemExit trap, and modified the exit |
|
2775 | 2779 | confirmation to have yes as the default. |
|
2776 | 2780 | |
|
2777 | 2781 | * IPython/UserConfig/ipythonrc.py: removed 'session' option from |
|
2778 | 2782 | this file. It's been gone from the code for a long time, this was |
|
2779 | 2783 | simply leftover junk. |
|
2780 | 2784 | |
|
2781 | 2785 | 2002-11-01 Fernando Perez <fperez@colorado.edu> |
|
2782 | 2786 | |
|
2783 | 2787 | * IPython/UserConfig/ipythonrc.py: new confirm_exit option |
|
2784 | 2788 | added. If set, IPython now traps EOF and asks for |
|
2785 | 2789 | confirmation. After a request by François Pinard. |
|
2786 | 2790 | |
|
2787 | 2791 | * IPython/Magic.py (Magic.magic_Exit): New @Exit and @Quit instead |
|
2788 | 2792 | of @abort, and with a new (better) mechanism for handling the |
|
2789 | 2793 | exceptions. |
|
2790 | 2794 | |
|
2791 | 2795 | 2002-10-27 Fernando Perez <fperez@colorado.edu> |
|
2792 | 2796 | |
|
2793 | 2797 | * IPython/usage.py (__doc__): updated the --help information and |
|
2794 | 2798 | the ipythonrc file to indicate that -log generates |
|
2795 | 2799 | ./ipython.log. Also fixed the corresponding info in @logstart. |
|
2796 | 2800 | This and several other fixes in the manuals thanks to reports by |
|
2797 | 2801 | François Pinard <pinard-AT-iro.umontreal.ca>. |
|
2798 | 2802 | |
|
2799 | 2803 | * IPython/Logger.py (Logger.switch_log): Fixed error message to |
|
2800 | 2804 | refer to @logstart (instead of @log, which doesn't exist). |
|
2801 | 2805 | |
|
2802 | 2806 | * IPython/iplib.py (InteractiveShell._prefilter): fixed |
|
2803 | 2807 | AttributeError crash. Thanks to Christopher Armstrong |
|
2804 | 2808 | <radix-AT-twistedmatrix.com> for the report/fix. This bug had been |
|
2805 | 2809 | introduced recently (in 0.2.14pre37) with the fix to the eval |
|
2806 | 2810 | problem mentioned below. |
|
2807 | 2811 | |
|
2808 | 2812 | 2002-10-17 Fernando Perez <fperez@colorado.edu> |
|
2809 | 2813 | |
|
2810 | 2814 | * IPython/ConfigLoader.py (ConfigLoader.load): Fixes for Windows |
|
2811 | 2815 | installation. Thanks to Leonardo Santagada <retype-AT-terra.com.br>. |
|
2812 | 2816 | |
|
2813 | 2817 | * IPython/iplib.py (InteractiveShell._prefilter): Many changes to |
|
2814 | 2818 | this function to fix a problem reported by Alex Schmolck. He saw |
|
2815 | 2819 | it with list comprehensions and generators, which were getting |
|
2816 | 2820 | called twice. The real problem was an 'eval' call in testing for |
|
2817 | 2821 | automagic which was evaluating the input line silently. |
|
2818 | 2822 | |
|
2819 | 2823 | This is a potentially very nasty bug, if the input has side |
|
2820 | 2824 | effects which must not be repeated. The code is much cleaner now, |
|
2821 | 2825 | without any blanket 'except' left and with a regexp test for |
|
2822 | 2826 | actual function names. |
|
2823 | 2827 | |
|
2824 | 2828 | But an eval remains, which I'm not fully comfortable with. I just |
|
2825 | 2829 | don't know how to find out if an expression could be a callable in |
|
2826 | 2830 | the user's namespace without doing an eval on the string. However |
|
2827 | 2831 | that string is now much more strictly checked so that no code |
|
2828 | 2832 | slips by, so the eval should only happen for things that can |
|
2829 | 2833 | really be only function/method names. |
|
2830 | 2834 | |
|
2831 | 2835 | 2002-10-15 Fernando Perez <fperez@colorado.edu> |
|
2832 | 2836 | |
|
2833 | 2837 | * Updated LyX to 1.2.1 so I can work on the docs again. Added Mac |
|
2834 | 2838 | OSX information to main manual, removed README_Mac_OSX file from |
|
2835 | 2839 | distribution. Also updated credits for recent additions. |
|
2836 | 2840 | |
|
2837 | 2841 | 2002-10-10 Fernando Perez <fperez@colorado.edu> |
|
2838 | 2842 | |
|
2839 | 2843 | * README_Mac_OSX: Added a README for Mac OSX users for fixing |
|
2840 | 2844 | terminal-related issues. Many thanks to Andrea Riciputi |
|
2841 | 2845 | <andrea.riciputi-AT-libero.it> for writing it. |
|
2842 | 2846 | |
|
2843 | 2847 | * IPython/UserConfig/ipythonrc.py: Fixes to various small issues, |
|
2844 | 2848 | thanks to Thorsten Kampe <thorsten-AT-thorstenkampe.de>. |
|
2845 | 2849 | |
|
2846 | 2850 | * setup.py (make_shortcut): Fixes for Windows installation. Thanks |
|
2847 | 2851 | to Fredrik Kant <fredrik.kant-AT-front.com> and Syver Enstad |
|
2848 | 2852 | <syver-en-AT-online.no> who both submitted patches for this problem. |
|
2849 | 2853 | |
|
2850 | 2854 | * IPython/iplib.py (InteractiveShell.embed_mainloop): Patch for |
|
2851 | 2855 | global embedding to make sure that things don't overwrite user |
|
2852 | 2856 | globals accidentally. Thanks to Richard <rxe-AT-renre-europe.com> |
|
2853 | 2857 | |
|
2854 | 2858 | * IPython/Gnuplot2.py (gp): Patch for Gnuplot.py 1.6 |
|
2855 | 2859 | compatibility. Thanks to Hayden Callow |
|
2856 | 2860 | <h.callow-AT-elec.canterbury.ac.nz> |
|
2857 | 2861 | |
|
2858 | 2862 | 2002-10-04 Fernando Perez <fperez@colorado.edu> |
|
2859 | 2863 | |
|
2860 | 2864 | * IPython/Gnuplot2.py (PlotItem): Added 'index' option for |
|
2861 | 2865 | Gnuplot.File objects. |
|
2862 | 2866 | |
|
2863 | 2867 | 2002-07-23 Fernando Perez <fperez@colorado.edu> |
|
2864 | 2868 | |
|
2865 | 2869 | * IPython/genutils.py (timing): Added timings() and timing() for |
|
2866 | 2870 | quick access to the most commonly needed data, the execution |
|
2867 | 2871 | times. Old timing() renamed to timings_out(). |
|
2868 | 2872 | |
|
2869 | 2873 | 2002-07-18 Fernando Perez <fperez@colorado.edu> |
|
2870 | 2874 | |
|
2871 | 2875 | * IPython/Shell.py (IPShellEmbed.restore_system_completer): fixed |
|
2872 | 2876 | bug with nested instances disrupting the parent's tab completion. |
|
2873 | 2877 | |
|
2874 | 2878 | * IPython/iplib.py (all_completions): Added Alex Schmolck's |
|
2875 | 2879 | all_completions code to begin the emacs integration. |
|
2876 | 2880 | |
|
2877 | 2881 | * IPython/Gnuplot2.py (zip_items): Added optional 'titles' |
|
2878 | 2882 | argument to allow titling individual arrays when plotting. |
|
2879 | 2883 | |
|
2880 | 2884 | 2002-07-15 Fernando Perez <fperez@colorado.edu> |
|
2881 | 2885 | |
|
2882 | 2886 | * setup.py (make_shortcut): changed to retrieve the value of |
|
2883 | 2887 | 'Program Files' directory from the registry (this value changes in |
|
2884 | 2888 | non-english versions of Windows). Thanks to Thomas Fanslau |
|
2885 | 2889 | <tfanslau-AT-gmx.de> for the report. |
|
2886 | 2890 | |
|
2887 | 2891 | 2002-07-10 Fernando Perez <fperez@colorado.edu> |
|
2888 | 2892 | |
|
2889 | 2893 | * IPython/ultraTB.py (VerboseTB.debugger): enabled workaround for |
|
2890 | 2894 | a bug in pdb, which crashes if a line with only whitespace is |
|
2891 | 2895 | entered. Bug report submitted to sourceforge. |
|
2892 | 2896 | |
|
2893 | 2897 | 2002-07-09 Fernando Perez <fperez@colorado.edu> |
|
2894 | 2898 | |
|
2895 | 2899 | * IPython/ultraTB.py (VerboseTB.nullrepr): fixed rare crash when |
|
2896 | 2900 | reporting exceptions (it's a bug in inspect.py, I just set a |
|
2897 | 2901 | workaround). |
|
2898 | 2902 | |
|
2899 | 2903 | 2002-07-08 Fernando Perez <fperez@colorado.edu> |
|
2900 | 2904 | |
|
2901 | 2905 | * IPython/iplib.py (InteractiveShell.__init__): fixed reference to |
|
2902 | 2906 | __IPYTHON__ in __builtins__ to show up in user_ns. |
|
2903 | 2907 | |
|
2904 | 2908 | 2002-07-03 Fernando Perez <fperez@colorado.edu> |
|
2905 | 2909 | |
|
2906 | 2910 | * IPython/GnuplotInteractive.py (magic_gp_set_default): changed |
|
2907 | 2911 | name from @gp_set_instance to @gp_set_default. |
|
2908 | 2912 | |
|
2909 | 2913 | * IPython/ipmaker.py (make_IPython): default editor value set to |
|
2910 | 2914 | '0' (a string), to match the rc file. Otherwise will crash when |
|
2911 | 2915 | .strip() is called on it. |
|
2912 | 2916 | |
|
2913 | 2917 | |
|
2914 | 2918 | 2002-06-28 Fernando Perez <fperez@colorado.edu> |
|
2915 | 2919 | |
|
2916 | 2920 | * IPython/iplib.py (InteractiveShell.safe_execfile): fix importing |
|
2917 | 2921 | of files in current directory when a file is executed via |
|
2918 | 2922 | @run. Patch also by RA <ralf_ahlbrink-AT-web.de>. |
|
2919 | 2923 | |
|
2920 | 2924 | * setup.py (manfiles): fix for rpm builds, submitted by RA |
|
2921 | 2925 | <ralf_ahlbrink-AT-web.de>. Now we have RPMs! |
|
2922 | 2926 | |
|
2923 | 2927 | * IPython/ipmaker.py (make_IPython): fixed lookup of default |
|
2924 | 2928 | editor when set to '0'. Problem was, '0' evaluates to True (it's a |
|
2925 | 2929 | string!). A. Schmolck caught this one. |
|
2926 | 2930 | |
|
2927 | 2931 | 2002-06-27 Fernando Perez <fperez@colorado.edu> |
|
2928 | 2932 | |
|
2929 | 2933 | * IPython/ipmaker.py (make_IPython): fixed bug when running user |
|
2930 | 2934 | defined files at the cmd line. __name__ wasn't being set to |
|
2931 | 2935 | __main__. |
|
2932 | 2936 | |
|
2933 | 2937 | * IPython/Gnuplot2.py (zip_items): improved it so it can plot also |
|
2934 | 2938 | regular lists and tuples besides Numeric arrays. |
|
2935 | 2939 | |
|
2936 | 2940 | * IPython/Prompts.py (CachedOutput.__call__): Added output |
|
2937 | 2941 | supression for input ending with ';'. Similar to Mathematica and |
|
2938 | 2942 | Matlab. The _* vars and Out[] list are still updated, just like |
|
2939 | 2943 | Mathematica behaves. |
|
2940 | 2944 | |
|
2941 | 2945 | 2002-06-25 Fernando Perez <fperez@colorado.edu> |
|
2942 | 2946 | |
|
2943 | 2947 | * IPython/ConfigLoader.py (ConfigLoader.load): fixed checking of |
|
2944 | 2948 | .ini extensions for profiels under Windows. |
|
2945 | 2949 | |
|
2946 | 2950 | * IPython/OInspect.py (Inspector.pinfo): improved alignment of |
|
2947 | 2951 | string form. Fix contributed by Alexander Schmolck |
|
2948 | 2952 | <a.schmolck-AT-gmx.net> |
|
2949 | 2953 | |
|
2950 | 2954 | * IPython/GnuplotRuntime.py (gp_new): new function. Returns a |
|
2951 | 2955 | pre-configured Gnuplot instance. |
|
2952 | 2956 | |
|
2953 | 2957 | 2002-06-21 Fernando Perez <fperez@colorado.edu> |
|
2954 | 2958 | |
|
2955 | 2959 | * IPython/numutils.py (exp_safe): new function, works around the |
|
2956 | 2960 | underflow problems in Numeric. |
|
2957 | 2961 | (log2): New fn. Safe log in base 2: returns exact integer answer |
|
2958 | 2962 | for exact integer powers of 2. |
|
2959 | 2963 | |
|
2960 | 2964 | * IPython/Magic.py (get_py_filename): fixed it not expanding '~' |
|
2961 | 2965 | properly. |
|
2962 | 2966 | |
|
2963 | 2967 | 2002-06-20 Fernando Perez <fperez@colorado.edu> |
|
2964 | 2968 | |
|
2965 | 2969 | * IPython/genutils.py (timing): new function like |
|
2966 | 2970 | Mathematica's. Similar to time_test, but returns more info. |
|
2967 | 2971 | |
|
2968 | 2972 | 2002-06-18 Fernando Perez <fperez@colorado.edu> |
|
2969 | 2973 | |
|
2970 | 2974 | * IPython/Magic.py (Magic.magic_save): modified @save and @r |
|
2971 | 2975 | according to Mike Heeter's suggestions. |
|
2972 | 2976 | |
|
2973 | 2977 | 2002-06-16 Fernando Perez <fperez@colorado.edu> |
|
2974 | 2978 | |
|
2975 | 2979 | * IPython/GnuplotRuntime.py: Massive overhaul to the Gnuplot |
|
2976 | 2980 | system. GnuplotMagic is gone as a user-directory option. New files |
|
2977 | 2981 | make it easier to use all the gnuplot stuff both from external |
|
2978 | 2982 | programs as well as from IPython. Had to rewrite part of |
|
2979 | 2983 | hardcopy() b/c of a strange bug: often the ps files simply don't |
|
2980 | 2984 | get created, and require a repeat of the command (often several |
|
2981 | 2985 | times). |
|
2982 | 2986 | |
|
2983 | 2987 | * IPython/ultraTB.py (AutoFormattedTB.__call__): changed to |
|
2984 | 2988 | resolve output channel at call time, so that if sys.stderr has |
|
2985 | 2989 | been redirected by user this gets honored. |
|
2986 | 2990 | |
|
2987 | 2991 | 2002-06-13 Fernando Perez <fperez@colorado.edu> |
|
2988 | 2992 | |
|
2989 | 2993 | * IPython/Shell.py (IPShell.__init__): Changed IPythonShell to |
|
2990 | 2994 | IPShell. Kept a copy with the old names to avoid breaking people's |
|
2991 | 2995 | embedded code. |
|
2992 | 2996 | |
|
2993 | 2997 | * IPython/ipython: simplified it to the bare minimum after |
|
2994 | 2998 | Holger's suggestions. Added info about how to use it in |
|
2995 | 2999 | PYTHONSTARTUP. |
|
2996 | 3000 | |
|
2997 | 3001 | * IPython/Shell.py (IPythonShell): changed the options passing |
|
2998 | 3002 | from a string with funky %s replacements to a straight list. Maybe |
|
2999 | 3003 | a bit more typing, but it follows sys.argv conventions, so there's |
|
3000 | 3004 | less special-casing to remember. |
|
3001 | 3005 | |
|
3002 | 3006 | 2002-06-12 Fernando Perez <fperez@colorado.edu> |
|
3003 | 3007 | |
|
3004 | 3008 | * IPython/Magic.py (Magic.magic_r): new magic auto-repeat |
|
3005 | 3009 | command. Thanks to a suggestion by Mike Heeter. |
|
3006 | 3010 | (Magic.magic_pfile): added behavior to look at filenames if given |
|
3007 | 3011 | arg is not a defined object. |
|
3008 | 3012 | (Magic.magic_save): New @save function to save code snippets. Also |
|
3009 | 3013 | a Mike Heeter idea. |
|
3010 | 3014 | |
|
3011 | 3015 | * IPython/UserConfig/GnuplotMagic.py (plot): Improvements to |
|
3012 | 3016 | plot() and replot(). Much more convenient now, especially for |
|
3013 | 3017 | interactive use. |
|
3014 | 3018 | |
|
3015 | 3019 | * IPython/Magic.py (Magic.magic_run): Added .py automatically to |
|
3016 | 3020 | filenames. |
|
3017 | 3021 | |
|
3018 | 3022 | 2002-06-02 Fernando Perez <fperez@colorado.edu> |
|
3019 | 3023 | |
|
3020 | 3024 | * IPython/Struct.py (Struct.__init__): modified to admit |
|
3021 | 3025 | initialization via another struct. |
|
3022 | 3026 | |
|
3023 | 3027 | * IPython/genutils.py (SystemExec.__init__): New stateful |
|
3024 | 3028 | interface to xsys and bq. Useful for writing system scripts. |
|
3025 | 3029 | |
|
3026 | 3030 | 2002-05-30 Fernando Perez <fperez@colorado.edu> |
|
3027 | 3031 | |
|
3028 | 3032 | * MANIFEST.in: Changed docfile selection to exclude all the lyx |
|
3029 | 3033 | documents. This will make the user download smaller (it's getting |
|
3030 | 3034 | too big). |
|
3031 | 3035 | |
|
3032 | 3036 | 2002-05-29 Fernando Perez <fperez@colorado.edu> |
|
3033 | 3037 | |
|
3034 | 3038 | * IPython/iplib.py (_FakeModule.__init__): New class introduced to |
|
3035 | 3039 | fix problems with shelve and pickle. Seems to work, but I don't |
|
3036 | 3040 | know if corner cases break it. Thanks to Mike Heeter |
|
3037 | 3041 | <korora-AT-SDF.LONESTAR.ORG> for the bug reports and test cases. |
|
3038 | 3042 | |
|
3039 | 3043 | 2002-05-24 Fernando Perez <fperez@colorado.edu> |
|
3040 | 3044 | |
|
3041 | 3045 | * IPython/Magic.py (Macro.__init__): fixed magics embedded in |
|
3042 | 3046 | macros having broken. |
|
3043 | 3047 | |
|
3044 | 3048 | 2002-05-21 Fernando Perez <fperez@colorado.edu> |
|
3045 | 3049 | |
|
3046 | 3050 | * IPython/Magic.py (Magic.magic_logstart): fixed recently |
|
3047 | 3051 | introduced logging bug: all history before logging started was |
|
3048 | 3052 | being written one character per line! This came from the redesign |
|
3049 | 3053 | of the input history as a special list which slices to strings, |
|
3050 | 3054 | not to lists. |
|
3051 | 3055 | |
|
3052 | 3056 | 2002-05-20 Fernando Perez <fperez@colorado.edu> |
|
3053 | 3057 | |
|
3054 | 3058 | * IPython/Prompts.py (CachedOutput.__init__): made the color table |
|
3055 | 3059 | be an attribute of all classes in this module. The design of these |
|
3056 | 3060 | classes needs some serious overhauling. |
|
3057 | 3061 | |
|
3058 | 3062 | * IPython/DPyGetOpt.py (DPyGetOpt.setPosixCompliance): fixed bug |
|
3059 | 3063 | which was ignoring '_' in option names. |
|
3060 | 3064 | |
|
3061 | 3065 | * IPython/ultraTB.py (FormattedTB.__init__): Changed |
|
3062 | 3066 | 'Verbose_novars' to 'Context' and made it the new default. It's a |
|
3063 | 3067 | bit more readable and also safer than verbose. |
|
3064 | 3068 | |
|
3065 | 3069 | * IPython/PyColorize.py (Parser.__call__): Fixed coloring of |
|
3066 | 3070 | triple-quoted strings. |
|
3067 | 3071 | |
|
3068 | 3072 | * IPython/OInspect.py (__all__): new module exposing the object |
|
3069 | 3073 | introspection facilities. Now the corresponding magics are dummy |
|
3070 | 3074 | wrappers around this. Having this module will make it much easier |
|
3071 | 3075 | to put these functions into our modified pdb. |
|
3072 | 3076 | This new object inspector system uses the new colorizing module, |
|
3073 | 3077 | so source code and other things are nicely syntax highlighted. |
|
3074 | 3078 | |
|
3075 | 3079 | 2002-05-18 Fernando Perez <fperez@colorado.edu> |
|
3076 | 3080 | |
|
3077 | 3081 | * IPython/ColorANSI.py: Split the coloring tools into a separate |
|
3078 | 3082 | module so I can use them in other code easier (they were part of |
|
3079 | 3083 | ultraTB). |
|
3080 | 3084 | |
|
3081 | 3085 | 2002-05-17 Fernando Perez <fperez@colorado.edu> |
|
3082 | 3086 | |
|
3083 | 3087 | * IPython/UserConfig/GnuplotMagic.py (magic_gp_set_instance): |
|
3084 | 3088 | fixed it to set the global 'g' also to the called instance, as |
|
3085 | 3089 | long as 'g' was still a gnuplot instance (so it doesn't overwrite |
|
3086 | 3090 | user's 'g' variables). |
|
3087 | 3091 | |
|
3088 | 3092 | * IPython/iplib.py (InteractiveShell.__init__): Added In/Out |
|
3089 | 3093 | global variables (aliases to _ih,_oh) so that users which expect |
|
3090 | 3094 | In[5] or Out[7] to work aren't unpleasantly surprised. |
|
3091 | 3095 | (InputList.__getslice__): new class to allow executing slices of |
|
3092 | 3096 | input history directly. Very simple class, complements the use of |
|
3093 | 3097 | macros. |
|
3094 | 3098 | |
|
3095 | 3099 | 2002-05-16 Fernando Perez <fperez@colorado.edu> |
|
3096 | 3100 | |
|
3097 | 3101 | * setup.py (docdirbase): make doc directory be just doc/IPython |
|
3098 | 3102 | without version numbers, it will reduce clutter for users. |
|
3099 | 3103 | |
|
3100 | 3104 | * IPython/Magic.py (Magic.magic_run): Add explicit local dict to |
|
3101 | 3105 | execfile call to prevent possible memory leak. See for details: |
|
3102 | 3106 | http://mail.python.org/pipermail/python-list/2002-February/088476.html |
|
3103 | 3107 | |
|
3104 | 3108 | 2002-05-15 Fernando Perez <fperez@colorado.edu> |
|
3105 | 3109 | |
|
3106 | 3110 | * IPython/Magic.py (Magic.magic_psource): made the object |
|
3107 | 3111 | introspection names be more standard: pdoc, pdef, pfile and |
|
3108 | 3112 | psource. They all print/page their output, and it makes |
|
3109 | 3113 | remembering them easier. Kept old names for compatibility as |
|
3110 | 3114 | aliases. |
|
3111 | 3115 | |
|
3112 | 3116 | 2002-05-14 Fernando Perez <fperez@colorado.edu> |
|
3113 | 3117 | |
|
3114 | 3118 | * IPython/UserConfig/GnuplotMagic.py: I think I finally understood |
|
3115 | 3119 | what the mouse problem was. The trick is to use gnuplot with temp |
|
3116 | 3120 | files and NOT with pipes (for data communication), because having |
|
3117 | 3121 | both pipes and the mouse on is bad news. |
|
3118 | 3122 | |
|
3119 | 3123 | 2002-05-13 Fernando Perez <fperez@colorado.edu> |
|
3120 | 3124 | |
|
3121 | 3125 | * IPython/Magic.py (Magic._ofind): fixed namespace order search |
|
3122 | 3126 | bug. Information would be reported about builtins even when |
|
3123 | 3127 | user-defined functions overrode them. |
|
3124 | 3128 | |
|
3125 | 3129 | 2002-05-11 Fernando Perez <fperez@colorado.edu> |
|
3126 | 3130 | |
|
3127 | 3131 | * IPython/__init__.py (__all__): removed FlexCompleter from |
|
3128 | 3132 | __all__ so that things don't fail in platforms without readline. |
|
3129 | 3133 | |
|
3130 | 3134 | 2002-05-10 Fernando Perez <fperez@colorado.edu> |
|
3131 | 3135 | |
|
3132 | 3136 | * IPython/__init__.py (__all__): removed numutils from __all__ b/c |
|
3133 | 3137 | it requires Numeric, effectively making Numeric a dependency for |
|
3134 | 3138 | IPython. |
|
3135 | 3139 | |
|
3136 | 3140 | * Released 0.2.13 |
|
3137 | 3141 | |
|
3138 | 3142 | * IPython/Magic.py (Magic.magic_prun): big overhaul to the |
|
3139 | 3143 | profiler interface. Now all the major options from the profiler |
|
3140 | 3144 | module are directly supported in IPython, both for single |
|
3141 | 3145 | expressions (@prun) and for full programs (@run -p). |
|
3142 | 3146 | |
|
3143 | 3147 | 2002-05-09 Fernando Perez <fperez@colorado.edu> |
|
3144 | 3148 | |
|
3145 | 3149 | * IPython/Magic.py (Magic.magic_doc): fixed to show docstrings of |
|
3146 | 3150 | magic properly formatted for screen. |
|
3147 | 3151 | |
|
3148 | 3152 | * setup.py (make_shortcut): Changed things to put pdf version in |
|
3149 | 3153 | doc/ instead of doc/manual (had to change lyxport a bit). |
|
3150 | 3154 | |
|
3151 | 3155 | * IPython/Magic.py (Profile.string_stats): made profile runs go |
|
3152 | 3156 | through pager (they are long and a pager allows searching, saving, |
|
3153 | 3157 | etc.) |
|
3154 | 3158 | |
|
3155 | 3159 | 2002-05-08 Fernando Perez <fperez@colorado.edu> |
|
3156 | 3160 | |
|
3157 | 3161 | * Released 0.2.12 |
|
3158 | 3162 | |
|
3159 | 3163 | 2002-05-06 Fernando Perez <fperez@colorado.edu> |
|
3160 | 3164 | |
|
3161 | 3165 | * IPython/Magic.py (Magic.magic_hist): small bug fixed (recently |
|
3162 | 3166 | introduced); 'hist n1 n2' was broken. |
|
3163 | 3167 | (Magic.magic_pdb): added optional on/off arguments to @pdb |
|
3164 | 3168 | (Magic.magic_run): added option -i to @run, which executes code in |
|
3165 | 3169 | the IPython namespace instead of a clean one. Also added @irun as |
|
3166 | 3170 | an alias to @run -i. |
|
3167 | 3171 | |
|
3168 | 3172 | * IPython/UserConfig/GnuplotMagic.py (magic_gp_set_instance): |
|
3169 | 3173 | fixed (it didn't really do anything, the namespaces were wrong). |
|
3170 | 3174 | |
|
3171 | 3175 | * IPython/Debugger.py (__init__): Added workaround for python 2.1 |
|
3172 | 3176 | |
|
3173 | 3177 | * IPython/__init__.py (__all__): Fixed package namespace, now |
|
3174 | 3178 | 'import IPython' does give access to IPython.<all> as |
|
3175 | 3179 | expected. Also renamed __release__ to Release. |
|
3176 | 3180 | |
|
3177 | 3181 | * IPython/Debugger.py (__license__): created new Pdb class which |
|
3178 | 3182 | functions like a drop-in for the normal pdb.Pdb but does NOT |
|
3179 | 3183 | import readline by default. This way it doesn't muck up IPython's |
|
3180 | 3184 | readline handling, and now tab-completion finally works in the |
|
3181 | 3185 | debugger -- sort of. It completes things globally visible, but the |
|
3182 | 3186 | completer doesn't track the stack as pdb walks it. That's a bit |
|
3183 | 3187 | tricky, and I'll have to implement it later. |
|
3184 | 3188 | |
|
3185 | 3189 | 2002-05-05 Fernando Perez <fperez@colorado.edu> |
|
3186 | 3190 | |
|
3187 | 3191 | * IPython/Magic.py (Magic.magic_oinfo): fixed formatting bug for |
|
3188 | 3192 | magic docstrings when printed via ? (explicit \'s were being |
|
3189 | 3193 | printed). |
|
3190 | 3194 | |
|
3191 | 3195 | * IPython/ipmaker.py (make_IPython): fixed namespace |
|
3192 | 3196 | identification bug. Now variables loaded via logs or command-line |
|
3193 | 3197 | files are recognized in the interactive namespace by @who. |
|
3194 | 3198 | |
|
3195 | 3199 | * IPython/iplib.py (InteractiveShell.safe_execfile): Fixed bug in |
|
3196 | 3200 | log replay system stemming from the string form of Structs. |
|
3197 | 3201 | |
|
3198 | 3202 | * IPython/Magic.py (Macro.__init__): improved macros to properly |
|
3199 | 3203 | handle magic commands in them. |
|
3200 | 3204 | (Magic.magic_logstart): usernames are now expanded so 'logstart |
|
3201 | 3205 | ~/mylog' now works. |
|
3202 | 3206 | |
|
3203 | 3207 | * IPython/iplib.py (complete): fixed bug where paths starting with |
|
3204 | 3208 | '/' would be completed as magic names. |
|
3205 | 3209 | |
|
3206 | 3210 | 2002-05-04 Fernando Perez <fperez@colorado.edu> |
|
3207 | 3211 | |
|
3208 | 3212 | * IPython/Magic.py (Magic.magic_run): added options -p and -f to |
|
3209 | 3213 | allow running full programs under the profiler's control. |
|
3210 | 3214 | |
|
3211 | 3215 | * IPython/ultraTB.py (FormattedTB.__init__): Added Verbose_novars |
|
3212 | 3216 | mode to report exceptions verbosely but without formatting |
|
3213 | 3217 | variables. This addresses the issue of ipython 'freezing' (it's |
|
3214 | 3218 | not frozen, but caught in an expensive formatting loop) when huge |
|
3215 | 3219 | variables are in the context of an exception. |
|
3216 | 3220 | (VerboseTB.text): Added '--->' markers at line where exception was |
|
3217 | 3221 | triggered. Much clearer to read, especially in NoColor modes. |
|
3218 | 3222 | |
|
3219 | 3223 | * IPython/Magic.py (Magic.magic_run): bugfix: -n option had been |
|
3220 | 3224 | implemented in reverse when changing to the new parse_options(). |
|
3221 | 3225 | |
|
3222 | 3226 | 2002-05-03 Fernando Perez <fperez@colorado.edu> |
|
3223 | 3227 | |
|
3224 | 3228 | * IPython/Magic.py (Magic.parse_options): new function so that |
|
3225 | 3229 | magics can parse options easier. |
|
3226 | 3230 | (Magic.magic_prun): new function similar to profile.run(), |
|
3227 | 3231 | suggested by Chris Hart. |
|
3228 | 3232 | (Magic.magic_cd): fixed behavior so that it only changes if |
|
3229 | 3233 | directory actually is in history. |
|
3230 | 3234 | |
|
3231 | 3235 | * IPython/usage.py (__doc__): added information about potential |
|
3232 | 3236 | slowness of Verbose exception mode when there are huge data |
|
3233 | 3237 | structures to be formatted (thanks to Archie Paulson). |
|
3234 | 3238 | |
|
3235 | 3239 | * IPython/ipmaker.py (make_IPython): Changed default logging |
|
3236 | 3240 | (when simply called with -log) to use curr_dir/ipython.log in |
|
3237 | 3241 | rotate mode. Fixed crash which was occuring with -log before |
|
3238 | 3242 | (thanks to Jim Boyle). |
|
3239 | 3243 | |
|
3240 | 3244 | 2002-05-01 Fernando Perez <fperez@colorado.edu> |
|
3241 | 3245 | |
|
3242 | 3246 | * Released 0.2.11 for these fixes (mainly the ultraTB one which |
|
3243 | 3247 | was nasty -- though somewhat of a corner case). |
|
3244 | 3248 | |
|
3245 | 3249 | * IPython/ultraTB.py (AutoFormattedTB.text): renamed __text to |
|
3246 | 3250 | text (was a bug). |
|
3247 | 3251 | |
|
3248 | 3252 | 2002-04-30 Fernando Perez <fperez@colorado.edu> |
|
3249 | 3253 | |
|
3250 | 3254 | * IPython/UserConfig/GnuplotMagic.py (magic_gp): Minor fix to add |
|
3251 | 3255 | a print after ^D or ^C from the user so that the In[] prompt |
|
3252 | 3256 | doesn't over-run the gnuplot one. |
|
3253 | 3257 | |
|
3254 | 3258 | 2002-04-29 Fernando Perez <fperez@colorado.edu> |
|
3255 | 3259 | |
|
3256 | 3260 | * Released 0.2.10 |
|
3257 | 3261 | |
|
3258 | 3262 | * IPython/__release__.py (version): get date dynamically. |
|
3259 | 3263 | |
|
3260 | 3264 | * Misc. documentation updates thanks to Arnd's comments. Also ran |
|
3261 | 3265 | a full spellcheck on the manual (hadn't been done in a while). |
|
3262 | 3266 | |
|
3263 | 3267 | 2002-04-27 Fernando Perez <fperez@colorado.edu> |
|
3264 | 3268 | |
|
3265 | 3269 | * IPython/Magic.py (Magic.magic_logstart): Fixed bug where |
|
3266 | 3270 | starting a log in mid-session would reset the input history list. |
|
3267 | 3271 | |
|
3268 | 3272 | 2002-04-26 Fernando Perez <fperez@colorado.edu> |
|
3269 | 3273 | |
|
3270 | 3274 | * IPython/iplib.py (InteractiveShell.wait): Fixed bug where not |
|
3271 | 3275 | all files were being included in an update. Now anything in |
|
3272 | 3276 | UserConfig that matches [A-Za-z]*.py will go (this excludes |
|
3273 | 3277 | __init__.py) |
|
3274 | 3278 | |
|
3275 | 3279 | 2002-04-25 Fernando Perez <fperez@colorado.edu> |
|
3276 | 3280 | |
|
3277 | 3281 | * IPython/iplib.py (InteractiveShell.__init__): Added __IPYTHON__ |
|
3278 | 3282 | to __builtins__ so that any form of embedded or imported code can |
|
3279 | 3283 | test for being inside IPython. |
|
3280 | 3284 | |
|
3281 | 3285 | * IPython/UserConfig/GnuplotMagic.py: (magic_gp_set_instance): |
|
3282 | 3286 | changed to GnuplotMagic because it's now an importable module, |
|
3283 | 3287 | this makes the name follow that of the standard Gnuplot module. |
|
3284 | 3288 | GnuplotMagic can now be loaded at any time in mid-session. |
|
3285 | 3289 | |
|
3286 | 3290 | 2002-04-24 Fernando Perez <fperez@colorado.edu> |
|
3287 | 3291 | |
|
3288 | 3292 | * IPython/numutils.py: removed SIUnits. It doesn't properly set |
|
3289 | 3293 | the globals (IPython has its own namespace) and the |
|
3290 | 3294 | PhysicalQuantity stuff is much better anyway. |
|
3291 | 3295 | |
|
3292 | 3296 | * IPython/UserConfig/example-gnuplot.py (g2): Added gnuplot |
|
3293 | 3297 | embedding example to standard user directory for |
|
3294 | 3298 | distribution. Also put it in the manual. |
|
3295 | 3299 | |
|
3296 | 3300 | * IPython/numutils.py (gnuplot_exec): Changed to take a gnuplot |
|
3297 | 3301 | instance as first argument (so it doesn't rely on some obscure |
|
3298 | 3302 | hidden global). |
|
3299 | 3303 | |
|
3300 | 3304 | * IPython/UserConfig/ipythonrc.py: put () back in accepted |
|
3301 | 3305 | delimiters. While it prevents ().TAB from working, it allows |
|
3302 | 3306 | completions in open (... expressions. This is by far a more common |
|
3303 | 3307 | case. |
|
3304 | 3308 | |
|
3305 | 3309 | 2002-04-23 Fernando Perez <fperez@colorado.edu> |
|
3306 | 3310 | |
|
3307 | 3311 | * IPython/Extensions/InterpreterPasteInput.py: new |
|
3308 | 3312 | syntax-processing module for pasting lines with >>> or ... at the |
|
3309 | 3313 | start. |
|
3310 | 3314 | |
|
3311 | 3315 | * IPython/Extensions/PhysicalQ_Interactive.py |
|
3312 | 3316 | (PhysicalQuantityInteractive.__int__): fixed to work with either |
|
3313 | 3317 | Numeric or math. |
|
3314 | 3318 | |
|
3315 | 3319 | * IPython/UserConfig/ipythonrc-numeric.py: reorganized the |
|
3316 | 3320 | provided profiles. Now we have: |
|
3317 | 3321 | -math -> math module as * and cmath with its own namespace. |
|
3318 | 3322 | -numeric -> Numeric as *, plus gnuplot & grace |
|
3319 | 3323 | -physics -> same as before |
|
3320 | 3324 | |
|
3321 | 3325 | * IPython/Magic.py (Magic.magic_magic): Fixed bug where |
|
3322 | 3326 | user-defined magics wouldn't be found by @magic if they were |
|
3323 | 3327 | defined as class methods. Also cleaned up the namespace search |
|
3324 | 3328 | logic and the string building (to use %s instead of many repeated |
|
3325 | 3329 | string adds). |
|
3326 | 3330 | |
|
3327 | 3331 | * IPython/UserConfig/example-magic.py (magic_foo): updated example |
|
3328 | 3332 | of user-defined magics to operate with class methods (cleaner, in |
|
3329 | 3333 | line with the gnuplot code). |
|
3330 | 3334 | |
|
3331 | 3335 | 2002-04-22 Fernando Perez <fperez@colorado.edu> |
|
3332 | 3336 | |
|
3333 | 3337 | * setup.py: updated dependency list so that manual is updated when |
|
3334 | 3338 | all included files change. |
|
3335 | 3339 | |
|
3336 | 3340 | * IPython/ipmaker.py (make_IPython): Fixed bug which was ignoring |
|
3337 | 3341 | the delimiter removal option (the fix is ugly right now). |
|
3338 | 3342 | |
|
3339 | 3343 | * IPython/UserConfig/ipythonrc-physics.py: simplified not to load |
|
3340 | 3344 | all of the math profile (quicker loading, no conflict between |
|
3341 | 3345 | g-9.8 and g-gnuplot). |
|
3342 | 3346 | |
|
3343 | 3347 | * IPython/CrashHandler.py (CrashHandler.__call__): changed default |
|
3344 | 3348 | name of post-mortem files to IPython_crash_report.txt. |
|
3345 | 3349 | |
|
3346 | 3350 | * Cleanup/update of the docs. Added all the new readline info and |
|
3347 | 3351 | formatted all lists as 'real lists'. |
|
3348 | 3352 | |
|
3349 | 3353 | * IPython/ipmaker.py (make_IPython): removed now-obsolete |
|
3350 | 3354 | tab-completion options, since the full readline parse_and_bind is |
|
3351 | 3355 | now accessible. |
|
3352 | 3356 | |
|
3353 | 3357 | * IPython/iplib.py (InteractiveShell.init_readline): Changed |
|
3354 | 3358 | handling of readline options. Now users can specify any string to |
|
3355 | 3359 | be passed to parse_and_bind(), as well as the delimiters to be |
|
3356 | 3360 | removed. |
|
3357 | 3361 | (InteractiveShell.__init__): Added __name__ to the global |
|
3358 | 3362 | namespace so that things like Itpl which rely on its existence |
|
3359 | 3363 | don't crash. |
|
3360 | 3364 | (InteractiveShell._prefilter): Defined the default with a _ so |
|
3361 | 3365 | that prefilter() is easier to override, while the default one |
|
3362 | 3366 | remains available. |
|
3363 | 3367 | |
|
3364 | 3368 | 2002-04-18 Fernando Perez <fperez@colorado.edu> |
|
3365 | 3369 | |
|
3366 | 3370 | * Added information about pdb in the docs. |
|
3367 | 3371 | |
|
3368 | 3372 | 2002-04-17 Fernando Perez <fperez@colorado.edu> |
|
3369 | 3373 | |
|
3370 | 3374 | * IPython/ipmaker.py (make_IPython): added rc_override option to |
|
3371 | 3375 | allow passing config options at creation time which may override |
|
3372 | 3376 | anything set in the config files or command line. This is |
|
3373 | 3377 | particularly useful for configuring embedded instances. |
|
3374 | 3378 | |
|
3375 | 3379 | 2002-04-15 Fernando Perez <fperez@colorado.edu> |
|
3376 | 3380 | |
|
3377 | 3381 | * IPython/Logger.py (Logger.log): Fixed a nasty bug which could |
|
3378 | 3382 | crash embedded instances because of the input cache falling out of |
|
3379 | 3383 | sync with the output counter. |
|
3380 | 3384 | |
|
3381 | 3385 | * IPython/Shell.py (IPythonShellEmbed.__init__): added a debug |
|
3382 | 3386 | mode which calls pdb after an uncaught exception in IPython itself. |
|
3383 | 3387 | |
|
3384 | 3388 | 2002-04-14 Fernando Perez <fperez@colorado.edu> |
|
3385 | 3389 | |
|
3386 | 3390 | * IPython/iplib.py (InteractiveShell.showtraceback): pdb mucks up |
|
3387 | 3391 | readline, fix it back after each call. |
|
3388 | 3392 | |
|
3389 | 3393 | * IPython/ultraTB.py (AutoFormattedTB.__text): made text a private |
|
3390 | 3394 | method to force all access via __call__(), which guarantees that |
|
3391 | 3395 | traceback references are properly deleted. |
|
3392 | 3396 | |
|
3393 | 3397 | * IPython/Prompts.py (CachedOutput._display): minor fixes to |
|
3394 | 3398 | improve printing when pprint is in use. |
|
3395 | 3399 | |
|
3396 | 3400 | 2002-04-13 Fernando Perez <fperez@colorado.edu> |
|
3397 | 3401 | |
|
3398 | 3402 | * IPython/Shell.py (IPythonShellEmbed.__call__): SystemExit |
|
3399 | 3403 | exceptions aren't caught anymore. If the user triggers one, he |
|
3400 | 3404 | should know why he's doing it and it should go all the way up, |
|
3401 | 3405 | just like any other exception. So now @abort will fully kill the |
|
3402 | 3406 | embedded interpreter and the embedding code (unless that happens |
|
3403 | 3407 | to catch SystemExit). |
|
3404 | 3408 | |
|
3405 | 3409 | * IPython/ultraTB.py (VerboseTB.__init__): added a call_pdb flag |
|
3406 | 3410 | and a debugger() method to invoke the interactive pdb debugger |
|
3407 | 3411 | after printing exception information. Also added the corresponding |
|
3408 | 3412 | -pdb option and @pdb magic to control this feature, and updated |
|
3409 | 3413 | the docs. After a suggestion from Christopher Hart |
|
3410 | 3414 | (hart-AT-caltech.edu). |
|
3411 | 3415 | |
|
3412 | 3416 | 2002-04-12 Fernando Perez <fperez@colorado.edu> |
|
3413 | 3417 | |
|
3414 | 3418 | * IPython/Shell.py (IPythonShellEmbed.__init__): modified to use |
|
3415 | 3419 | the exception handlers defined by the user (not the CrashHandler) |
|
3416 | 3420 | so that user exceptions don't trigger an ipython bug report. |
|
3417 | 3421 | |
|
3418 | 3422 | * IPython/ultraTB.py (ColorTB.__init__): made the color scheme |
|
3419 | 3423 | configurable (it should have always been so). |
|
3420 | 3424 | |
|
3421 | 3425 | 2002-03-26 Fernando Perez <fperez@colorado.edu> |
|
3422 | 3426 | |
|
3423 | 3427 | * IPython/Shell.py (IPythonShellEmbed.__call__): many changes here |
|
3424 | 3428 | and there to fix embedding namespace issues. This should all be |
|
3425 | 3429 | done in a more elegant way. |
|
3426 | 3430 | |
|
3427 | 3431 | 2002-03-25 Fernando Perez <fperez@colorado.edu> |
|
3428 | 3432 | |
|
3429 | 3433 | * IPython/genutils.py (get_home_dir): Try to make it work under |
|
3430 | 3434 | win9x also. |
|
3431 | 3435 | |
|
3432 | 3436 | 2002-03-20 Fernando Perez <fperez@colorado.edu> |
|
3433 | 3437 | |
|
3434 | 3438 | * IPython/Shell.py (IPythonShellEmbed.__init__): leave |
|
3435 | 3439 | sys.displayhook untouched upon __init__. |
|
3436 | 3440 | |
|
3437 | 3441 | 2002-03-19 Fernando Perez <fperez@colorado.edu> |
|
3438 | 3442 | |
|
3439 | 3443 | * Released 0.2.9 (for embedding bug, basically). |
|
3440 | 3444 | |
|
3441 | 3445 | * IPython/Shell.py (IPythonShellEmbed.__call__): Trap SystemExit |
|
3442 | 3446 | exceptions so that enclosing shell's state can be restored. |
|
3443 | 3447 | |
|
3444 | 3448 | * Changed magic_gnuplot.py to magic-gnuplot.py to standardize |
|
3445 | 3449 | naming conventions in the .ipython/ dir. |
|
3446 | 3450 | |
|
3447 | 3451 | * IPython/iplib.py (InteractiveShell.init_readline): removed '-' |
|
3448 | 3452 | from delimiters list so filenames with - in them get expanded. |
|
3449 | 3453 | |
|
3450 | 3454 | * IPython/Shell.py (IPythonShellEmbed.__call__): fixed bug with |
|
3451 | 3455 | sys.displayhook not being properly restored after an embedded call. |
|
3452 | 3456 | |
|
3453 | 3457 | 2002-03-18 Fernando Perez <fperez@colorado.edu> |
|
3454 | 3458 | |
|
3455 | 3459 | * Released 0.2.8 |
|
3456 | 3460 | |
|
3457 | 3461 | * IPython/iplib.py (InteractiveShell.user_setup): fixed bug where |
|
3458 | 3462 | some files weren't being included in a -upgrade. |
|
3459 | 3463 | (InteractiveShell.init_readline): Added 'set show-all-if-ambiguous |
|
3460 | 3464 | on' so that the first tab completes. |
|
3461 | 3465 | (InteractiveShell.handle_magic): fixed bug with spaces around |
|
3462 | 3466 | quotes breaking many magic commands. |
|
3463 | 3467 | |
|
3464 | 3468 | * setup.py: added note about ignoring the syntax error messages at |
|
3465 | 3469 | installation. |
|
3466 | 3470 | |
|
3467 | 3471 | * IPython/UserConfig/magic_gnuplot.py (magic_gp): finished |
|
3468 | 3472 | streamlining the gnuplot interface, now there's only one magic @gp. |
|
3469 | 3473 | |
|
3470 | 3474 | 2002-03-17 Fernando Perez <fperez@colorado.edu> |
|
3471 | 3475 | |
|
3472 | 3476 | * IPython/UserConfig/magic_gnuplot.py: new name for the |
|
3473 | 3477 | example-magic_pm.py file. Much enhanced system, now with a shell |
|
3474 | 3478 | for communicating directly with gnuplot, one command at a time. |
|
3475 | 3479 | |
|
3476 | 3480 | * IPython/Magic.py (Magic.magic_run): added option -n to prevent |
|
3477 | 3481 | setting __name__=='__main__'. |
|
3478 | 3482 | |
|
3479 | 3483 | * IPython/UserConfig/example-magic_pm.py (magic_pm): Added |
|
3480 | 3484 | mini-shell for accessing gnuplot from inside ipython. Should |
|
3481 | 3485 | extend it later for grace access too. Inspired by Arnd's |
|
3482 | 3486 | suggestion. |
|
3483 | 3487 | |
|
3484 | 3488 | * IPython/iplib.py (InteractiveShell.handle_magic): fixed bug when |
|
3485 | 3489 | calling magic functions with () in their arguments. Thanks to Arnd |
|
3486 | 3490 | Baecker for pointing this to me. |
|
3487 | 3491 | |
|
3488 | 3492 | * IPython/numutils.py (sum_flat): fixed bug. Would recurse |
|
3489 | 3493 | infinitely for integer or complex arrays (only worked with floats). |
|
3490 | 3494 | |
|
3491 | 3495 | 2002-03-16 Fernando Perez <fperez@colorado.edu> |
|
3492 | 3496 | |
|
3493 | 3497 | * setup.py: Merged setup and setup_windows into a single script |
|
3494 | 3498 | which properly handles things for windows users. |
|
3495 | 3499 | |
|
3496 | 3500 | 2002-03-15 Fernando Perez <fperez@colorado.edu> |
|
3497 | 3501 | |
|
3498 | 3502 | * Big change to the manual: now the magics are all automatically |
|
3499 | 3503 | documented. This information is generated from their docstrings |
|
3500 | 3504 | and put in a latex file included by the manual lyx file. This way |
|
3501 | 3505 | we get always up to date information for the magics. The manual |
|
3502 | 3506 | now also has proper version information, also auto-synced. |
|
3503 | 3507 | |
|
3504 | 3508 | For this to work, an undocumented --magic_docstrings option was added. |
|
3505 | 3509 | |
|
3506 | 3510 | 2002-03-13 Fernando Perez <fperez@colorado.edu> |
|
3507 | 3511 | |
|
3508 | 3512 | * IPython/ultraTB.py (TermColors): fixed problem with dark colors |
|
3509 | 3513 | under CDE terminals. An explicit ;2 color reset is needed in the escapes. |
|
3510 | 3514 | |
|
3511 | 3515 | 2002-03-12 Fernando Perez <fperez@colorado.edu> |
|
3512 | 3516 | |
|
3513 | 3517 | * IPython/ultraTB.py (TermColors): changed color escapes again to |
|
3514 | 3518 | fix the (old, reintroduced) line-wrapping bug. Basically, if |
|
3515 | 3519 | \001..\002 aren't given in the color escapes, lines get wrapped |
|
3516 | 3520 | weirdly. But giving those screws up old xterms and emacs terms. So |
|
3517 | 3521 | I added some logic for emacs terms to be ok, but I can't identify old |
|
3518 | 3522 | xterms separately ($TERM=='xterm' for many terminals, like konsole). |
|
3519 | 3523 | |
|
3520 | 3524 | 2002-03-10 Fernando Perez <fperez@colorado.edu> |
|
3521 | 3525 | |
|
3522 | 3526 | * IPython/usage.py (__doc__): Various documentation cleanups and |
|
3523 | 3527 | updates, both in usage docstrings and in the manual. |
|
3524 | 3528 | |
|
3525 | 3529 | * IPython/Prompts.py (CachedOutput.set_colors): cleanups for |
|
3526 | 3530 | handling of caching. Set minimum acceptabe value for having a |
|
3527 | 3531 | cache at 20 values. |
|
3528 | 3532 | |
|
3529 | 3533 | * IPython/iplib.py (InteractiveShell.user_setup): moved the |
|
3530 | 3534 | install_first_time function to a method, renamed it and added an |
|
3531 | 3535 | 'upgrade' mode. Now people can update their config directory with |
|
3532 | 3536 | a simple command line switch (-upgrade, also new). |
|
3533 | 3537 | |
|
3534 | 3538 | * IPython/Magic.py (Magic.magic_pfile): Made @pfile an alias to |
|
3535 | 3539 | @file (convenient for automagic users under Python >= 2.2). |
|
3536 | 3540 | Removed @files (it seemed more like a plural than an abbrev. of |
|
3537 | 3541 | 'file show'). |
|
3538 | 3542 | |
|
3539 | 3543 | * IPython/iplib.py (install_first_time): Fixed crash if there were |
|
3540 | 3544 | backup files ('~') in .ipython/ install directory. |
|
3541 | 3545 | |
|
3542 | 3546 | * IPython/ipmaker.py (make_IPython): fixes for new prompt |
|
3543 | 3547 | system. Things look fine, but these changes are fairly |
|
3544 | 3548 | intrusive. Test them for a few days. |
|
3545 | 3549 | |
|
3546 | 3550 | * IPython/Prompts.py (CachedOutput.__init__): Massive rewrite of |
|
3547 | 3551 | the prompts system. Now all in/out prompt strings are user |
|
3548 | 3552 | controllable. This is particularly useful for embedding, as one |
|
3549 | 3553 | can tag embedded instances with particular prompts. |
|
3550 | 3554 | |
|
3551 | 3555 | Also removed global use of sys.ps1/2, which now allows nested |
|
3552 | 3556 | embeddings without any problems. Added command-line options for |
|
3553 | 3557 | the prompt strings. |
|
3554 | 3558 | |
|
3555 | 3559 | 2002-03-08 Fernando Perez <fperez@colorado.edu> |
|
3556 | 3560 | |
|
3557 | 3561 | * IPython/UserConfig/example-embed-short.py (ipshell): added |
|
3558 | 3562 | example file with the bare minimum code for embedding. |
|
3559 | 3563 | |
|
3560 | 3564 | * IPython/Shell.py (IPythonShellEmbed.set_dummy_mode): added |
|
3561 | 3565 | functionality for the embeddable shell to be activated/deactivated |
|
3562 | 3566 | either globally or at each call. |
|
3563 | 3567 | |
|
3564 | 3568 | * IPython/Prompts.py (Prompt1.auto_rewrite): Fixes the problem of |
|
3565 | 3569 | rewriting the prompt with '--->' for auto-inputs with proper |
|
3566 | 3570 | coloring. Now the previous UGLY hack in handle_auto() is gone, and |
|
3567 | 3571 | this is handled by the prompts class itself, as it should. |
|
3568 | 3572 | |
|
3569 | 3573 | 2002-03-05 Fernando Perez <fperez@colorado.edu> |
|
3570 | 3574 | |
|
3571 | 3575 | * IPython/Magic.py (Magic.magic_logstart): Changed @log to |
|
3572 | 3576 | @logstart to avoid name clashes with the math log function. |
|
3573 | 3577 | |
|
3574 | 3578 | * Big updates to X/Emacs section of the manual. |
|
3575 | 3579 | |
|
3576 | 3580 | * Removed ipython_emacs. Milan explained to me how to pass |
|
3577 | 3581 | arguments to ipython through Emacs. Some day I'm going to end up |
|
3578 | 3582 | learning some lisp... |
|
3579 | 3583 | |
|
3580 | 3584 | 2002-03-04 Fernando Perez <fperez@colorado.edu> |
|
3581 | 3585 | |
|
3582 | 3586 | * IPython/ipython_emacs: Created script to be used as the |
|
3583 | 3587 | py-python-command Emacs variable so we can pass IPython |
|
3584 | 3588 | parameters. I can't figure out how to tell Emacs directly to pass |
|
3585 | 3589 | parameters to IPython, so a dummy shell script will do it. |
|
3586 | 3590 | |
|
3587 | 3591 | Other enhancements made for things to work better under Emacs' |
|
3588 | 3592 | various types of terminals. Many thanks to Milan Zamazal |
|
3589 | 3593 | <pdm-AT-zamazal.org> for all the suggestions and pointers. |
|
3590 | 3594 | |
|
3591 | 3595 | 2002-03-01 Fernando Perez <fperez@colorado.edu> |
|
3592 | 3596 | |
|
3593 | 3597 | * IPython/ipmaker.py (make_IPython): added a --readline! option so |
|
3594 | 3598 | that loading of readline is now optional. This gives better |
|
3595 | 3599 | control to emacs users. |
|
3596 | 3600 | |
|
3597 | 3601 | * IPython/ultraTB.py (__date__): Modified color escape sequences |
|
3598 | 3602 | and now things work fine under xterm and in Emacs' term buffers |
|
3599 | 3603 | (though not shell ones). Well, in emacs you get colors, but all |
|
3600 | 3604 | seem to be 'light' colors (no difference between dark and light |
|
3601 | 3605 | ones). But the garbage chars are gone, and also in xterms. It |
|
3602 | 3606 | seems that now I'm using 'cleaner' ansi sequences. |
|
3603 | 3607 | |
|
3604 | 3608 | 2002-02-21 Fernando Perez <fperez@colorado.edu> |
|
3605 | 3609 | |
|
3606 | 3610 | * Released 0.2.7 (mainly to publish the scoping fix). |
|
3607 | 3611 | |
|
3608 | 3612 | * IPython/Logger.py (Logger.logstate): added. A corresponding |
|
3609 | 3613 | @logstate magic was created. |
|
3610 | 3614 | |
|
3611 | 3615 | * IPython/Magic.py: fixed nested scoping problem under Python |
|
3612 | 3616 | 2.1.x (automagic wasn't working). |
|
3613 | 3617 | |
|
3614 | 3618 | 2002-02-20 Fernando Perez <fperez@colorado.edu> |
|
3615 | 3619 | |
|
3616 | 3620 | * Released 0.2.6. |
|
3617 | 3621 | |
|
3618 | 3622 | * IPython/OutputTrap.py (OutputTrap.__init__): added a 'quiet' |
|
3619 | 3623 | option so that logs can come out without any headers at all. |
|
3620 | 3624 | |
|
3621 | 3625 | * IPython/UserConfig/ipythonrc-scipy.py: created a profile for |
|
3622 | 3626 | SciPy. |
|
3623 | 3627 | |
|
3624 | 3628 | * IPython/iplib.py (InteractiveShell.embed_mainloop): Changed so |
|
3625 | 3629 | that embedded IPython calls don't require vars() to be explicitly |
|
3626 | 3630 | passed. Now they are extracted from the caller's frame (code |
|
3627 | 3631 | snatched from Eric Jones' weave). Added better documentation to |
|
3628 | 3632 | the section on embedding and the example file. |
|
3629 | 3633 | |
|
3630 | 3634 | * IPython/genutils.py (page): Changed so that under emacs, it just |
|
3631 | 3635 | prints the string. You can then page up and down in the emacs |
|
3632 | 3636 | buffer itself. This is how the builtin help() works. |
|
3633 | 3637 | |
|
3634 | 3638 | * IPython/Prompts.py (CachedOutput.__call__): Fixed issue with |
|
3635 | 3639 | macro scoping: macros need to be executed in the user's namespace |
|
3636 | 3640 | to work as if they had been typed by the user. |
|
3637 | 3641 | |
|
3638 | 3642 | * IPython/Magic.py (Magic.magic_macro): Changed macros so they |
|
3639 | 3643 | execute automatically (no need to type 'exec...'). They then |
|
3640 | 3644 | behave like 'true macros'. The printing system was also modified |
|
3641 | 3645 | for this to work. |
|
3642 | 3646 | |
|
3643 | 3647 | 2002-02-19 Fernando Perez <fperez@colorado.edu> |
|
3644 | 3648 | |
|
3645 | 3649 | * IPython/genutils.py (page_file): new function for paging files |
|
3646 | 3650 | in an OS-independent way. Also necessary for file viewing to work |
|
3647 | 3651 | well inside Emacs buffers. |
|
3648 | 3652 | (page): Added checks for being in an emacs buffer. |
|
3649 | 3653 | (page): fixed bug for Windows ($TERM isn't set in Windows). Fixed |
|
3650 | 3654 | same bug in iplib. |
|
3651 | 3655 | |
|
3652 | 3656 | 2002-02-18 Fernando Perez <fperez@colorado.edu> |
|
3653 | 3657 | |
|
3654 | 3658 | * IPython/iplib.py (InteractiveShell.init_readline): modified use |
|
3655 | 3659 | of readline so that IPython can work inside an Emacs buffer. |
|
3656 | 3660 | |
|
3657 | 3661 | * IPython/ultraTB.py (AutoFormattedTB.__call__): some fixes to |
|
3658 | 3662 | method signatures (they weren't really bugs, but it looks cleaner |
|
3659 | 3663 | and keeps PyChecker happy). |
|
3660 | 3664 | |
|
3661 | 3665 | * IPython/ipmaker.py (make_IPython): added hooks Struct to __IP |
|
3662 | 3666 | for implementing various user-defined hooks. Currently only |
|
3663 | 3667 | display is done. |
|
3664 | 3668 | |
|
3665 | 3669 | * IPython/Prompts.py (CachedOutput._display): changed display |
|
3666 | 3670 | functions so that they can be dynamically changed by users easily. |
|
3667 | 3671 | |
|
3668 | 3672 | * IPython/Extensions/numeric_formats.py (num_display): added an |
|
3669 | 3673 | extension for printing NumPy arrays in flexible manners. It |
|
3670 | 3674 | doesn't do anything yet, but all the structure is in |
|
3671 | 3675 | place. Ultimately the plan is to implement output format control |
|
3672 | 3676 | like in Octave. |
|
3673 | 3677 | |
|
3674 | 3678 | * IPython/Magic.py (Magic.lsmagic): changed so that bound magic |
|
3675 | 3679 | methods are found at run-time by all the automatic machinery. |
|
3676 | 3680 | |
|
3677 | 3681 | 2002-02-17 Fernando Perez <fperez@colorado.edu> |
|
3678 | 3682 | |
|
3679 | 3683 | * setup_Windows.py (make_shortcut): documented. Cleaned up the |
|
3680 | 3684 | whole file a little. |
|
3681 | 3685 | |
|
3682 | 3686 | * ToDo: closed this document. Now there's a new_design.lyx |
|
3683 | 3687 | document for all new ideas. Added making a pdf of it for the |
|
3684 | 3688 | end-user distro. |
|
3685 | 3689 | |
|
3686 | 3690 | * IPython/Logger.py (Logger.switch_log): Created this to replace |
|
3687 | 3691 | logon() and logoff(). It also fixes a nasty crash reported by |
|
3688 | 3692 | Philip Hisley <compsys-AT-starpower.net>. Many thanks to him. |
|
3689 | 3693 | |
|
3690 | 3694 | * IPython/iplib.py (complete): got auto-completion to work with |
|
3691 | 3695 | automagic (I had wanted this for a long time). |
|
3692 | 3696 | |
|
3693 | 3697 | * IPython/Magic.py (Magic.magic_files): Added @files as an alias |
|
3694 | 3698 | to @file, since file() is now a builtin and clashes with automagic |
|
3695 | 3699 | for @file. |
|
3696 | 3700 | |
|
3697 | 3701 | * Made some new files: Prompts, CrashHandler, Magic, Logger. All |
|
3698 | 3702 | of this was previously in iplib, which had grown to more than 2000 |
|
3699 | 3703 | lines, way too long. No new functionality, but it makes managing |
|
3700 | 3704 | the code a bit easier. |
|
3701 | 3705 | |
|
3702 | 3706 | * IPython/iplib.py (IPythonCrashHandler.__call__): Added version |
|
3703 | 3707 | information to crash reports. |
|
3704 | 3708 | |
|
3705 | 3709 | 2002-02-12 Fernando Perez <fperez@colorado.edu> |
|
3706 | 3710 | |
|
3707 | 3711 | * Released 0.2.5. |
|
3708 | 3712 | |
|
3709 | 3713 | 2002-02-11 Fernando Perez <fperez@colorado.edu> |
|
3710 | 3714 | |
|
3711 | 3715 | * Wrote a relatively complete Windows installer. It puts |
|
3712 | 3716 | everything in place, creates Start Menu entries and fixes the |
|
3713 | 3717 | color issues. Nothing fancy, but it works. |
|
3714 | 3718 | |
|
3715 | 3719 | 2002-02-10 Fernando Perez <fperez@colorado.edu> |
|
3716 | 3720 | |
|
3717 | 3721 | * IPython/iplib.py (InteractiveShell.safe_execfile): added an |
|
3718 | 3722 | os.path.expanduser() call so that we can type @run ~/myfile.py and |
|
3719 | 3723 | have thigs work as expected. |
|
3720 | 3724 | |
|
3721 | 3725 | * IPython/genutils.py (page): fixed exception handling so things |
|
3722 | 3726 | work both in Unix and Windows correctly. Quitting a pager triggers |
|
3723 | 3727 | an IOError/broken pipe in Unix, and in windows not finding a pager |
|
3724 | 3728 | is also an IOError, so I had to actually look at the return value |
|
3725 | 3729 | of the exception, not just the exception itself. Should be ok now. |
|
3726 | 3730 | |
|
3727 | 3731 | * IPython/ultraTB.py (ColorSchemeTable.set_active_scheme): |
|
3728 | 3732 | modified to allow case-insensitive color scheme changes. |
|
3729 | 3733 | |
|
3730 | 3734 | 2002-02-09 Fernando Perez <fperez@colorado.edu> |
|
3731 | 3735 | |
|
3732 | 3736 | * IPython/genutils.py (native_line_ends): new function to leave |
|
3733 | 3737 | user config files with os-native line-endings. |
|
3734 | 3738 | |
|
3735 | 3739 | * README and manual updates. |
|
3736 | 3740 | |
|
3737 | 3741 | * IPython/genutils.py: fixed unicode bug: use types.StringTypes |
|
3738 | 3742 | instead of StringType to catch Unicode strings. |
|
3739 | 3743 | |
|
3740 | 3744 | * IPython/genutils.py (filefind): fixed bug for paths with |
|
3741 | 3745 | embedded spaces (very common in Windows). |
|
3742 | 3746 | |
|
3743 | 3747 | * IPython/ipmaker.py (make_IPython): added a '.ini' to the rc |
|
3744 | 3748 | files under Windows, so that they get automatically associated |
|
3745 | 3749 | with a text editor. Windows makes it a pain to handle |
|
3746 | 3750 | extension-less files. |
|
3747 | 3751 | |
|
3748 | 3752 | * IPython/iplib.py (InteractiveShell.init_readline): Made the |
|
3749 | 3753 | warning about readline only occur for Posix. In Windows there's no |
|
3750 | 3754 | way to get readline, so why bother with the warning. |
|
3751 | 3755 | |
|
3752 | 3756 | * IPython/Struct.py (Struct.__str__): fixed to use self.__dict__ |
|
3753 | 3757 | for __str__ instead of dir(self), since dir() changed in 2.2. |
|
3754 | 3758 | |
|
3755 | 3759 | * Ported to Windows! Tested on XP, I suspect it should work fine |
|
3756 | 3760 | on NT/2000, but I don't think it will work on 98 et al. That |
|
3757 | 3761 | series of Windows is such a piece of junk anyway that I won't try |
|
3758 | 3762 | porting it there. The XP port was straightforward, showed a few |
|
3759 | 3763 | bugs here and there (fixed all), in particular some string |
|
3760 | 3764 | handling stuff which required considering Unicode strings (which |
|
3761 | 3765 | Windows uses). This is good, but hasn't been too tested :) No |
|
3762 | 3766 | fancy installer yet, I'll put a note in the manual so people at |
|
3763 | 3767 | least make manually a shortcut. |
|
3764 | 3768 | |
|
3765 | 3769 | * IPython/iplib.py (Magic.magic_colors): Unified the color options |
|
3766 | 3770 | into a single one, "colors". This now controls both prompt and |
|
3767 | 3771 | exception color schemes, and can be changed both at startup |
|
3768 | 3772 | (either via command-line switches or via ipythonrc files) and at |
|
3769 | 3773 | runtime, with @colors. |
|
3770 | 3774 | (Magic.magic_run): renamed @prun to @run and removed the old |
|
3771 | 3775 | @run. The two were too similar to warrant keeping both. |
|
3772 | 3776 | |
|
3773 | 3777 | 2002-02-03 Fernando Perez <fperez@colorado.edu> |
|
3774 | 3778 | |
|
3775 | 3779 | * IPython/iplib.py (install_first_time): Added comment on how to |
|
3776 | 3780 | configure the color options for first-time users. Put a <return> |
|
3777 | 3781 | request at the end so that small-terminal users get a chance to |
|
3778 | 3782 | read the startup info. |
|
3779 | 3783 | |
|
3780 | 3784 | 2002-01-23 Fernando Perez <fperez@colorado.edu> |
|
3781 | 3785 | |
|
3782 | 3786 | * IPython/iplib.py (CachedOutput.update): Changed output memory |
|
3783 | 3787 | variable names from _o,_oo,_ooo,_o<n> to simply _,__,___,_<n>. For |
|
3784 | 3788 | input history we still use _i. Did this b/c these variable are |
|
3785 | 3789 | very commonly used in interactive work, so the less we need to |
|
3786 | 3790 | type the better off we are. |
|
3787 | 3791 | (Magic.magic_prun): updated @prun to better handle the namespaces |
|
3788 | 3792 | the file will run in, including a fix for __name__ not being set |
|
3789 | 3793 | before. |
|
3790 | 3794 | |
|
3791 | 3795 | 2002-01-20 Fernando Perez <fperez@colorado.edu> |
|
3792 | 3796 | |
|
3793 | 3797 | * IPython/ultraTB.py (VerboseTB.linereader): Fixed printing of |
|
3794 | 3798 | extra garbage for Python 2.2. Need to look more carefully into |
|
3795 | 3799 | this later. |
|
3796 | 3800 | |
|
3797 | 3801 | 2002-01-19 Fernando Perez <fperez@colorado.edu> |
|
3798 | 3802 | |
|
3799 | 3803 | * IPython/iplib.py (InteractiveShell.showtraceback): fixed to |
|
3800 | 3804 | display SyntaxError exceptions properly formatted when they occur |
|
3801 | 3805 | (they can be triggered by imported code). |
|
3802 | 3806 | |
|
3803 | 3807 | 2002-01-18 Fernando Perez <fperez@colorado.edu> |
|
3804 | 3808 | |
|
3805 | 3809 | * IPython/iplib.py (InteractiveShell.safe_execfile): now |
|
3806 | 3810 | SyntaxError exceptions are reported nicely formatted, instead of |
|
3807 | 3811 | spitting out only offset information as before. |
|
3808 | 3812 | (Magic.magic_prun): Added the @prun function for executing |
|
3809 | 3813 | programs with command line args inside IPython. |
|
3810 | 3814 | |
|
3811 | 3815 | 2002-01-16 Fernando Perez <fperez@colorado.edu> |
|
3812 | 3816 | |
|
3813 | 3817 | * IPython/iplib.py (Magic.magic_hist): Changed @hist and @dhist |
|
3814 | 3818 | to *not* include the last item given in a range. This brings their |
|
3815 | 3819 | behavior in line with Python's slicing: |
|
3816 | 3820 | a[n1:n2] -> a[n1]...a[n2-1] |
|
3817 | 3821 | It may be a bit less convenient, but I prefer to stick to Python's |
|
3818 | 3822 | conventions *everywhere*, so users never have to wonder. |
|
3819 | 3823 | (Magic.magic_macro): Added @macro function to ease the creation of |
|
3820 | 3824 | macros. |
|
3821 | 3825 | |
|
3822 | 3826 | 2002-01-05 Fernando Perez <fperez@colorado.edu> |
|
3823 | 3827 | |
|
3824 | 3828 | * Released 0.2.4. |
|
3825 | 3829 | |
|
3826 | 3830 | * IPython/iplib.py (Magic.magic_pdef): |
|
3827 | 3831 | (InteractiveShell.safe_execfile): report magic lines and error |
|
3828 | 3832 | lines without line numbers so one can easily copy/paste them for |
|
3829 | 3833 | re-execution. |
|
3830 | 3834 | |
|
3831 | 3835 | * Updated manual with recent changes. |
|
3832 | 3836 | |
|
3833 | 3837 | * IPython/iplib.py (Magic.magic_oinfo): added constructor |
|
3834 | 3838 | docstring printing when class? is called. Very handy for knowing |
|
3835 | 3839 | how to create class instances (as long as __init__ is well |
|
3836 | 3840 | documented, of course :) |
|
3837 | 3841 | (Magic.magic_doc): print both class and constructor docstrings. |
|
3838 | 3842 | (Magic.magic_pdef): give constructor info if passed a class and |
|
3839 | 3843 | __call__ info for callable object instances. |
|
3840 | 3844 | |
|
3841 | 3845 | 2002-01-04 Fernando Perez <fperez@colorado.edu> |
|
3842 | 3846 | |
|
3843 | 3847 | * Made deep_reload() off by default. It doesn't always work |
|
3844 | 3848 | exactly as intended, so it's probably safer to have it off. It's |
|
3845 | 3849 | still available as dreload() anyway, so nothing is lost. |
|
3846 | 3850 | |
|
3847 | 3851 | 2002-01-02 Fernando Perez <fperez@colorado.edu> |
|
3848 | 3852 | |
|
3849 | 3853 | * Released 0.2.3 (contacted R.Singh at CU about biopython course, |
|
3850 | 3854 | so I wanted an updated release). |
|
3851 | 3855 | |
|
3852 | 3856 | 2001-12-27 Fernando Perez <fperez@colorado.edu> |
|
3853 | 3857 | |
|
3854 | 3858 | * IPython/iplib.py (InteractiveShell.interact): Added the original |
|
3855 | 3859 | code from 'code.py' for this module in order to change the |
|
3856 | 3860 | handling of a KeyboardInterrupt. This was necessary b/c otherwise |
|
3857 | 3861 | the history cache would break when the user hit Ctrl-C, and |
|
3858 | 3862 | interact() offers no way to add any hooks to it. |
|
3859 | 3863 | |
|
3860 | 3864 | 2001-12-23 Fernando Perez <fperez@colorado.edu> |
|
3861 | 3865 | |
|
3862 | 3866 | * setup.py: added check for 'MANIFEST' before trying to remove |
|
3863 | 3867 | it. Thanks to Sean Reifschneider. |
|
3864 | 3868 | |
|
3865 | 3869 | 2001-12-22 Fernando Perez <fperez@colorado.edu> |
|
3866 | 3870 | |
|
3867 | 3871 | * Released 0.2.2. |
|
3868 | 3872 | |
|
3869 | 3873 | * Finished (reasonably) writing the manual. Later will add the |
|
3870 | 3874 | python-standard navigation stylesheets, but for the time being |
|
3871 | 3875 | it's fairly complete. Distribution will include html and pdf |
|
3872 | 3876 | versions. |
|
3873 | 3877 | |
|
3874 | 3878 | * Bugfix: '.' wasn't being added to sys.path. Thanks to Prabhu |
|
3875 | 3879 | (MayaVi author). |
|
3876 | 3880 | |
|
3877 | 3881 | 2001-12-21 Fernando Perez <fperez@colorado.edu> |
|
3878 | 3882 | |
|
3879 | 3883 | * Released 0.2.1. Barring any nasty bugs, this is it as far as a |
|
3880 | 3884 | good public release, I think (with the manual and the distutils |
|
3881 | 3885 | installer). The manual can use some work, but that can go |
|
3882 | 3886 | slowly. Otherwise I think it's quite nice for end users. Next |
|
3883 | 3887 | summer, rewrite the guts of it... |
|
3884 | 3888 | |
|
3885 | 3889 | * Changed format of ipythonrc files to use whitespace as the |
|
3886 | 3890 | separator instead of an explicit '='. Cleaner. |
|
3887 | 3891 | |
|
3888 | 3892 | 2001-12-20 Fernando Perez <fperez@colorado.edu> |
|
3889 | 3893 | |
|
3890 | 3894 | * Started a manual in LyX. For now it's just a quick merge of the |
|
3891 | 3895 | various internal docstrings and READMEs. Later it may grow into a |
|
3892 | 3896 | nice, full-blown manual. |
|
3893 | 3897 | |
|
3894 | 3898 | * Set up a distutils based installer. Installation should now be |
|
3895 | 3899 | trivially simple for end-users. |
|
3896 | 3900 | |
|
3897 | 3901 | 2001-12-11 Fernando Perez <fperez@colorado.edu> |
|
3898 | 3902 | |
|
3899 | 3903 | * Released 0.2.0. First public release, announced it at |
|
3900 | 3904 | comp.lang.python. From now on, just bugfixes... |
|
3901 | 3905 | |
|
3902 | 3906 | * Went through all the files, set copyright/license notices and |
|
3903 | 3907 | cleaned up things. Ready for release. |
|
3904 | 3908 | |
|
3905 | 3909 | 2001-12-10 Fernando Perez <fperez@colorado.edu> |
|
3906 | 3910 | |
|
3907 | 3911 | * Changed the first-time installer not to use tarfiles. It's more |
|
3908 | 3912 | robust now and less unix-dependent. Also makes it easier for |
|
3909 | 3913 | people to later upgrade versions. |
|
3910 | 3914 | |
|
3911 | 3915 | * Changed @exit to @abort to reflect the fact that it's pretty |
|
3912 | 3916 | brutal (a sys.exit()). The difference between @abort and Ctrl-D |
|
3913 | 3917 | becomes significant only when IPyhton is embedded: in that case, |
|
3914 | 3918 | C-D closes IPython only, but @abort kills the enclosing program |
|
3915 | 3919 | too (unless it had called IPython inside a try catching |
|
3916 | 3920 | SystemExit). |
|
3917 | 3921 | |
|
3918 | 3922 | * Created Shell module which exposes the actuall IPython Shell |
|
3919 | 3923 | classes, currently the normal and the embeddable one. This at |
|
3920 | 3924 | least offers a stable interface we won't need to change when |
|
3921 | 3925 | (later) the internals are rewritten. That rewrite will be confined |
|
3922 | 3926 | to iplib and ipmaker, but the Shell interface should remain as is. |
|
3923 | 3927 | |
|
3924 | 3928 | * Added embed module which offers an embeddable IPShell object, |
|
3925 | 3929 | useful to fire up IPython *inside* a running program. Great for |
|
3926 | 3930 | debugging or dynamical data analysis. |
|
3927 | 3931 | |
|
3928 | 3932 | 2001-12-08 Fernando Perez <fperez@colorado.edu> |
|
3929 | 3933 | |
|
3930 | 3934 | * Fixed small bug preventing seeing info from methods of defined |
|
3931 | 3935 | objects (incorrect namespace in _ofind()). |
|
3932 | 3936 | |
|
3933 | 3937 | * Documentation cleanup. Moved the main usage docstrings to a |
|
3934 | 3938 | separate file, usage.py (cleaner to maintain, and hopefully in the |
|
3935 | 3939 | future some perlpod-like way of producing interactive, man and |
|
3936 | 3940 | html docs out of it will be found). |
|
3937 | 3941 | |
|
3938 | 3942 | * Added @profile to see your profile at any time. |
|
3939 | 3943 | |
|
3940 | 3944 | * Added @p as an alias for 'print'. It's especially convenient if |
|
3941 | 3945 | using automagic ('p x' prints x). |
|
3942 | 3946 | |
|
3943 | 3947 | * Small cleanups and fixes after a pychecker run. |
|
3944 | 3948 | |
|
3945 | 3949 | * Changed the @cd command to handle @cd - and @cd -<n> for |
|
3946 | 3950 | visiting any directory in _dh. |
|
3947 | 3951 | |
|
3948 | 3952 | * Introduced _dh, a history of visited directories. @dhist prints |
|
3949 | 3953 | it out with numbers. |
|
3950 | 3954 | |
|
3951 | 3955 | 2001-12-07 Fernando Perez <fperez@colorado.edu> |
|
3952 | 3956 | |
|
3953 | 3957 | * Released 0.1.22 |
|
3954 | 3958 | |
|
3955 | 3959 | * Made initialization a bit more robust against invalid color |
|
3956 | 3960 | options in user input (exit, not traceback-crash). |
|
3957 | 3961 | |
|
3958 | 3962 | * Changed the bug crash reporter to write the report only in the |
|
3959 | 3963 | user's .ipython directory. That way IPython won't litter people's |
|
3960 | 3964 | hard disks with crash files all over the place. Also print on |
|
3961 | 3965 | screen the necessary mail command. |
|
3962 | 3966 | |
|
3963 | 3967 | * With the new ultraTB, implemented LightBG color scheme for light |
|
3964 | 3968 | background terminals. A lot of people like white backgrounds, so I |
|
3965 | 3969 | guess we should at least give them something readable. |
|
3966 | 3970 | |
|
3967 | 3971 | 2001-12-06 Fernando Perez <fperez@colorado.edu> |
|
3968 | 3972 | |
|
3969 | 3973 | * Modified the structure of ultraTB. Now there's a proper class |
|
3970 | 3974 | for tables of color schemes which allow adding schemes easily and |
|
3971 | 3975 | switching the active scheme without creating a new instance every |
|
3972 | 3976 | time (which was ridiculous). The syntax for creating new schemes |
|
3973 | 3977 | is also cleaner. I think ultraTB is finally done, with a clean |
|
3974 | 3978 | class structure. Names are also much cleaner (now there's proper |
|
3975 | 3979 | color tables, no need for every variable to also have 'color' in |
|
3976 | 3980 | its name). |
|
3977 | 3981 | |
|
3978 | 3982 | * Broke down genutils into separate files. Now genutils only |
|
3979 | 3983 | contains utility functions, and classes have been moved to their |
|
3980 | 3984 | own files (they had enough independent functionality to warrant |
|
3981 | 3985 | it): ConfigLoader, OutputTrap, Struct. |
|
3982 | 3986 | |
|
3983 | 3987 | 2001-12-05 Fernando Perez <fperez@colorado.edu> |
|
3984 | 3988 | |
|
3985 | 3989 | * IPython turns 21! Released version 0.1.21, as a candidate for |
|
3986 | 3990 | public consumption. If all goes well, release in a few days. |
|
3987 | 3991 | |
|
3988 | 3992 | * Fixed path bug (files in Extensions/ directory wouldn't be found |
|
3989 | 3993 | unless IPython/ was explicitly in sys.path). |
|
3990 | 3994 | |
|
3991 | 3995 | * Extended the FlexCompleter class as MagicCompleter to allow |
|
3992 | 3996 | completion of @-starting lines. |
|
3993 | 3997 | |
|
3994 | 3998 | * Created __release__.py file as a central repository for release |
|
3995 | 3999 | info that other files can read from. |
|
3996 | 4000 | |
|
3997 | 4001 | * Fixed small bug in logging: when logging was turned on in |
|
3998 | 4002 | mid-session, old lines with special meanings (!@?) were being |
|
3999 | 4003 | logged without the prepended comment, which is necessary since |
|
4000 | 4004 | they are not truly valid python syntax. This should make session |
|
4001 | 4005 | restores produce less errors. |
|
4002 | 4006 | |
|
4003 | 4007 | * The namespace cleanup forced me to make a FlexCompleter class |
|
4004 | 4008 | which is nothing but a ripoff of rlcompleter, but with selectable |
|
4005 | 4009 | namespace (rlcompleter only works in __main__.__dict__). I'll try |
|
4006 | 4010 | to submit a note to the authors to see if this change can be |
|
4007 | 4011 | incorporated in future rlcompleter releases (Dec.6: done) |
|
4008 | 4012 | |
|
4009 | 4013 | * More fixes to namespace handling. It was a mess! Now all |
|
4010 | 4014 | explicit references to __main__.__dict__ are gone (except when |
|
4011 | 4015 | really needed) and everything is handled through the namespace |
|
4012 | 4016 | dicts in the IPython instance. We seem to be getting somewhere |
|
4013 | 4017 | with this, finally... |
|
4014 | 4018 | |
|
4015 | 4019 | * Small documentation updates. |
|
4016 | 4020 | |
|
4017 | 4021 | * Created the Extensions directory under IPython (with an |
|
4018 | 4022 | __init__.py). Put the PhysicalQ stuff there. This directory should |
|
4019 | 4023 | be used for all special-purpose extensions. |
|
4020 | 4024 | |
|
4021 | 4025 | * File renaming: |
|
4022 | 4026 | ipythonlib --> ipmaker |
|
4023 | 4027 | ipplib --> iplib |
|
4024 | 4028 | This makes a bit more sense in terms of what these files actually do. |
|
4025 | 4029 | |
|
4026 | 4030 | * Moved all the classes and functions in ipythonlib to ipplib, so |
|
4027 | 4031 | now ipythonlib only has make_IPython(). This will ease up its |
|
4028 | 4032 | splitting in smaller functional chunks later. |
|
4029 | 4033 | |
|
4030 | 4034 | * Cleaned up (done, I think) output of @whos. Better column |
|
4031 | 4035 | formatting, and now shows str(var) for as much as it can, which is |
|
4032 | 4036 | typically what one gets with a 'print var'. |
|
4033 | 4037 | |
|
4034 | 4038 | 2001-12-04 Fernando Perez <fperez@colorado.edu> |
|
4035 | 4039 | |
|
4036 | 4040 | * Fixed namespace problems. Now builtin/IPyhton/user names get |
|
4037 | 4041 | properly reported in their namespace. Internal namespace handling |
|
4038 | 4042 | is finally getting decent (not perfect yet, but much better than |
|
4039 | 4043 | the ad-hoc mess we had). |
|
4040 | 4044 | |
|
4041 | 4045 | * Removed -exit option. If people just want to run a python |
|
4042 | 4046 | script, that's what the normal interpreter is for. Less |
|
4043 | 4047 | unnecessary options, less chances for bugs. |
|
4044 | 4048 | |
|
4045 | 4049 | * Added a crash handler which generates a complete post-mortem if |
|
4046 | 4050 | IPython crashes. This will help a lot in tracking bugs down the |
|
4047 | 4051 | road. |
|
4048 | 4052 | |
|
4049 | 4053 | * Fixed nasty bug in auto-evaluation part of prefilter(). Names |
|
4050 | 4054 | which were boud to functions being reassigned would bypass the |
|
4051 | 4055 | logger, breaking the sync of _il with the prompt counter. This |
|
4052 | 4056 | would then crash IPython later when a new line was logged. |
|
4053 | 4057 | |
|
4054 | 4058 | 2001-12-02 Fernando Perez <fperez@colorado.edu> |
|
4055 | 4059 | |
|
4056 | 4060 | * Made IPython a package. This means people don't have to clutter |
|
4057 | 4061 | their sys.path with yet another directory. Changed the INSTALL |
|
4058 | 4062 | file accordingly. |
|
4059 | 4063 | |
|
4060 | 4064 | * Cleaned up the output of @who_ls, @who and @whos. @who_ls now |
|
4061 | 4065 | sorts its output (so @who shows it sorted) and @whos formats the |
|
4062 | 4066 | table according to the width of the first column. Nicer, easier to |
|
4063 | 4067 | read. Todo: write a generic table_format() which takes a list of |
|
4064 | 4068 | lists and prints it nicely formatted, with optional row/column |
|
4065 | 4069 | separators and proper padding and justification. |
|
4066 | 4070 | |
|
4067 | 4071 | * Released 0.1.20 |
|
4068 | 4072 | |
|
4069 | 4073 | * Fixed bug in @log which would reverse the inputcache list (a |
|
4070 | 4074 | copy operation was missing). |
|
4071 | 4075 | |
|
4072 | 4076 | * Code cleanup. @config was changed to use page(). Better, since |
|
4073 | 4077 | its output is always quite long. |
|
4074 | 4078 | |
|
4075 | 4079 | * Itpl is back as a dependency. I was having too many problems |
|
4076 | 4080 | getting the parametric aliases to work reliably, and it's just |
|
4077 | 4081 | easier to code weird string operations with it than playing %()s |
|
4078 | 4082 | games. It's only ~6k, so I don't think it's too big a deal. |
|
4079 | 4083 | |
|
4080 | 4084 | * Found (and fixed) a very nasty bug with history. !lines weren't |
|
4081 | 4085 | getting cached, and the out of sync caches would crash |
|
4082 | 4086 | IPython. Fixed it by reorganizing the prefilter/handlers/logger |
|
4083 | 4087 | division of labor a bit better. Bug fixed, cleaner structure. |
|
4084 | 4088 | |
|
4085 | 4089 | 2001-12-01 Fernando Perez <fperez@colorado.edu> |
|
4086 | 4090 | |
|
4087 | 4091 | * Released 0.1.19 |
|
4088 | 4092 | |
|
4089 | 4093 | * Added option -n to @hist to prevent line number printing. Much |
|
4090 | 4094 | easier to copy/paste code this way. |
|
4091 | 4095 | |
|
4092 | 4096 | * Created global _il to hold the input list. Allows easy |
|
4093 | 4097 | re-execution of blocks of code by slicing it (inspired by Janko's |
|
4094 | 4098 | comment on 'macros'). |
|
4095 | 4099 | |
|
4096 | 4100 | * Small fixes and doc updates. |
|
4097 | 4101 | |
|
4098 | 4102 | * Rewrote @history function (was @h). Renamed it to @hist, @h is |
|
4099 | 4103 | much too fragile with automagic. Handles properly multi-line |
|
4100 | 4104 | statements and takes parameters. |
|
4101 | 4105 | |
|
4102 | 4106 | 2001-11-30 Fernando Perez <fperez@colorado.edu> |
|
4103 | 4107 | |
|
4104 | 4108 | * Version 0.1.18 released. |
|
4105 | 4109 | |
|
4106 | 4110 | * Fixed nasty namespace bug in initial module imports. |
|
4107 | 4111 | |
|
4108 | 4112 | * Added copyright/license notes to all code files (except |
|
4109 | 4113 | DPyGetOpt). For the time being, LGPL. That could change. |
|
4110 | 4114 | |
|
4111 | 4115 | * Rewrote a much nicer README, updated INSTALL, cleaned up |
|
4112 | 4116 | ipythonrc-* samples. |
|
4113 | 4117 | |
|
4114 | 4118 | * Overall code/documentation cleanup. Basically ready for |
|
4115 | 4119 | release. Only remaining thing: licence decision (LGPL?). |
|
4116 | 4120 | |
|
4117 | 4121 | * Converted load_config to a class, ConfigLoader. Now recursion |
|
4118 | 4122 | control is better organized. Doesn't include the same file twice. |
|
4119 | 4123 | |
|
4120 | 4124 | 2001-11-29 Fernando Perez <fperez@colorado.edu> |
|
4121 | 4125 | |
|
4122 | 4126 | * Got input history working. Changed output history variables from |
|
4123 | 4127 | _p to _o so that _i is for input and _o for output. Just cleaner |
|
4124 | 4128 | convention. |
|
4125 | 4129 | |
|
4126 | 4130 | * Implemented parametric aliases. This pretty much allows the |
|
4127 | 4131 | alias system to offer full-blown shell convenience, I think. |
|
4128 | 4132 | |
|
4129 | 4133 | * Version 0.1.17 released, 0.1.18 opened. |
|
4130 | 4134 | |
|
4131 | 4135 | * dot_ipython/ipythonrc (alias): added documentation. |
|
4132 | 4136 | (xcolor): Fixed small bug (xcolors -> xcolor) |
|
4133 | 4137 | |
|
4134 | 4138 | * Changed the alias system. Now alias is a magic command to define |
|
4135 | 4139 | aliases just like the shell. Rationale: the builtin magics should |
|
4136 | 4140 | be there for things deeply connected to IPython's |
|
4137 | 4141 | architecture. And this is a much lighter system for what I think |
|
4138 | 4142 | is the really important feature: allowing users to define quickly |
|
4139 | 4143 | magics that will do shell things for them, so they can customize |
|
4140 | 4144 | IPython easily to match their work habits. If someone is really |
|
4141 | 4145 | desperate to have another name for a builtin alias, they can |
|
4142 | 4146 | always use __IP.magic_newname = __IP.magic_oldname. Hackish but |
|
4143 | 4147 | works. |
|
4144 | 4148 | |
|
4145 | 4149 | 2001-11-28 Fernando Perez <fperez@colorado.edu> |
|
4146 | 4150 | |
|
4147 | 4151 | * Changed @file so that it opens the source file at the proper |
|
4148 | 4152 | line. Since it uses less, if your EDITOR environment is |
|
4149 | 4153 | configured, typing v will immediately open your editor of choice |
|
4150 | 4154 | right at the line where the object is defined. Not as quick as |
|
4151 | 4155 | having a direct @edit command, but for all intents and purposes it |
|
4152 | 4156 | works. And I don't have to worry about writing @edit to deal with |
|
4153 | 4157 | all the editors, less does that. |
|
4154 | 4158 | |
|
4155 | 4159 | * Version 0.1.16 released, 0.1.17 opened. |
|
4156 | 4160 | |
|
4157 | 4161 | * Fixed some nasty bugs in the page/page_dumb combo that could |
|
4158 | 4162 | crash IPython. |
|
4159 | 4163 | |
|
4160 | 4164 | 2001-11-27 Fernando Perez <fperez@colorado.edu> |
|
4161 | 4165 | |
|
4162 | 4166 | * Version 0.1.15 released, 0.1.16 opened. |
|
4163 | 4167 | |
|
4164 | 4168 | * Finally got ? and ?? to work for undefined things: now it's |
|
4165 | 4169 | possible to type {}.get? and get information about the get method |
|
4166 | 4170 | of dicts, or os.path? even if only os is defined (so technically |
|
4167 | 4171 | os.path isn't). Works at any level. For example, after import os, |
|
4168 | 4172 | os?, os.path?, os.path.abspath? all work. This is great, took some |
|
4169 | 4173 | work in _ofind. |
|
4170 | 4174 | |
|
4171 | 4175 | * Fixed more bugs with logging. The sanest way to do it was to add |
|
4172 | 4176 | to @log a 'mode' parameter. Killed two in one shot (this mode |
|
4173 | 4177 | option was a request of Janko's). I think it's finally clean |
|
4174 | 4178 | (famous last words). |
|
4175 | 4179 | |
|
4176 | 4180 | * Added a page_dumb() pager which does a decent job of paging on |
|
4177 | 4181 | screen, if better things (like less) aren't available. One less |
|
4178 | 4182 | unix dependency (someday maybe somebody will port this to |
|
4179 | 4183 | windows). |
|
4180 | 4184 | |
|
4181 | 4185 | * Fixed problem in magic_log: would lock of logging out if log |
|
4182 | 4186 | creation failed (because it would still think it had succeeded). |
|
4183 | 4187 | |
|
4184 | 4188 | * Improved the page() function using curses to auto-detect screen |
|
4185 | 4189 | size. Now it can make a much better decision on whether to print |
|
4186 | 4190 | or page a string. Option screen_length was modified: a value 0 |
|
4187 | 4191 | means auto-detect, and that's the default now. |
|
4188 | 4192 | |
|
4189 | 4193 | * Version 0.1.14 released, 0.1.15 opened. I think this is ready to |
|
4190 | 4194 | go out. I'll test it for a few days, then talk to Janko about |
|
4191 | 4195 | licences and announce it. |
|
4192 | 4196 | |
|
4193 | 4197 | * Fixed the length of the auto-generated ---> prompt which appears |
|
4194 | 4198 | for auto-parens and auto-quotes. Getting this right isn't trivial, |
|
4195 | 4199 | with all the color escapes, different prompt types and optional |
|
4196 | 4200 | separators. But it seems to be working in all the combinations. |
|
4197 | 4201 | |
|
4198 | 4202 | 2001-11-26 Fernando Perez <fperez@colorado.edu> |
|
4199 | 4203 | |
|
4200 | 4204 | * Wrote a regexp filter to get option types from the option names |
|
4201 | 4205 | string. This eliminates the need to manually keep two duplicate |
|
4202 | 4206 | lists. |
|
4203 | 4207 | |
|
4204 | 4208 | * Removed the unneeded check_option_names. Now options are handled |
|
4205 | 4209 | in a much saner manner and it's easy to visually check that things |
|
4206 | 4210 | are ok. |
|
4207 | 4211 | |
|
4208 | 4212 | * Updated version numbers on all files I modified to carry a |
|
4209 | 4213 | notice so Janko and Nathan have clear version markers. |
|
4210 | 4214 | |
|
4211 | 4215 | * Updated docstring for ultraTB with my changes. I should send |
|
4212 | 4216 | this to Nathan. |
|
4213 | 4217 | |
|
4214 | 4218 | * Lots of small fixes. Ran everything through pychecker again. |
|
4215 | 4219 | |
|
4216 | 4220 | * Made loading of deep_reload an cmd line option. If it's not too |
|
4217 | 4221 | kosher, now people can just disable it. With -nodeep_reload it's |
|
4218 | 4222 | still available as dreload(), it just won't overwrite reload(). |
|
4219 | 4223 | |
|
4220 | 4224 | * Moved many options to the no| form (-opt and -noopt |
|
4221 | 4225 | accepted). Cleaner. |
|
4222 | 4226 | |
|
4223 | 4227 | * Changed magic_log so that if called with no parameters, it uses |
|
4224 | 4228 | 'rotate' mode. That way auto-generated logs aren't automatically |
|
4225 | 4229 | over-written. For normal logs, now a backup is made if it exists |
|
4226 | 4230 | (only 1 level of backups). A new 'backup' mode was added to the |
|
4227 | 4231 | Logger class to support this. This was a request by Janko. |
|
4228 | 4232 | |
|
4229 | 4233 | * Added @logoff/@logon to stop/restart an active log. |
|
4230 | 4234 | |
|
4231 | 4235 | * Fixed a lot of bugs in log saving/replay. It was pretty |
|
4232 | 4236 | broken. Now special lines (!@,/) appear properly in the command |
|
4233 | 4237 | history after a log replay. |
|
4234 | 4238 | |
|
4235 | 4239 | * Tried and failed to implement full session saving via pickle. My |
|
4236 | 4240 | idea was to pickle __main__.__dict__, but modules can't be |
|
4237 | 4241 | pickled. This would be a better alternative to replaying logs, but |
|
4238 | 4242 | seems quite tricky to get to work. Changed -session to be called |
|
4239 | 4243 | -logplay, which more accurately reflects what it does. And if we |
|
4240 | 4244 | ever get real session saving working, -session is now available. |
|
4241 | 4245 | |
|
4242 | 4246 | * Implemented color schemes for prompts also. As for tracebacks, |
|
4243 | 4247 | currently only NoColor and Linux are supported. But now the |
|
4244 | 4248 | infrastructure is in place, based on a generic ColorScheme |
|
4245 | 4249 | class. So writing and activating new schemes both for the prompts |
|
4246 | 4250 | and the tracebacks should be straightforward. |
|
4247 | 4251 | |
|
4248 | 4252 | * Version 0.1.13 released, 0.1.14 opened. |
|
4249 | 4253 | |
|
4250 | 4254 | * Changed handling of options for output cache. Now counter is |
|
4251 | 4255 | hardwired starting at 1 and one specifies the maximum number of |
|
4252 | 4256 | entries *in the outcache* (not the max prompt counter). This is |
|
4253 | 4257 | much better, since many statements won't increase the cache |
|
4254 | 4258 | count. It also eliminated some confusing options, now there's only |
|
4255 | 4259 | one: cache_size. |
|
4256 | 4260 | |
|
4257 | 4261 | * Added 'alias' magic function and magic_alias option in the |
|
4258 | 4262 | ipythonrc file. Now the user can easily define whatever names he |
|
4259 | 4263 | wants for the magic functions without having to play weird |
|
4260 | 4264 | namespace games. This gives IPython a real shell-like feel. |
|
4261 | 4265 | |
|
4262 | 4266 | * Fixed doc/?/?? for magics. Now all work, in all forms (explicit |
|
4263 | 4267 | @ or not). |
|
4264 | 4268 | |
|
4265 | 4269 | This was one of the last remaining 'visible' bugs (that I know |
|
4266 | 4270 | of). I think if I can clean up the session loading so it works |
|
4267 | 4271 | 100% I'll release a 0.2.0 version on c.p.l (talk to Janko first |
|
4268 | 4272 | about licensing). |
|
4269 | 4273 | |
|
4270 | 4274 | 2001-11-25 Fernando Perez <fperez@colorado.edu> |
|
4271 | 4275 | |
|
4272 | 4276 | * Rewrote somewhat oinfo (?/??). Nicer, now uses page() and |
|
4273 | 4277 | there's a cleaner distinction between what ? and ?? show. |
|
4274 | 4278 | |
|
4275 | 4279 | * Added screen_length option. Now the user can define his own |
|
4276 | 4280 | screen size for page() operations. |
|
4277 | 4281 | |
|
4278 | 4282 | * Implemented magic shell-like functions with automatic code |
|
4279 | 4283 | generation. Now adding another function is just a matter of adding |
|
4280 | 4284 | an entry to a dict, and the function is dynamically generated at |
|
4281 | 4285 | run-time. Python has some really cool features! |
|
4282 | 4286 | |
|
4283 | 4287 | * Renamed many options to cleanup conventions a little. Now all |
|
4284 | 4288 | are lowercase, and only underscores where needed. Also in the code |
|
4285 | 4289 | option name tables are clearer. |
|
4286 | 4290 | |
|
4287 | 4291 | * Changed prompts a little. Now input is 'In [n]:' instead of |
|
4288 | 4292 | 'In[n]:='. This allows it the numbers to be aligned with the |
|
4289 | 4293 | Out[n] numbers, and removes usage of ':=' which doesn't exist in |
|
4290 | 4294 | Python (it was a Mathematica thing). The '...' continuation prompt |
|
4291 | 4295 | was also changed a little to align better. |
|
4292 | 4296 | |
|
4293 | 4297 | * Fixed bug when flushing output cache. Not all _p<n> variables |
|
4294 | 4298 | exist, so their deletion needs to be wrapped in a try: |
|
4295 | 4299 | |
|
4296 | 4300 | * Figured out how to properly use inspect.formatargspec() (it |
|
4297 | 4301 | requires the args preceded by *). So I removed all the code from |
|
4298 | 4302 | _get_pdef in Magic, which was just replicating that. |
|
4299 | 4303 | |
|
4300 | 4304 | * Added test to prefilter to allow redefining magic function names |
|
4301 | 4305 | as variables. This is ok, since the @ form is always available, |
|
4302 | 4306 | but whe should allow the user to define a variable called 'ls' if |
|
4303 | 4307 | he needs it. |
|
4304 | 4308 | |
|
4305 | 4309 | * Moved the ToDo information from README into a separate ToDo. |
|
4306 | 4310 | |
|
4307 | 4311 | * General code cleanup and small bugfixes. I think it's close to a |
|
4308 | 4312 | state where it can be released, obviously with a big 'beta' |
|
4309 | 4313 | warning on it. |
|
4310 | 4314 | |
|
4311 | 4315 | * Got the magic function split to work. Now all magics are defined |
|
4312 | 4316 | in a separate class. It just organizes things a bit, and now |
|
4313 | 4317 | Xemacs behaves nicer (it was choking on InteractiveShell b/c it |
|
4314 | 4318 | was too long). |
|
4315 | 4319 | |
|
4316 | 4320 | * Changed @clear to @reset to avoid potential confusions with |
|
4317 | 4321 | the shell command clear. Also renamed @cl to @clear, which does |
|
4318 | 4322 | exactly what people expect it to from their shell experience. |
|
4319 | 4323 | |
|
4320 | 4324 | Added a check to the @reset command (since it's so |
|
4321 | 4325 | destructive, it's probably a good idea to ask for confirmation). |
|
4322 | 4326 | But now reset only works for full namespace resetting. Since the |
|
4323 | 4327 | del keyword is already there for deleting a few specific |
|
4324 | 4328 | variables, I don't see the point of having a redundant magic |
|
4325 | 4329 | function for the same task. |
|
4326 | 4330 | |
|
4327 | 4331 | 2001-11-24 Fernando Perez <fperez@colorado.edu> |
|
4328 | 4332 | |
|
4329 | 4333 | * Updated the builtin docs (esp. the ? ones). |
|
4330 | 4334 | |
|
4331 | 4335 | * Ran all the code through pychecker. Not terribly impressed with |
|
4332 | 4336 | it: lots of spurious warnings and didn't really find anything of |
|
4333 | 4337 | substance (just a few modules being imported and not used). |
|
4334 | 4338 | |
|
4335 | 4339 | * Implemented the new ultraTB functionality into IPython. New |
|
4336 | 4340 | option: xcolors. This chooses color scheme. xmode now only selects |
|
4337 | 4341 | between Plain and Verbose. Better orthogonality. |
|
4338 | 4342 | |
|
4339 | 4343 | * Large rewrite of ultraTB. Much cleaner now, with a separation of |
|
4340 | 4344 | mode and color scheme for the exception handlers. Now it's |
|
4341 | 4345 | possible to have the verbose traceback with no coloring. |
|
4342 | 4346 | |
|
4343 | 4347 | 2001-11-23 Fernando Perez <fperez@colorado.edu> |
|
4344 | 4348 | |
|
4345 | 4349 | * Version 0.1.12 released, 0.1.13 opened. |
|
4346 | 4350 | |
|
4347 | 4351 | * Removed option to set auto-quote and auto-paren escapes by |
|
4348 | 4352 | user. The chances of breaking valid syntax are just too high. If |
|
4349 | 4353 | someone *really* wants, they can always dig into the code. |
|
4350 | 4354 | |
|
4351 | 4355 | * Made prompt separators configurable. |
|
4352 | 4356 | |
|
4353 | 4357 | 2001-11-22 Fernando Perez <fperez@colorado.edu> |
|
4354 | 4358 | |
|
4355 | 4359 | * Small bugfixes in many places. |
|
4356 | 4360 | |
|
4357 | 4361 | * Removed the MyCompleter class from ipplib. It seemed redundant |
|
4358 | 4362 | with the C-p,C-n history search functionality. Less code to |
|
4359 | 4363 | maintain. |
|
4360 | 4364 | |
|
4361 | 4365 | * Moved all the original ipython.py code into ipythonlib.py. Right |
|
4362 | 4366 | now it's just one big dump into a function called make_IPython, so |
|
4363 | 4367 | no real modularity has been gained. But at least it makes the |
|
4364 | 4368 | wrapper script tiny, and since ipythonlib is a module, it gets |
|
4365 | 4369 | compiled and startup is much faster. |
|
4366 | 4370 | |
|
4367 | 4371 | This is a reasobably 'deep' change, so we should test it for a |
|
4368 | 4372 | while without messing too much more with the code. |
|
4369 | 4373 | |
|
4370 | 4374 | 2001-11-21 Fernando Perez <fperez@colorado.edu> |
|
4371 | 4375 | |
|
4372 | 4376 | * Version 0.1.11 released, 0.1.12 opened for further work. |
|
4373 | 4377 | |
|
4374 | 4378 | * Removed dependency on Itpl. It was only needed in one place. It |
|
4375 | 4379 | would be nice if this became part of python, though. It makes life |
|
4376 | 4380 | *a lot* easier in some cases. |
|
4377 | 4381 | |
|
4378 | 4382 | * Simplified the prefilter code a bit. Now all handlers are |
|
4379 | 4383 | expected to explicitly return a value (at least a blank string). |
|
4380 | 4384 | |
|
4381 | 4385 | * Heavy edits in ipplib. Removed the help system altogether. Now |
|
4382 | 4386 | obj?/?? is used for inspecting objects, a magic @doc prints |
|
4383 | 4387 | docstrings, and full-blown Python help is accessed via the 'help' |
|
4384 | 4388 | keyword. This cleans up a lot of code (less to maintain) and does |
|
4385 | 4389 | the job. Since 'help' is now a standard Python component, might as |
|
4386 | 4390 | well use it and remove duplicate functionality. |
|
4387 | 4391 | |
|
4388 | 4392 | Also removed the option to use ipplib as a standalone program. By |
|
4389 | 4393 | now it's too dependent on other parts of IPython to function alone. |
|
4390 | 4394 | |
|
4391 | 4395 | * Fixed bug in genutils.pager. It would crash if the pager was |
|
4392 | 4396 | exited immediately after opening (broken pipe). |
|
4393 | 4397 | |
|
4394 | 4398 | * Trimmed down the VerboseTB reporting a little. The header is |
|
4395 | 4399 | much shorter now and the repeated exception arguments at the end |
|
4396 | 4400 | have been removed. For interactive use the old header seemed a bit |
|
4397 | 4401 | excessive. |
|
4398 | 4402 | |
|
4399 | 4403 | * Fixed small bug in output of @whos for variables with multi-word |
|
4400 | 4404 | types (only first word was displayed). |
|
4401 | 4405 | |
|
4402 | 4406 | 2001-11-17 Fernando Perez <fperez@colorado.edu> |
|
4403 | 4407 | |
|
4404 | 4408 | * Version 0.1.10 released, 0.1.11 opened for further work. |
|
4405 | 4409 | |
|
4406 | 4410 | * Modified dirs and friends. dirs now *returns* the stack (not |
|
4407 | 4411 | prints), so one can manipulate it as a variable. Convenient to |
|
4408 | 4412 | travel along many directories. |
|
4409 | 4413 | |
|
4410 | 4414 | * Fixed bug in magic_pdef: would only work with functions with |
|
4411 | 4415 | arguments with default values. |
|
4412 | 4416 | |
|
4413 | 4417 | 2001-11-14 Fernando Perez <fperez@colorado.edu> |
|
4414 | 4418 | |
|
4415 | 4419 | * Added the PhysicsInput stuff to dot_ipython so it ships as an |
|
4416 | 4420 | example with IPython. Various other minor fixes and cleanups. |
|
4417 | 4421 | |
|
4418 | 4422 | * Version 0.1.9 released, 0.1.10 opened for further work. |
|
4419 | 4423 | |
|
4420 | 4424 | * Added sys.path to the list of directories searched in the |
|
4421 | 4425 | execfile= option. It used to be the current directory and the |
|
4422 | 4426 | user's IPYTHONDIR only. |
|
4423 | 4427 | |
|
4424 | 4428 | 2001-11-13 Fernando Perez <fperez@colorado.edu> |
|
4425 | 4429 | |
|
4426 | 4430 | * Reinstated the raw_input/prefilter separation that Janko had |
|
4427 | 4431 | initially. This gives a more convenient setup for extending the |
|
4428 | 4432 | pre-processor from the outside: raw_input always gets a string, |
|
4429 | 4433 | and prefilter has to process it. We can then redefine prefilter |
|
4430 | 4434 | from the outside and implement extensions for special |
|
4431 | 4435 | purposes. |
|
4432 | 4436 | |
|
4433 | 4437 | Today I got one for inputting PhysicalQuantity objects |
|
4434 | 4438 | (from Scientific) without needing any function calls at |
|
4435 | 4439 | all. Extremely convenient, and it's all done as a user-level |
|
4436 | 4440 | extension (no IPython code was touched). Now instead of: |
|
4437 | 4441 | a = PhysicalQuantity(4.2,'m/s**2') |
|
4438 | 4442 | one can simply say |
|
4439 | 4443 | a = 4.2 m/s**2 |
|
4440 | 4444 | or even |
|
4441 | 4445 | a = 4.2 m/s^2 |
|
4442 | 4446 | |
|
4443 | 4447 | I use this, but it's also a proof of concept: IPython really is |
|
4444 | 4448 | fully user-extensible, even at the level of the parsing of the |
|
4445 | 4449 | command line. It's not trivial, but it's perfectly doable. |
|
4446 | 4450 | |
|
4447 | 4451 | * Added 'add_flip' method to inclusion conflict resolver. Fixes |
|
4448 | 4452 | the problem of modules being loaded in the inverse order in which |
|
4449 | 4453 | they were defined in |
|
4450 | 4454 | |
|
4451 | 4455 | * Version 0.1.8 released, 0.1.9 opened for further work. |
|
4452 | 4456 | |
|
4453 | 4457 | * Added magics pdef, source and file. They respectively show the |
|
4454 | 4458 | definition line ('prototype' in C), source code and full python |
|
4455 | 4459 | file for any callable object. The object inspector oinfo uses |
|
4456 | 4460 | these to show the same information. |
|
4457 | 4461 | |
|
4458 | 4462 | * Version 0.1.7 released, 0.1.8 opened for further work. |
|
4459 | 4463 | |
|
4460 | 4464 | * Separated all the magic functions into a class called Magic. The |
|
4461 | 4465 | InteractiveShell class was becoming too big for Xemacs to handle |
|
4462 | 4466 | (de-indenting a line would lock it up for 10 seconds while it |
|
4463 | 4467 | backtracked on the whole class!) |
|
4464 | 4468 | |
|
4465 | 4469 | FIXME: didn't work. It can be done, but right now namespaces are |
|
4466 | 4470 | all messed up. Do it later (reverted it for now, so at least |
|
4467 | 4471 | everything works as before). |
|
4468 | 4472 | |
|
4469 | 4473 | * Got the object introspection system (magic_oinfo) working! I |
|
4470 | 4474 | think this is pretty much ready for release to Janko, so he can |
|
4471 | 4475 | test it for a while and then announce it. Pretty much 100% of what |
|
4472 | 4476 | I wanted for the 'phase 1' release is ready. Happy, tired. |
|
4473 | 4477 | |
|
4474 | 4478 | 2001-11-12 Fernando Perez <fperez@colorado.edu> |
|
4475 | 4479 | |
|
4476 | 4480 | * Version 0.1.6 released, 0.1.7 opened for further work. |
|
4477 | 4481 | |
|
4478 | 4482 | * Fixed bug in printing: it used to test for truth before |
|
4479 | 4483 | printing, so 0 wouldn't print. Now checks for None. |
|
4480 | 4484 | |
|
4481 | 4485 | * Fixed bug where auto-execs increase the prompt counter by 2 (b/c |
|
4482 | 4486 | they have to call len(str(sys.ps1)) ). But the fix is ugly, it |
|
4483 | 4487 | reaches by hand into the outputcache. Think of a better way to do |
|
4484 | 4488 | this later. |
|
4485 | 4489 | |
|
4486 | 4490 | * Various small fixes thanks to Nathan's comments. |
|
4487 | 4491 | |
|
4488 | 4492 | * Changed magic_pprint to magic_Pprint. This way it doesn't |
|
4489 | 4493 | collide with pprint() and the name is consistent with the command |
|
4490 | 4494 | line option. |
|
4491 | 4495 | |
|
4492 | 4496 | * Changed prompt counter behavior to be fully like |
|
4493 | 4497 | Mathematica's. That is, even input that doesn't return a result |
|
4494 | 4498 | raises the prompt counter. The old behavior was kind of confusing |
|
4495 | 4499 | (getting the same prompt number several times if the operation |
|
4496 | 4500 | didn't return a result). |
|
4497 | 4501 | |
|
4498 | 4502 | * Fixed Nathan's last name in a couple of places (Gray, not Graham). |
|
4499 | 4503 | |
|
4500 | 4504 | * Fixed -Classic mode (wasn't working anymore). |
|
4501 | 4505 | |
|
4502 | 4506 | * Added colored prompts using Nathan's new code. Colors are |
|
4503 | 4507 | currently hardwired, they can be user-configurable. For |
|
4504 | 4508 | developers, they can be chosen in file ipythonlib.py, at the |
|
4505 | 4509 | beginning of the CachedOutput class def. |
|
4506 | 4510 | |
|
4507 | 4511 | 2001-11-11 Fernando Perez <fperez@colorado.edu> |
|
4508 | 4512 | |
|
4509 | 4513 | * Version 0.1.5 released, 0.1.6 opened for further work. |
|
4510 | 4514 | |
|
4511 | 4515 | * Changed magic_env to *return* the environment as a dict (not to |
|
4512 | 4516 | print it). This way it prints, but it can also be processed. |
|
4513 | 4517 | |
|
4514 | 4518 | * Added Verbose exception reporting to interactive |
|
4515 | 4519 | exceptions. Very nice, now even 1/0 at the prompt gives a verbose |
|
4516 | 4520 | traceback. Had to make some changes to the ultraTB file. This is |
|
4517 | 4521 | probably the last 'big' thing in my mental todo list. This ties |
|
4518 | 4522 | in with the next entry: |
|
4519 | 4523 | |
|
4520 | 4524 | * Changed -Xi and -Xf to a single -xmode option. Now all the user |
|
4521 | 4525 | has to specify is Plain, Color or Verbose for all exception |
|
4522 | 4526 | handling. |
|
4523 | 4527 | |
|
4524 | 4528 | * Removed ShellServices option. All this can really be done via |
|
4525 | 4529 | the magic system. It's easier to extend, cleaner and has automatic |
|
4526 | 4530 | namespace protection and documentation. |
|
4527 | 4531 | |
|
4528 | 4532 | 2001-11-09 Fernando Perez <fperez@colorado.edu> |
|
4529 | 4533 | |
|
4530 | 4534 | * Fixed bug in output cache flushing (missing parameter to |
|
4531 | 4535 | __init__). Other small bugs fixed (found using pychecker). |
|
4532 | 4536 | |
|
4533 | 4537 | * Version 0.1.4 opened for bugfixing. |
|
4534 | 4538 | |
|
4535 | 4539 | 2001-11-07 Fernando Perez <fperez@colorado.edu> |
|
4536 | 4540 | |
|
4537 | 4541 | * Version 0.1.3 released, mainly because of the raw_input bug. |
|
4538 | 4542 | |
|
4539 | 4543 | * Fixed NASTY bug in raw_input: input line wasn't properly parsed |
|
4540 | 4544 | and when testing for whether things were callable, a call could |
|
4541 | 4545 | actually be made to certain functions. They would get called again |
|
4542 | 4546 | once 'really' executed, with a resulting double call. A disaster |
|
4543 | 4547 | in many cases (list.reverse() would never work!). |
|
4544 | 4548 | |
|
4545 | 4549 | * Removed prefilter() function, moved its code to raw_input (which |
|
4546 | 4550 | after all was just a near-empty caller for prefilter). This saves |
|
4547 | 4551 | a function call on every prompt, and simplifies the class a tiny bit. |
|
4548 | 4552 | |
|
4549 | 4553 | * Fix _ip to __ip name in magic example file. |
|
4550 | 4554 | |
|
4551 | 4555 | * Changed 'tar -x -f' to 'tar xvf' in auto-installer. This should |
|
4552 | 4556 | work with non-gnu versions of tar. |
|
4553 | 4557 | |
|
4554 | 4558 | 2001-11-06 Fernando Perez <fperez@colorado.edu> |
|
4555 | 4559 | |
|
4556 | 4560 | * Version 0.1.2. Just to keep track of the recent changes. |
|
4557 | 4561 | |
|
4558 | 4562 | * Fixed nasty bug in output prompt routine. It used to check 'if |
|
4559 | 4563 | arg != None...'. Problem is, this fails if arg implements a |
|
4560 | 4564 | special comparison (__cmp__) which disallows comparing to |
|
4561 | 4565 | None. Found it when trying to use the PhysicalQuantity module from |
|
4562 | 4566 | ScientificPython. |
|
4563 | 4567 | |
|
4564 | 4568 | 2001-11-05 Fernando Perez <fperez@colorado.edu> |
|
4565 | 4569 | |
|
4566 | 4570 | * Also added dirs. Now the pushd/popd/dirs family functions |
|
4567 | 4571 | basically like the shell, with the added convenience of going home |
|
4568 | 4572 | when called with no args. |
|
4569 | 4573 | |
|
4570 | 4574 | * pushd/popd slightly modified to mimic shell behavior more |
|
4571 | 4575 | closely. |
|
4572 | 4576 | |
|
4573 | 4577 | * Added env,pushd,popd from ShellServices as magic functions. I |
|
4574 | 4578 | think the cleanest will be to port all desired functions from |
|
4575 | 4579 | ShellServices as magics and remove ShellServices altogether. This |
|
4576 | 4580 | will provide a single, clean way of adding functionality |
|
4577 | 4581 | (shell-type or otherwise) to IP. |
|
4578 | 4582 | |
|
4579 | 4583 | 2001-11-04 Fernando Perez <fperez@colorado.edu> |
|
4580 | 4584 | |
|
4581 | 4585 | * Added .ipython/ directory to sys.path. This way users can keep |
|
4582 | 4586 | customizations there and access them via import. |
|
4583 | 4587 | |
|
4584 | 4588 | 2001-11-03 Fernando Perez <fperez@colorado.edu> |
|
4585 | 4589 | |
|
4586 | 4590 | * Opened version 0.1.1 for new changes. |
|
4587 | 4591 | |
|
4588 | 4592 | * Changed version number to 0.1.0: first 'public' release, sent to |
|
4589 | 4593 | Nathan and Janko. |
|
4590 | 4594 | |
|
4591 | 4595 | * Lots of small fixes and tweaks. |
|
4592 | 4596 | |
|
4593 | 4597 | * Minor changes to whos format. Now strings are shown, snipped if |
|
4594 | 4598 | too long. |
|
4595 | 4599 | |
|
4596 | 4600 | * Changed ShellServices to work on __main__ so they show up in @who |
|
4597 | 4601 | |
|
4598 | 4602 | * Help also works with ? at the end of a line: |
|
4599 | 4603 | ?sin and sin? |
|
4600 | 4604 | both produce the same effect. This is nice, as often I use the |
|
4601 | 4605 | tab-complete to find the name of a method, but I used to then have |
|
4602 | 4606 | to go to the beginning of the line to put a ? if I wanted more |
|
4603 | 4607 | info. Now I can just add the ? and hit return. Convenient. |
|
4604 | 4608 | |
|
4605 | 4609 | 2001-11-02 Fernando Perez <fperez@colorado.edu> |
|
4606 | 4610 | |
|
4607 | 4611 | * Python version check (>=2.1) added. |
|
4608 | 4612 | |
|
4609 | 4613 | * Added LazyPython documentation. At this point the docs are quite |
|
4610 | 4614 | a mess. A cleanup is in order. |
|
4611 | 4615 | |
|
4612 | 4616 | * Auto-installer created. For some bizarre reason, the zipfiles |
|
4613 | 4617 | module isn't working on my system. So I made a tar version |
|
4614 | 4618 | (hopefully the command line options in various systems won't kill |
|
4615 | 4619 | me). |
|
4616 | 4620 | |
|
4617 | 4621 | * Fixes to Struct in genutils. Now all dictionary-like methods are |
|
4618 | 4622 | protected (reasonably). |
|
4619 | 4623 | |
|
4620 | 4624 | * Added pager function to genutils and changed ? to print usage |
|
4621 | 4625 | note through it (it was too long). |
|
4622 | 4626 | |
|
4623 | 4627 | * Added the LazyPython functionality. Works great! I changed the |
|
4624 | 4628 | auto-quote escape to ';', it's on home row and next to '. But |
|
4625 | 4629 | both auto-quote and auto-paren (still /) escapes are command-line |
|
4626 | 4630 | parameters. |
|
4627 | 4631 | |
|
4628 | 4632 | |
|
4629 | 4633 | 2001-11-01 Fernando Perez <fperez@colorado.edu> |
|
4630 | 4634 | |
|
4631 | 4635 | * Version changed to 0.0.7. Fairly large change: configuration now |
|
4632 | 4636 | is all stored in a directory, by default .ipython. There, all |
|
4633 | 4637 | config files have normal looking names (not .names) |
|
4634 | 4638 | |
|
4635 | 4639 | * Version 0.0.6 Released first to Lucas and Archie as a test |
|
4636 | 4640 | run. Since it's the first 'semi-public' release, change version to |
|
4637 | 4641 | > 0.0.6 for any changes now. |
|
4638 | 4642 | |
|
4639 | 4643 | * Stuff I had put in the ipplib.py changelog: |
|
4640 | 4644 | |
|
4641 | 4645 | Changes to InteractiveShell: |
|
4642 | 4646 | |
|
4643 | 4647 | - Made the usage message a parameter. |
|
4644 | 4648 | |
|
4645 | 4649 | - Require the name of the shell variable to be given. It's a bit |
|
4646 | 4650 | of a hack, but allows the name 'shell' not to be hardwire in the |
|
4647 | 4651 | magic (@) handler, which is problematic b/c it requires |
|
4648 | 4652 | polluting the global namespace with 'shell'. This in turn is |
|
4649 | 4653 | fragile: if a user redefines a variable called shell, things |
|
4650 | 4654 | break. |
|
4651 | 4655 | |
|
4652 | 4656 | - magic @: all functions available through @ need to be defined |
|
4653 | 4657 | as magic_<name>, even though they can be called simply as |
|
4654 | 4658 | @<name>. This allows the special command @magic to gather |
|
4655 | 4659 | information automatically about all existing magic functions, |
|
4656 | 4660 | even if they are run-time user extensions, by parsing the shell |
|
4657 | 4661 | instance __dict__ looking for special magic_ names. |
|
4658 | 4662 | |
|
4659 | 4663 | - mainloop: added *two* local namespace parameters. This allows |
|
4660 | 4664 | the class to differentiate between parameters which were there |
|
4661 | 4665 | before and after command line initialization was processed. This |
|
4662 | 4666 | way, later @who can show things loaded at startup by the |
|
4663 | 4667 | user. This trick was necessary to make session saving/reloading |
|
4664 | 4668 | really work: ideally after saving/exiting/reloading a session, |
|
4665 | 4669 | *everythin* should look the same, including the output of @who. I |
|
4666 | 4670 | was only able to make this work with this double namespace |
|
4667 | 4671 | trick. |
|
4668 | 4672 | |
|
4669 | 4673 | - added a header to the logfile which allows (almost) full |
|
4670 | 4674 | session restoring. |
|
4671 | 4675 | |
|
4672 | 4676 | - prepend lines beginning with @ or !, with a and log |
|
4673 | 4677 | them. Why? !lines: may be useful to know what you did @lines: |
|
4674 | 4678 | they may affect session state. So when restoring a session, at |
|
4675 | 4679 | least inform the user of their presence. I couldn't quite get |
|
4676 | 4680 | them to properly re-execute, but at least the user is warned. |
|
4677 | 4681 | |
|
4678 | 4682 | * Started ChangeLog. |
@@ -1,1345 +1,1358 b'' | |||
|
1 | 1 | #LyX 1.3 created this file. For more info see http://www.lyx.org/ |
|
2 | 2 | \lyxformat 221 |
|
3 | 3 | \textclass article |
|
4 | 4 | \begin_preamble |
|
5 | 5 | \usepackage{ae,aecompl} |
|
6 | 6 | \usepackage{hyperref} |
|
7 | 7 | \usepackage{html} |
|
8 | 8 | \end_preamble |
|
9 | 9 | \language english |
|
10 | 10 | \inputencoding auto |
|
11 | 11 | \fontscheme default |
|
12 | 12 | \graphics default |
|
13 | 13 | \paperfontsize default |
|
14 | 14 | \spacing single |
|
15 | 15 | \papersize Default |
|
16 | 16 | \paperpackage a4 |
|
17 | 17 | \use_geometry 1 |
|
18 | 18 | \use_amsmath 0 |
|
19 | 19 | \use_natbib 0 |
|
20 | 20 | \use_numerical_citations 0 |
|
21 | 21 | \paperorientation portrait |
|
22 | 22 | \leftmargin 1.25in |
|
23 | 23 | \topmargin 1in |
|
24 | 24 | \rightmargin 1.25in |
|
25 | 25 | \bottommargin 1in |
|
26 | 26 | \secnumdepth 3 |
|
27 | 27 | \tocdepth 3 |
|
28 | 28 | \paragraph_separation skip |
|
29 | 29 | \defskip medskip |
|
30 | 30 | \quotes_language english |
|
31 | 31 | \quotes_times 2 |
|
32 | 32 | \papercolumns 1 |
|
33 | 33 | \papersides 1 |
|
34 | 34 | \paperpagestyle default |
|
35 | 35 | |
|
36 | 36 | \layout Title |
|
37 | 37 | |
|
38 | 38 | IPython |
|
39 | 39 | \newline |
|
40 | 40 | |
|
41 | 41 | \size larger |
|
42 | 42 | New design notes |
|
43 | 43 | \layout Author |
|
44 | 44 | |
|
45 | 45 | Fernando PοΏ½rez |
|
46 | 46 | \layout Section |
|
47 | 47 | |
|
48 | 48 | Introduction |
|
49 | 49 | \layout Standard |
|
50 | 50 | |
|
51 | 51 | This is a draft document with notes and ideas for the IPython rewrite. |
|
52 | 52 | The section order and structure of this document roughly reflects in which |
|
53 | 53 | order things should be done and what the dependencies are. |
|
54 | 54 | This document is mainly a draft for developers, a pdf version is provided |
|
55 | 55 | with the standard distribution in case regular users are interested and |
|
56 | 56 | wish to contribute ideas. |
|
57 | 57 | \layout Standard |
|
58 | 58 | |
|
59 | 59 | A tentative plan for the future: |
|
60 | 60 | \layout Itemize |
|
61 | 61 | |
|
62 | 62 | 0.6.x series: in practice, enough people are using IPython for real work that |
|
63 | 63 | I think it warrants a higher number. |
|
64 | 64 | This series will continue to evolve with bugfixes and incremental improvements. |
|
65 | 65 | \layout Itemize |
|
66 | 66 | |
|
67 | 67 | 0.7.x series: (maybe) If resources allow, there may be a branch for 'unstable' |
|
68 | 68 | development, where the architectural rewrite may take place. |
|
69 | 69 | \layout Standard |
|
70 | 70 | |
|
71 | 71 | However, I am starting to doubt it is feasible to keep two separate branches. |
|
72 | 72 | I am leaning more towards a |
|
73 | 73 | \begin_inset ERT |
|
74 | 74 | status Collapsed |
|
75 | 75 | |
|
76 | 76 | \layout Standard |
|
77 | 77 | |
|
78 | 78 | \backslash |
|
79 | 79 | LyX |
|
80 | 80 | \end_inset |
|
81 | 81 | |
|
82 | 82 | -like approach, where the main branch slowly transforms and evolves. |
|
83 | 83 | Having CVS support now makes this a reasonable alternative, as I don't |
|
84 | 84 | have to make pre-releases as often. |
|
85 | 85 | The active branch can remain the mainline of development, and users interested |
|
86 | 86 | in the bleeding-edge stuff can always grab the CVS code. |
|
87 | 87 | \layout Standard |
|
88 | 88 | |
|
89 | 89 | Ideally, IPython should have a clean class setup that would allow further |
|
90 | 90 | extensions for special-purpose systems. |
|
91 | 91 | I view IPython as a base system that provides a great interactive environment |
|
92 | 92 | with full access to the Python language, and which could be used in many |
|
93 | 93 | different contexts. |
|
94 | 94 | The basic hooks are there: the magic extension syntax and the flexible |
|
95 | 95 | system of recursive configuration files and profiles. |
|
96 | 96 | But with a code as messy as the current one, nobody is going to touch it. |
|
97 | 97 | \layout Section |
|
98 | 98 | |
|
99 | 99 | Immediate TODO and bug list |
|
100 | 100 | \layout Standard |
|
101 | 101 | |
|
102 | 102 | Things that should be done for the current series, before starting major |
|
103 | 103 | changes. |
|
104 | 104 | \layout Itemize |
|
105 | 105 | |
|
106 | 106 | Fix any bugs reported at the online bug tracker. |
|
107 | 107 | \layout Itemize |
|
108 | 108 | |
|
109 | History bug: I often see that, under certain circumstances, the input history | |
|
110 | is incorrect. | |
|
111 | The problem is that so far, I've failed to find a simple way to reproduce | |
|
112 | it consistently, so I can't easily track it down. | |
|
113 | It seems to me that it happens when output is generated multiple times | |
|
114 | for the same input (for i in range(10): i will do it). | |
|
115 | But even this isn't reliable... | |
|
116 | Ultimately the right solution for this is to cleanly separate the dataflow | |
|
117 | for input/output history management; right now that happens all over the | |
|
118 | place, which makes the code impossible to debug, and almost guaranteed | |
|
119 | to be buggy in the first place. | |
|
120 | \layout Itemize | |
|
121 | ||
|
109 | 122 | |
|
110 | 123 | \series bold |
|
111 | 124 | Redesign the output traps. |
|
112 | 125 | |
|
113 | 126 | \series default |
|
114 | 127 | They cause problems when users try to execute code which relies on sys.stdout |
|
115 | 128 | being the 'true' sys.stdout. |
|
116 | 129 | They also prevent scripts which use raw_input() to work as command-line |
|
117 | 130 | arguments. |
|
118 | 131 | \newline |
|
119 | 132 | The best solution is probably to print the banner first, and then just execute |
|
120 | 133 | all the user code straight with no output traps at all. |
|
121 | 134 | Whatever comes out comes out. |
|
122 | 135 | This makes the ipython code actually simpler, and eliminates the problem |
|
123 | 136 | altogether. |
|
124 | 137 | \newline |
|
125 | 138 | These things need to be ripped out, they cause no end of problems. |
|
126 | 139 | For example, if user code requires acces to stdin during startup, the process |
|
127 | 140 | just hangs indefinitely. |
|
128 | 141 | For now I've just disabled them, and I'll live with the ugly error messages. |
|
129 | 142 | \layout Itemize |
|
130 | 143 | |
|
131 | 144 | The prompt specials dictionary should be turned into a class which does |
|
132 | 145 | proper namespace management, since the prompt specials need to be evaluated |
|
133 | 146 | in a certain namespace. |
|
134 | 147 | Currently it's just globals, which need to be managed manually by code |
|
135 | 148 | below. |
|
136 | 149 | |
|
137 | 150 | \layout Itemize |
|
138 | 151 | |
|
139 | 152 | Fix coloring of prompts: the pysh color strings don't have any effect on |
|
140 | 153 | prompt numbers, b/c these are controlled by the global scheme. |
|
141 | 154 | Make the prompts fully user-definable, colors and all. |
|
142 | 155 | This is what I said to a user: |
|
143 | 156 | \newline |
|
144 | 157 | As far as the green |
|
145 | 158 | \backslash |
|
146 | 159 | #, this is a minor bug of the coloring code due to the vagaries of history. |
|
147 | 160 | While the color strings allow you to control the coloring of most elements, |
|
148 | 161 | there are a few which are still controlled by the old ipython internal |
|
149 | 162 | coloring code, which only accepts a global 'color scheme' choice. |
|
150 | 163 | So basically the input/output numbers are hardwired to the choice in the |
|
151 | 164 | color scheme, and there are only 'Linux', 'LightBG' and 'NoColor' schemes |
|
152 | 165 | to choose from. |
|
153 | 166 | |
|
154 | 167 | \layout Itemize |
|
155 | 168 | |
|
156 | 169 | Clean up FakeModule issues. |
|
157 | 170 | Currently, unittesting with embedded ipython breaks because a FakeModule |
|
158 | 171 | instance overwrites __main__. |
|
159 | 172 | Maybe ipython should revert back to using __main__ directly as the user |
|
160 | 173 | namespace? Handling a separate namespace is proving |
|
161 | 174 | \emph on |
|
162 | 175 | very |
|
163 | 176 | \emph default |
|
164 | 177 | tricky in all corner cases. |
|
165 | 178 | \layout Itemize |
|
166 | 179 | |
|
167 | 180 | Make the output cache depth independent of the input one. |
|
168 | 181 | This way one can have say only the last 10 results stored and still have |
|
169 | 182 | a long input history/cache. |
|
170 | 183 | \layout Itemize |
|
171 | 184 | |
|
172 | 185 | Fix the fact that importing a shell for embedding screws up the command-line |
|
173 | 186 | history. |
|
174 | 187 | This can be done by not importing the history file when the shell is already |
|
175 | 188 | inside ipython. |
|
176 | 189 | \layout Itemize |
|
177 | 190 | |
|
178 | 191 | Lay out the class structure so that embedding into a gtk/wx/qt app is trivial, |
|
179 | 192 | much like the multithreaded gui shells now provide command-line coexistence |
|
180 | 193 | with the gui toolkits. |
|
181 | 194 | See |
|
182 | 195 | \begin_inset LatexCommand \url{http://www.livejournal.com/users/glyf/32396.html} |
|
183 | 196 | |
|
184 | 197 | \end_inset |
|
185 | 198 | |
|
186 | 199 | |
|
187 | 200 | \layout Itemize |
|
188 | 201 | |
|
189 | 202 | Get Holger's completer in, once he adds filename completion. |
|
190 | 203 | \layout Standard |
|
191 | 204 | |
|
192 | 205 | Lower priority stuff: |
|
193 | 206 | \layout Itemize |
|
194 | 207 | |
|
195 | 208 | Add @showopt/@setopt (decide name) for viewing/setting all options. |
|
196 | 209 | The existing option-setting magics should become aliases for setopt calls. |
|
197 | 210 | \layout Itemize |
|
198 | 211 | |
|
199 | 212 | It would be nice to be able to continue with python stuff after an @ command. |
|
200 | 213 | For instance "@run something; test_stuff()" in order to test stuff even |
|
201 | 214 | faster. |
|
202 | 215 | Suggestion by Kasper Souren <Kasper.Souren@ircam.fr> |
|
203 | 216 | \layout Itemize |
|
204 | 217 | |
|
205 | 218 | Run a 'first time wizard' which configures a few things for the user, such |
|
206 | 219 | as color_info, editor and the like. |
|
207 | 220 | \layout Itemize |
|
208 | 221 | |
|
209 | 222 | Logging: @logstart and -log should start logfiles in ~.ipython, but with |
|
210 | 223 | unique names in case of collisions. |
|
211 | 224 | This would prevent ipython.log files all over while also allowing multiple |
|
212 | 225 | sessions. |
|
213 | 226 | Also the -log option should take an optional filename, instead of having |
|
214 | 227 | a separate -logfile option. |
|
215 | 228 | \newline |
|
216 | 229 | In general the logging system needs a serious cleanup. |
|
217 | 230 | Many functions now in Magic should be moved to Logger, and the magic @s |
|
218 | 231 | should be very simple wrappers to the Logger methods. |
|
219 | 232 | \layout Section |
|
220 | 233 | |
|
221 | 234 | Lighten the code |
|
222 | 235 | \layout Standard |
|
223 | 236 | |
|
224 | 237 | If we decide to base future versions of IPython on Python 2.3, which has |
|
225 | 238 | the new Optik module (called optparse), it should be possible to drop DPyGetOpt. |
|
226 | 239 | We should also remove the need for Itpl. |
|
227 | 240 | Another area for trimming is the Gnuplot stuff: much of that could be merged |
|
228 | 241 | into the mainline project. |
|
229 | 242 | \layout Standard |
|
230 | 243 | |
|
231 | 244 | Double check whether we really need FlexCompleter. |
|
232 | 245 | This was written as an enhanced rlcompleter, but my patches went in for |
|
233 | 246 | python 2.2 (or 2.3, can't remember). |
|
234 | 247 | \layout Standard |
|
235 | 248 | |
|
236 | 249 | With these changes we could shed a fair bit of code from the main trunk. |
|
237 | 250 | \layout Section |
|
238 | 251 | |
|
239 | 252 | Unit testing |
|
240 | 253 | \layout Standard |
|
241 | 254 | |
|
242 | 255 | All new code should use a testing framework. |
|
243 | 256 | Python seems to have very good testing facilities, I just need to learn |
|
244 | 257 | how to use them. |
|
245 | 258 | I should also check out QMTest at |
|
246 | 259 | \begin_inset LatexCommand \htmlurl{http://www.codesourcery.com/qm/qmtest} |
|
247 | 260 | |
|
248 | 261 | \end_inset |
|
249 | 262 | |
|
250 | 263 | , it sounds interesting (it's Python-based too). |
|
251 | 264 | \layout Section |
|
252 | 265 | |
|
253 | 266 | Configuration system |
|
254 | 267 | \layout Standard |
|
255 | 268 | |
|
256 | 269 | Move away from the current ipythonrc format to using standard python files |
|
257 | 270 | for configuration. |
|
258 | 271 | This will require users to be slightly more careful in their syntax, but |
|
259 | 272 | reduces code in IPython, is more in line with Python's normal form (using |
|
260 | 273 | the $PYTHONSTARTUP file) and allows much more flexibility. |
|
261 | 274 | I also think it's more 'pythonic', in using a single language for everything. |
|
262 | 275 | \layout Standard |
|
263 | 276 | |
|
264 | 277 | Options can be set up with a function call which takes keywords and updates |
|
265 | 278 | the options Struct. |
|
266 | 279 | \layout Standard |
|
267 | 280 | |
|
268 | 281 | In order to maintain the recursive inclusion system, write an 'include' |
|
269 | 282 | function which is basically a wrapper around safe_execfile(). |
|
270 | 283 | Also for alias definitions an alias() function will do. |
|
271 | 284 | All functionality which we want to have at startup time for the users can |
|
272 | 285 | be wrapped in a small module so that config files look like: |
|
273 | 286 | \layout Standard |
|
274 | 287 | |
|
275 | 288 | |
|
276 | 289 | \family typewriter |
|
277 | 290 | from IPython.Startup import * |
|
278 | 291 | \newline |
|
279 | 292 | ... |
|
280 | 293 | \newline |
|
281 | 294 | set_options(automagic=1,colors='NoColor',...) |
|
282 | 295 | \newline |
|
283 | 296 | ... |
|
284 | 297 | \newline |
|
285 | 298 | include('mysetup.py') |
|
286 | 299 | \newline |
|
287 | 300 | ... |
|
288 | 301 | \newline |
|
289 | 302 | alias('ls ls --color -l') |
|
290 | 303 | \newline |
|
291 | 304 | ... |
|
292 | 305 | etc. |
|
293 | 306 | \layout Standard |
|
294 | 307 | |
|
295 | 308 | Also, put |
|
296 | 309 | \series bold |
|
297 | 310 | all |
|
298 | 311 | \series default |
|
299 | 312 | aliases in here, out of the core code. |
|
300 | 313 | \layout Standard |
|
301 | 314 | |
|
302 | 315 | The new system should allow for more seamless upgrading, so that: |
|
303 | 316 | \layout Itemize |
|
304 | 317 | |
|
305 | 318 | It automatically recognizes when the config files need updating and does |
|
306 | 319 | the upgrade. |
|
307 | 320 | \layout Itemize |
|
308 | 321 | |
|
309 | 322 | It simply adds the new options to the user's config file without overwriting |
|
310 | 323 | it. |
|
311 | 324 | The current system is annoying since users need to manually re-sync their |
|
312 | 325 | configuration after every update. |
|
313 | 326 | \layout Itemize |
|
314 | 327 | |
|
315 | 328 | It detects obsolete options and informs the user to remove them from his |
|
316 | 329 | config file. |
|
317 | 330 | \layout Standard |
|
318 | 331 | |
|
319 | 332 | Here's a copy of Arnd Baecker suggestions on the matter: |
|
320 | 333 | \layout Standard |
|
321 | 334 | |
|
322 | 335 | 1.) upgrade: it might be nice to have an "auto" upgrade procedure: i.e. |
|
323 | 336 | imagine that IPython is installed system-wide and gets upgraded, how does |
|
324 | 337 | a user know, that an upgrade of the stuff in ~/.ipython is necessary ? So |
|
325 | 338 | maybe one has to a keep a version number in ~/.ipython and if there is a |
|
326 | 339 | mismatch with the started ipython, then invoke the upgrade procedure. |
|
327 | 340 | \layout Standard |
|
328 | 341 | |
|
329 | 342 | 2.) upgrade: I find that replacing the old files in ~/.ipython (after copying |
|
330 | 343 | them to .old not optimal (for example, after every update, I have to change |
|
331 | 344 | my color settings (and some others) in ~/.ipython/ipthonrc). |
|
332 | 345 | So somehow keeping the old files and merging the new features would be |
|
333 | 346 | nice. |
|
334 | 347 | (but how to distinguish changes from version to version with changes made |
|
335 | 348 | by the user ?) For, example, I would have to change in GnuplotMagic.py gnuplot_m |
|
336 | 349 | ouse to 1 after every upgrade ... |
|
337 | 350 | \layout Standard |
|
338 | 351 | |
|
339 | 352 | This is surely a minor point - also things will change during the "BIG" |
|
340 | 353 | rewrite, but maybe this is a point to keep in mind for this ? |
|
341 | 354 | \layout Standard |
|
342 | 355 | |
|
343 | 356 | 3.) upgrade: old, sometimes obsolete files stay in the ~/.ipython subdirectory. |
|
344 | 357 | (hmm, maybe one could move all these into some subdirectory, but which |
|
345 | 358 | name for that (via version-number ?) ?) |
|
346 | 359 | \layout Subsection |
|
347 | 360 | |
|
348 | 361 | Command line options |
|
349 | 362 | \layout Standard |
|
350 | 363 | |
|
351 | 364 | It would be great to design the command-line processing system so that it |
|
352 | 365 | can be dynamically modified in some easy way. |
|
353 | 366 | This would allow systems based on IPython to include their own command-line |
|
354 | 367 | processing to either extend or fully replace IPython's. |
|
355 | 368 | Probably moving to the new optparse library (also known as optik) will |
|
356 | 369 | make this a lot easier. |
|
357 | 370 | \layout Section |
|
358 | 371 | |
|
359 | 372 | OS-dependent code |
|
360 | 373 | \layout Standard |
|
361 | 374 | |
|
362 | 375 | Options which are OS-dependent (such as colors and aliases) should be loaded |
|
363 | 376 | via include files. |
|
364 | 377 | That is, the general file will have: |
|
365 | 378 | \layout Standard |
|
366 | 379 | |
|
367 | 380 | |
|
368 | 381 | \family typewriter |
|
369 | 382 | if os.name == 'posix': |
|
370 | 383 | \newline |
|
371 | 384 | include('ipythonrc-posix.py') |
|
372 | 385 | \newline |
|
373 | 386 | elif os.name == 'nt': |
|
374 | 387 | \newline |
|
375 | 388 | include('ipythonrc-nt.py')... |
|
376 | 389 | \layout Standard |
|
377 | 390 | |
|
378 | 391 | In the |
|
379 | 392 | \family typewriter |
|
380 | 393 | -posix |
|
381 | 394 | \family default |
|
382 | 395 | , |
|
383 | 396 | \family typewriter |
|
384 | 397 | -nt |
|
385 | 398 | \family default |
|
386 | 399 | , etc. |
|
387 | 400 | files we'll set all os-specific options. |
|
388 | 401 | \layout Section |
|
389 | 402 | |
|
390 | 403 | Merging with other shell systems |
|
391 | 404 | \layout Standard |
|
392 | 405 | |
|
393 | 406 | This is listed before the big design issues, as it is something which should |
|
394 | 407 | be kept in mind when that design is made. |
|
395 | 408 | \layout Standard |
|
396 | 409 | |
|
397 | 410 | The following shell systems are out there and I think the whole design of |
|
398 | 411 | IPython should try to be modular enough to make it possible to integrate |
|
399 | 412 | its features into these. |
|
400 | 413 | In all cases IPython should exist as a stand-alone, terminal based program. |
|
401 | 414 | But it would be great if users of these other shells (some of them which |
|
402 | 415 | have very nice features of their own, especially the graphical ones) could |
|
403 | 416 | keep their environment but gain IPython's features. |
|
404 | 417 | \layout List |
|
405 | 418 | \labelwidthstring 00.00.0000 |
|
406 | 419 | |
|
407 | 420 | IDLE This is the standard, distributed as part of Python. |
|
408 | 421 | |
|
409 | 422 | \layout List |
|
410 | 423 | \labelwidthstring 00.00.0000 |
|
411 | 424 | |
|
412 | 425 | pyrepl |
|
413 | 426 | \begin_inset LatexCommand \htmlurl{http://starship.python.net/crew/mwh/hacks/pyrepl.html} |
|
414 | 427 | |
|
415 | 428 | \end_inset |
|
416 | 429 | |
|
417 | 430 | . |
|
418 | 431 | This is a text (curses-based) shell-like replacement which doesn't have |
|
419 | 432 | some of IPython's features, but has a crucially useful (and hard to implement) |
|
420 | 433 | one: full multi-line editing. |
|
421 | 434 | This turns the interactive interpreter into a true code testing and development |
|
422 | 435 | environment. |
|
423 | 436 | |
|
424 | 437 | \layout List |
|
425 | 438 | \labelwidthstring 00.00.0000 |
|
426 | 439 | |
|
427 | 440 | PyCrust |
|
428 | 441 | \begin_inset LatexCommand \htmlurl{http://sourceforge.net/projects/pycrust} |
|
429 | 442 | |
|
430 | 443 | \end_inset |
|
431 | 444 | |
|
432 | 445 | . |
|
433 | 446 | Very nice, wxWindows based system. |
|
434 | 447 | \layout List |
|
435 | 448 | \labelwidthstring 00.00.0000 |
|
436 | 449 | |
|
437 | 450 | PythonWin |
|
438 | 451 | \begin_inset LatexCommand \htmlurl{http://starship.python.net/crew/mhammond} |
|
439 | 452 | |
|
440 | 453 | \end_inset |
|
441 | 454 | |
|
442 | 455 | . |
|
443 | 456 | Similar to PyCrust in some respects, a very good and free Python development |
|
444 | 457 | environment for Windows systems. |
|
445 | 458 | \layout Section |
|
446 | 459 | |
|
447 | 460 | Class design |
|
448 | 461 | \layout Standard |
|
449 | 462 | |
|
450 | 463 | This is the big one. |
|
451 | 464 | Currently classes use each other in a very messy way, poking inside one |
|
452 | 465 | another for data and methods. |
|
453 | 466 | ipmaker() adds tons of stuff to the main __IP instance by hand, and the |
|
454 | 467 | mix-ins used (Logger, Magic, etc) mean the final __IP instance has a million |
|
455 | 468 | things in it. |
|
456 | 469 | All that needs to be cleanly broken down with well defined interfaces amongst |
|
457 | 470 | the different classes, and probably no mix-ins. |
|
458 | 471 | \layout Standard |
|
459 | 472 | |
|
460 | 473 | The best approach is probably to have all the sub-systems which are currently |
|
461 | 474 | mixins be fully independent classes which talk back only to the main instance |
|
462 | 475 | (and |
|
463 | 476 | \series bold |
|
464 | 477 | not |
|
465 | 478 | \series default |
|
466 | 479 | to each other). |
|
467 | 480 | In the main instance there should be an object whose job is to handle communica |
|
468 | 481 | tion with the sub-systems. |
|
469 | 482 | \layout Standard |
|
470 | 483 | |
|
471 | 484 | I should probably learn a little UML and diagram this whole thing before |
|
472 | 485 | I start coding. |
|
473 | 486 | \layout Subsection |
|
474 | 487 | |
|
475 | 488 | Magic |
|
476 | 489 | \layout Standard |
|
477 | 490 | |
|
478 | 491 | Now all methods which will become publicly available are called Magic.magic_name, |
|
479 | 492 | the magic_ should go away. |
|
480 | 493 | Then, Magic instead of being a mix-in should simply be an attribute of |
|
481 | 494 | __IP: |
|
482 | 495 | \layout Standard |
|
483 | 496 | |
|
484 | 497 | __IP.Magic = Magic() |
|
485 | 498 | \layout Standard |
|
486 | 499 | |
|
487 | 500 | This will then give all the magic functions as __IP.Magic.name(), which is |
|
488 | 501 | much cleaner. |
|
489 | 502 | This will also force a better separation so that Magic doesn't poke inside |
|
490 | 503 | __IP so much. |
|
491 | 504 | In the constructor, Magic should get whatever information it needs to know |
|
492 | 505 | about __IP (even if it means a pointer to __IP itself, but at least we'll |
|
493 | 506 | know where it is. |
|
494 | 507 | Right now since it's a mix-in, there's no way to know which variables belong |
|
495 | 508 | to whom). |
|
496 | 509 | \layout Standard |
|
497 | 510 | |
|
498 | 511 | Build a class MagicFunction so that adding new functions is a matter of: |
|
499 | 512 | \layout Standard |
|
500 | 513 | |
|
501 | 514 | |
|
502 | 515 | \family typewriter |
|
503 | 516 | my_magic = MagicFunction(category = 'System utilities') |
|
504 | 517 | \newline |
|
505 | 518 | my_magic.__call__ = ... |
|
506 | 519 | \layout Standard |
|
507 | 520 | |
|
508 | 521 | Features: |
|
509 | 522 | \layout Itemize |
|
510 | 523 | |
|
511 | 524 | The class constructor should automatically register the functions and keep |
|
512 | 525 | a table with category sections for easy sorting/viewing. |
|
513 | 526 | \layout Itemize |
|
514 | 527 | |
|
515 | 528 | The object interface must allow automatic building of a GUI for them. |
|
516 | 529 | This requires registering the options the command takes, the number of |
|
517 | 530 | arguments, etc, in a formal way. |
|
518 | 531 | The advantage of this approach is that it allows not only to add GUIs to |
|
519 | 532 | the magics, but also for a much more intelligent building of docstrings, |
|
520 | 533 | and better parsing of options and arguments. |
|
521 | 534 | \layout Standard |
|
522 | 535 | |
|
523 | 536 | Also think through better an alias system for magics. |
|
524 | 537 | Since the magic system is like a command shell inside ipython, the relation |
|
525 | 538 | between these aliases and system aliases should be cleanly thought out. |
|
526 | 539 | \layout Subsection |
|
527 | 540 | |
|
528 | 541 | Color schemes |
|
529 | 542 | \layout Standard |
|
530 | 543 | |
|
531 | 544 | These should be loaded from some kind of resource file so they are easier |
|
532 | 545 | to modify by the user. |
|
533 | 546 | \layout Section |
|
534 | 547 | |
|
535 | 548 | Hooks |
|
536 | 549 | \layout Standard |
|
537 | 550 | |
|
538 | 551 | IPython should have a modular system where functions can register themselves |
|
539 | 552 | for certain tasks. |
|
540 | 553 | Currently changing functionality requires overriding certain specific methods, |
|
541 | 554 | there should be a clean API for this to be done. |
|
542 | 555 | \layout Subsection |
|
543 | 556 | |
|
544 | 557 | whos hook |
|
545 | 558 | \layout Standard |
|
546 | 559 | |
|
547 | 560 | This was a very nice suggestion from Alexander Schmolck <a.schmolck@gmx.net>: |
|
548 | 561 | \layout Standard |
|
549 | 562 | |
|
550 | 563 | 2. |
|
551 | 564 | I think it would also be very helpful if there where some sort of hook |
|
552 | 565 | for ``whos`` that let one customize display formaters depending on the |
|
553 | 566 | object type. |
|
554 | 567 | \layout Standard |
|
555 | 568 | |
|
556 | 569 | For example I'd rather have a whos that formats an array like: |
|
557 | 570 | \layout Standard |
|
558 | 571 | |
|
559 | 572 | |
|
560 | 573 | \family typewriter |
|
561 | 574 | Variable Type Data/Length |
|
562 | 575 | \newline |
|
563 | 576 | ------------------------------ |
|
564 | 577 | \newline |
|
565 | 578 | a array size: 4x3 type: 'Float' |
|
566 | 579 | \layout Standard |
|
567 | 580 | |
|
568 | 581 | than |
|
569 | 582 | \layout Standard |
|
570 | 583 | |
|
571 | 584 | |
|
572 | 585 | \family typewriter |
|
573 | 586 | Variable Type Data/Length |
|
574 | 587 | \newline |
|
575 | 588 | ------------------------------ |
|
576 | 589 | \newline |
|
577 | 590 | a array [[ 0. |
|
578 | 591 | 1. |
|
579 | 592 | 2. |
|
580 | 593 | 3<...> 8. |
|
581 | 594 | 9. |
|
582 | 595 | 10. |
|
583 | 596 | 11.]] |
|
584 | 597 | \layout Section |
|
585 | 598 | |
|
586 | 599 | Parallel support |
|
587 | 600 | \layout Standard |
|
588 | 601 | |
|
589 | 602 | For integration with graphical shells and other systems, it will be best |
|
590 | 603 | if ipython is split into a kernel/client model, much like Mathematica works. |
|
591 | 604 | This simultaneously opens the door for support of interactive parallel |
|
592 | 605 | computing. |
|
593 | 606 | Currenlty %bg provides a threads-based proof of concept, and Brian Granger's |
|
594 | 607 | XGrid project is a much more realistic such system. |
|
595 | 608 | The new design should integrates ideas as core elements. |
|
596 | 609 | Some notes from Brian on this topic: |
|
597 | 610 | \layout Standard |
|
598 | 611 | |
|
599 | 612 | 1. |
|
600 | 613 | How should the remote python server/kernel be designed? Multithreaded? |
|
601 | 614 | Blocking? Connected/disconnected modes? Load balancing? |
|
602 | 615 | \layout Standard |
|
603 | 616 | |
|
604 | 617 | 2. |
|
605 | 618 | What APi/protocol should the server/kernel expose to clients? |
|
606 | 619 | \layout Standard |
|
607 | 620 | |
|
608 | 621 | 3. |
|
609 | 622 | How should the client classes (which the user uses to interact with the |
|
610 | 623 | cluster) be designed? |
|
611 | 624 | \layout Standard |
|
612 | 625 | |
|
613 | 626 | 4. |
|
614 | 627 | What API should the client classes expose? |
|
615 | 628 | \layout Standard |
|
616 | 629 | |
|
617 | 630 | 5. |
|
618 | 631 | How should the client API be wrapped in a few simple magic functions? |
|
619 | 632 | \layout Standard |
|
620 | 633 | |
|
621 | 634 | 6. |
|
622 | 635 | How should security be handled? |
|
623 | 636 | \layout Standard |
|
624 | 637 | |
|
625 | 638 | 7. |
|
626 | 639 | How to work around the issues of the GIL and threads? |
|
627 | 640 | \layout Standard |
|
628 | 641 | |
|
629 | 642 | I think the most important things to work out are the client API (#4) the |
|
630 | 643 | server/kernel API/protocol (#2) and the magic function API (#5). |
|
631 | 644 | We should let these determine the design and architecture of the components. |
|
632 | 645 | \layout Standard |
|
633 | 646 | |
|
634 | 647 | One other thing. |
|
635 | 648 | What is your impression of twisted? I have been looking at it and it looks |
|
636 | 649 | like a _very_ powerful set of tools for this type of stuff. |
|
637 | 650 | I am wondering if it might make sense to think about using twisted for |
|
638 | 651 | this project. |
|
639 | 652 | \layout Section |
|
640 | 653 | |
|
641 | 654 | Manuals |
|
642 | 655 | \layout Standard |
|
643 | 656 | |
|
644 | 657 | The documentation should be generated from docstrings for the command line |
|
645 | 658 | args and all the magic commands. |
|
646 | 659 | Look into one of the simple text markup systems to see if we can get latex |
|
647 | 660 | (for reLyXing later) out of this. |
|
648 | 661 | Part of the build command would then be to make an update of the docs based |
|
649 | 662 | on this, thus giving more complete manual (and guaranteed to be in sync |
|
650 | 663 | with the code docstrings). |
|
651 | 664 | \layout Standard |
|
652 | 665 | |
|
653 | 666 | [PARTLY DONE] At least now all magics are auto-documented, works farily |
|
654 | 667 | well. |
|
655 | 668 | Limited Latex formatting yet. |
|
656 | 669 | \layout Subsection |
|
657 | 670 | |
|
658 | 671 | Integration with pydoc-help |
|
659 | 672 | \layout Standard |
|
660 | 673 | |
|
661 | 674 | It should be possible to have access to the manual via the pydoc help system |
|
662 | 675 | somehow. |
|
663 | 676 | This might require subclassing the pydoc help, or figuring out how to add |
|
664 | 677 | the IPython docs in the right form so that help() finds them. |
|
665 | 678 | \layout Standard |
|
666 | 679 | |
|
667 | 680 | Some comments from Arnd and my reply on this topic: |
|
668 | 681 | \layout Standard |
|
669 | 682 | |
|
670 | 683 | > ((Generally I would like to have the nice documentation > more easily |
|
671 | 684 | accessable from within ipython ... |
|
672 | 685 | > Many people just don't read documentation, even if it is > as good as |
|
673 | 686 | the one of IPython )) |
|
674 | 687 | \layout Standard |
|
675 | 688 | |
|
676 | 689 | That's an excellent point. |
|
677 | 690 | I've added a note to this effect in new_design. |
|
678 | 691 | Basically I'd like help() to naturally access the IPython docs. |
|
679 | 692 | Since they are already there in html for the user, it's probably a matter |
|
680 | 693 | of playing a bit with pydoc to tell it where to find them. |
|
681 | 694 | It would definitely make for a much cleaner system. |
|
682 | 695 | Right now the information on IPython is: |
|
683 | 696 | \layout Standard |
|
684 | 697 | |
|
685 | 698 | -ipython --help at the command line: info on command line switches |
|
686 | 699 | \layout Standard |
|
687 | 700 | |
|
688 | 701 | -? at the ipython prompt: overview of IPython |
|
689 | 702 | \layout Standard |
|
690 | 703 | |
|
691 | 704 | -magic at the ipython prompt: overview of the magic system |
|
692 | 705 | \layout Standard |
|
693 | 706 | |
|
694 | 707 | -external docs (html/pdf) |
|
695 | 708 | \layout Standard |
|
696 | 709 | |
|
697 | 710 | All that should be better integrated seamlessly in the help() system, so |
|
698 | 711 | that you can simply say: |
|
699 | 712 | \layout Standard |
|
700 | 713 | |
|
701 | 714 | help ipython -> full documentation access |
|
702 | 715 | \layout Standard |
|
703 | 716 | |
|
704 | 717 | help magic -> magic overview |
|
705 | 718 | \layout Standard |
|
706 | 719 | |
|
707 | 720 | help profile -> help on current profile |
|
708 | 721 | \layout Standard |
|
709 | 722 | |
|
710 | 723 | help -> normal python help access. |
|
711 | 724 | \layout Section |
|
712 | 725 | |
|
713 | 726 | Graphical object browsers |
|
714 | 727 | \layout Standard |
|
715 | 728 | |
|
716 | 729 | I'd like a system for graphically browsing through objects. |
|
717 | 730 | |
|
718 | 731 | \family typewriter |
|
719 | 732 | @obrowse |
|
720 | 733 | \family default |
|
721 | 734 | should open a widged with all the things which |
|
722 | 735 | \family typewriter |
|
723 | 736 | @who |
|
724 | 737 | \family default |
|
725 | 738 | lists, but cliking on each object would open a dedicated object viewer |
|
726 | 739 | (also accessible as |
|
727 | 740 | \family typewriter |
|
728 | 741 | @oview <object> |
|
729 | 742 | \family default |
|
730 | 743 | ). |
|
731 | 744 | This object viewer could show a summary of what |
|
732 | 745 | \family typewriter |
|
733 | 746 | <object>? |
|
734 | 747 | \family default |
|
735 | 748 | currently shows, but also colorize source code and show it via an html |
|
736 | 749 | browser, show all attributes and methods of a given object (themselves |
|
737 | 750 | openable in their own viewers, since in Python everything is an object), |
|
738 | 751 | links to the parent classes, etc. |
|
739 | 752 | \layout Standard |
|
740 | 753 | |
|
741 | 754 | The object viewer widget should be extensible, so that one can add methods |
|
742 | 755 | to view certain types of objects in a special way (for example, plotting |
|
743 | 756 | Numeric arrays via grace or gnuplot). |
|
744 | 757 | This would be very useful when using IPython as part of an interactive |
|
745 | 758 | complex system for working with certain types of data. |
|
746 | 759 | \layout Standard |
|
747 | 760 | |
|
748 | 761 | I should look at what PyCrust has to offer along these lines, at least as |
|
749 | 762 | a starting point. |
|
750 | 763 | \layout Section |
|
751 | 764 | |
|
752 | 765 | Miscellaneous small things |
|
753 | 766 | \layout Itemize |
|
754 | 767 | |
|
755 | 768 | Collect whatever variables matter from the environment in some globals for |
|
756 | 769 | __IP, so we're not testing for them constantly (like $HOME, $TERM, etc.) |
|
757 | 770 | \layout Section |
|
758 | 771 | |
|
759 | 772 | Session restoring |
|
760 | 773 | \layout Standard |
|
761 | 774 | |
|
762 | 775 | I've convinced myself that session restore by log replay is too fragile |
|
763 | 776 | and tricky to ever work reliably. |
|
764 | 777 | Plus it can be dog slow. |
|
765 | 778 | I'd rather have a way of saving/restoring the *current* memory state of |
|
766 | 779 | IPython. |
|
767 | 780 | I tried with pickle but failed (can't pickle modules). |
|
768 | 781 | This seems the right way to do it to me, but it will have to wait until |
|
769 | 782 | someone tells me of a robust way of dumping/reloading *all* of the user |
|
770 | 783 | namespace in a file. |
|
771 | 784 | \layout Standard |
|
772 | 785 | |
|
773 | 786 | Probably the best approach will be to pickle as much as possible and record |
|
774 | 787 | what can not be pickled for manual reload (such as modules). |
|
775 | 788 | This is not trivial to get to work reliably, so it's best left for after |
|
776 | 789 | the code restructuring. |
|
777 | 790 | \layout Standard |
|
778 | 791 | |
|
779 | 792 | The following issues exist (old notes, see above paragraph for my current |
|
780 | 793 | take on the issue): |
|
781 | 794 | \layout Itemize |
|
782 | 795 | |
|
783 | 796 | magic lines aren't properly re-executed when a log file is reloaded (and |
|
784 | 797 | some of them, like clear or run, may change the environment). |
|
785 | 798 | So session restore isn't 100% perfect. |
|
786 | 799 | \layout Itemize |
|
787 | 800 | |
|
788 | 801 | auto-quote/parens lines aren't replayed either. |
|
789 | 802 | All this could be done, but it needs some work. |
|
790 | 803 | Basically it requires re-running the log through IPython itself, not through |
|
791 | 804 | python. |
|
792 | 805 | \layout Itemize |
|
793 | 806 | |
|
794 | 807 | _p variables aren't restored with a session. |
|
795 | 808 | Fix: same as above. |
|
796 | 809 | \layout Section |
|
797 | 810 | |
|
798 | 811 | Tips system |
|
799 | 812 | \layout Standard |
|
800 | 813 | |
|
801 | 814 | It would be nice to have a tip() function which gives tips to users in some |
|
802 | 815 | situations, but keeps track of already-given tips so they aren't given |
|
803 | 816 | every time. |
|
804 | 817 | This could be done by pickling a dict of given tips to IPYTHONDIR. |
|
805 | 818 | \layout Section |
|
806 | 819 | |
|
807 | 820 | TAB completer |
|
808 | 821 | \layout Standard |
|
809 | 822 | |
|
810 | 823 | Some suggestions from Arnd Baecker: |
|
811 | 824 | \layout Standard |
|
812 | 825 | |
|
813 | 826 | a) For file related commands (ls, cat, ...) it would be nice to be able to |
|
814 | 827 | TAB complete the files in the current directory. |
|
815 | 828 | (once you started typing something which is uniquely a file, this leads |
|
816 | 829 | to this effect, apart from going through the list of possible completions |
|
817 | 830 | ...). |
|
818 | 831 | (I know that this point is in your documentation.) |
|
819 | 832 | \layout Standard |
|
820 | 833 | |
|
821 | 834 | More general, this might lead to something like command specific completion |
|
822 | 835 | ? |
|
823 | 836 | \layout Standard |
|
824 | 837 | |
|
825 | 838 | Here's John Hunter's suggestion: |
|
826 | 839 | \layout Standard |
|
827 | 840 | |
|
828 | 841 | The *right way to do it* would be to make intelligent or customizable choices |
|
829 | 842 | about which namespace to add to the completion list depending on the string |
|
830 | 843 | match up to the prompt, eg programmed completions. |
|
831 | 844 | In the simplest implementation, one would only complete on files and directorie |
|
832 | 845 | s if the line preceding the tab press matched 'cd ' or 'run ' (eg you don't |
|
833 | 846 | want callable showing up in 'cd ca<TAB>') |
|
834 | 847 | \layout Standard |
|
835 | 848 | |
|
836 | 849 | In a more advanced scenario, you might imaging that functions supplied the |
|
837 | 850 | TAB namespace, and the user could configure a dictionary that mapped regular |
|
838 | 851 | expressions to namespace providing functions (with sensible defaults). |
|
839 | 852 | Something like |
|
840 | 853 | \layout Standard |
|
841 | 854 | |
|
842 | 855 | completed = { |
|
843 | 856 | \newline |
|
844 | 857 | '^cd |
|
845 | 858 | \backslash |
|
846 | 859 | s+(.*)' : complete_files_and_dirs, |
|
847 | 860 | \newline |
|
848 | 861 | '^run |
|
849 | 862 | \backslash |
|
850 | 863 | s+(.*)' : complete_files_and_dirs, |
|
851 | 864 | \newline |
|
852 | 865 | '^run |
|
853 | 866 | \backslash |
|
854 | 867 | s+(-.*)' : complete_run_options, |
|
855 | 868 | \newline |
|
856 | 869 | } |
|
857 | 870 | \layout Standard |
|
858 | 871 | |
|
859 | 872 | I don't know if this is feasible, but I really like programmed completions, |
|
860 | 873 | which I use extensively in tcsh. |
|
861 | 874 | My feeling is that something like this is eminently doable in ipython. |
|
862 | 875 | \layout Standard |
|
863 | 876 | |
|
864 | 877 | /JDH |
|
865 | 878 | \layout Standard |
|
866 | 879 | |
|
867 | 880 | For something like this to work cleanly, the magic command system needs |
|
868 | 881 | also a clean options framework, so all valid options for a given magic |
|
869 | 882 | can be extracted programatically. |
|
870 | 883 | \layout Section |
|
871 | 884 | |
|
872 | 885 | Debugger |
|
873 | 886 | \layout Standard |
|
874 | 887 | |
|
875 | 888 | Current system uses a minimally tweaked pdb. |
|
876 | 889 | Fine-tune it a bit, to provide at least: |
|
877 | 890 | \layout Itemize |
|
878 | 891 | |
|
879 | 892 | Tab-completion in each stack frame. |
|
880 | 893 | See email to Chris Hart for details. |
|
881 | 894 | \layout Itemize |
|
882 | 895 | |
|
883 | 896 | Object information via ? at least. |
|
884 | 897 | Break up magic_oinfo a bit so that pdb can call it without loading all |
|
885 | 898 | of IPython. |
|
886 | 899 | If possible, also have the other magics for object study: doc, source, |
|
887 | 900 | pdef and pfile. |
|
888 | 901 | \layout Itemize |
|
889 | 902 | |
|
890 | 903 | Shell access via ! |
|
891 | 904 | \layout Itemize |
|
892 | 905 | |
|
893 | 906 | Syntax highlighting in listings. |
|
894 | 907 | Use py2html code, implement color schemes. |
|
895 | 908 | \layout Section |
|
896 | 909 | |
|
897 | 910 | A Python-based system shell - pysh? |
|
898 | 911 | \layout Standard |
|
899 | 912 | |
|
900 | 913 | Note: as of IPython 0.6.1, most of this functionality has actually been implemente |
|
901 | 914 | d. |
|
902 | 915 | \layout Standard |
|
903 | 916 | |
|
904 | 917 | This section is meant as a working draft for discussions on the possibility |
|
905 | 918 | of having a python-based system shell. |
|
906 | 919 | It is the result of my own thinking about these issues as much of discussions |
|
907 | 920 | on the ipython lists. |
|
908 | 921 | I apologize in advance for not giving individual credit to the various |
|
909 | 922 | contributions, but right now I don't have the time to track down each message |
|
910 | 923 | from the archives. |
|
911 | 924 | So please consider this as the result of a collective effort by the ipython |
|
912 | 925 | user community. |
|
913 | 926 | \layout Standard |
|
914 | 927 | |
|
915 | 928 | While IPyhton is (and will remain) a python shell first, it does offer a |
|
916 | 929 | fair amount of system access functionality: |
|
917 | 930 | \layout Standard |
|
918 | 931 | |
|
919 | 932 | - ! and !! for direct system access, |
|
920 | 933 | \layout Standard |
|
921 | 934 | |
|
922 | 935 | - magic commands which wrap various system commands, |
|
923 | 936 | \layout Standard |
|
924 | 937 | |
|
925 | 938 | - @sc and @sx, for shell output capture into python variables, |
|
926 | 939 | \layout Standard |
|
927 | 940 | |
|
928 | 941 | - @alias, for aliasing system commands. |
|
929 | 942 | \layout Standard |
|
930 | 943 | |
|
931 | 944 | This has prompted many users, over time, to ask for a way of extending ipython |
|
932 | 945 | to the point where it could be used as a full-time replacement over typical |
|
933 | 946 | user shells like bash, csh or tcsh. |
|
934 | 947 | While my interest in ipython is such that I'll concentrate my personal |
|
935 | 948 | efforts on other fronts (debugging, architecture, improvements for scientific |
|
936 | 949 | use, gui access), I will be happy to do anything which could make such |
|
937 | 950 | a development possible. |
|
938 | 951 | It would be the responsibility of someone else to maintain the code, but |
|
939 | 952 | I would do all necessary architectural changes to ipython for such an extension |
|
940 | 953 | to be feasible. |
|
941 | 954 | \layout Standard |
|
942 | 955 | |
|
943 | 956 | I'll try to outline here what I see as the key issues which need to be taken |
|
944 | 957 | into account. |
|
945 | 958 | This document should be considered an evolving draft. |
|
946 | 959 | Feel free to submit comments/improvements, even in the form of patches. |
|
947 | 960 | \layout Standard |
|
948 | 961 | |
|
949 | 962 | In what follows, I'll represent the hypothetical python-based shell ('pysh' |
|
950 | 963 | for now) prompt with '>>'. |
|
951 | 964 | \layout Subsection |
|
952 | 965 | |
|
953 | 966 | Basic design principles |
|
954 | 967 | \layout Standard |
|
955 | 968 | |
|
956 | 969 | I think the basic design guideline should be the following: a hypothetical |
|
957 | 970 | python system shell should behave, as much as possible, like a normal shell |
|
958 | 971 | that users are familiar with (bash, tcsh, etc). |
|
959 | 972 | This means: |
|
960 | 973 | \layout Standard |
|
961 | 974 | |
|
962 | 975 | 1. |
|
963 | 976 | System commands can be issued directly at the prompt with no special syntax: |
|
964 | 977 | \layout Standard |
|
965 | 978 | |
|
966 | 979 | >> ls |
|
967 | 980 | \layout Standard |
|
968 | 981 | |
|
969 | 982 | >> xemacs |
|
970 | 983 | \layout Standard |
|
971 | 984 | |
|
972 | 985 | should just work like a user expects. |
|
973 | 986 | \layout Standard |
|
974 | 987 | |
|
975 | 988 | 2. |
|
976 | 989 | The facilities of the python language should always be available, like |
|
977 | 990 | they are in ipython: |
|
978 | 991 | \layout Standard |
|
979 | 992 | |
|
980 | 993 | >> 3+4 |
|
981 | 994 | \newline |
|
982 | 995 | 7 |
|
983 | 996 | \layout Standard |
|
984 | 997 | |
|
985 | 998 | 3. |
|
986 | 999 | It should be possible to easily capture shell output into a variable. |
|
987 | 1000 | bash and friends use backquotes, I think using a command (@sc) like ipython |
|
988 | 1001 | currently does is an acceptable compromise. |
|
989 | 1002 | \layout Standard |
|
990 | 1003 | |
|
991 | 1004 | 4. |
|
992 | 1005 | It should also be possible to expand python variables/commands in the middle |
|
993 | 1006 | of system commands. |
|
994 | 1007 | I thihk this will make it necessary to use $var for name expansions: |
|
995 | 1008 | \layout Standard |
|
996 | 1009 | |
|
997 | 1010 | >> var='hello' # var is a Python variable |
|
998 | 1011 | \newline |
|
999 | 1012 | >> print var hello # This is the result of a Python print command |
|
1000 | 1013 | \newline |
|
1001 | 1014 | >> echo $var hello # This calls the echo command, expanding 'var'. |
|
1002 | 1015 | \layout Standard |
|
1003 | 1016 | |
|
1004 | 1017 | 5. |
|
1005 | 1018 | The above capabilities should remain possible for multi-line commands. |
|
1006 | 1019 | One of the most annoying things I find about tcsh, is that I never quite |
|
1007 | 1020 | remember the syntactic details of looping. |
|
1008 | 1021 | I often want to do something at the shell which involves a simple loop, |
|
1009 | 1022 | but I can never remember how to do it in tcsh. |
|
1010 | 1023 | This often means I just write a quick throwaway python script to do it |
|
1011 | 1024 | (Perl is great for this kind of quick things, but I've forgotten most its |
|
1012 | 1025 | syntax as well). |
|
1013 | 1026 | \layout Standard |
|
1014 | 1027 | |
|
1015 | 1028 | It should be possible to write code like: |
|
1016 | 1029 | \layout Standard |
|
1017 | 1030 | |
|
1018 | 1031 | >> for ext in ['.jpg','.gif']: |
|
1019 | 1032 | \newline |
|
1020 | 1033 | .. |
|
1021 | 1034 | ls file$ext |
|
1022 | 1035 | \layout Standard |
|
1023 | 1036 | |
|
1024 | 1037 | And have it work as 'ls file.jpg;ls file.gif'. |
|
1025 | 1038 | \layout Subsection |
|
1026 | 1039 | |
|
1027 | 1040 | Smaller details |
|
1028 | 1041 | \layout Standard |
|
1029 | 1042 | |
|
1030 | 1043 | If the above are considered as valid guiding principles for how such a python |
|
1031 | 1044 | system shell should behave, then some smaller considerations and comments |
|
1032 | 1045 | to keep in mind are listed below. |
|
1033 | 1046 | \layout Standard |
|
1034 | 1047 | |
|
1035 | 1048 | - it's ok for shell builtins (in this case this includes the python language) |
|
1036 | 1049 | to override system commands on the path. |
|
1037 | 1050 | See tcsh's 'time' vs '/usr/bin/time'. |
|
1038 | 1051 | This settles the 'print' issue and related. |
|
1039 | 1052 | \layout Standard |
|
1040 | 1053 | |
|
1041 | 1054 | - pysh should take |
|
1042 | 1055 | \layout Standard |
|
1043 | 1056 | |
|
1044 | 1057 | foo args |
|
1045 | 1058 | \layout Standard |
|
1046 | 1059 | |
|
1047 | 1060 | as a command if (foo args is NOT valid python) and (foo is in $PATH). |
|
1048 | 1061 | \layout Standard |
|
1049 | 1062 | |
|
1050 | 1063 | If the user types |
|
1051 | 1064 | \layout Standard |
|
1052 | 1065 | |
|
1053 | 1066 | >> ./foo args |
|
1054 | 1067 | \layout Standard |
|
1055 | 1068 | |
|
1056 | 1069 | it should be considered a system command always. |
|
1057 | 1070 | \layout Standard |
|
1058 | 1071 | |
|
1059 | 1072 | - _, __ and ___ should automatically remember the previous 3 outputs captured |
|
1060 | 1073 | from stdout. |
|
1061 | 1074 | In parallel, there should be _e, __e and ___e for stderr. |
|
1062 | 1075 | Whether capture is done as a single string or in list mode should be a |
|
1063 | 1076 | user preference. |
|
1064 | 1077 | If users have numbered prompts, ipython's full In/Out cache system should |
|
1065 | 1078 | be available. |
|
1066 | 1079 | \layout Standard |
|
1067 | 1080 | |
|
1068 | 1081 | But regardless of how variables are captured, the printout should be like |
|
1069 | 1082 | that of a plain shell (without quotes or braces to indicate strings/lists). |
|
1070 | 1083 | The everyday 'feel' of pysh should be more that of bash/tcsh than that |
|
1071 | 1084 | of ipython. |
|
1072 | 1085 | \layout Standard |
|
1073 | 1086 | |
|
1074 | 1087 | - filename completion first. |
|
1075 | 1088 | Tab completion could work like in ipython, but with the order of priorities |
|
1076 | 1089 | reversed: first files, then python names. |
|
1077 | 1090 | \layout Standard |
|
1078 | 1091 | |
|
1079 | 1092 | - configuration via standard python files. |
|
1080 | 1093 | Instead of 'setenv' you'd simply write into the os.environ[] dictionary. |
|
1081 | 1094 | This assumes that IPython itself has been fixed to be configured via normal |
|
1082 | 1095 | python files, instead of the current clunky ipythonrc format. |
|
1083 | 1096 | \layout Standard |
|
1084 | 1097 | |
|
1085 | 1098 | - IPython can already configure the prompt in fairly generic ways. |
|
1086 | 1099 | It should be able to generate almost any kind of prompt which bash/tcsh |
|
1087 | 1100 | can (within reason). |
|
1088 | 1101 | \layout Standard |
|
1089 | 1102 | |
|
1090 | 1103 | - Keep the Magics system. |
|
1091 | 1104 | They provide a lightweight syntax for configuring and modifying the state |
|
1092 | 1105 | of the user's session itself. |
|
1093 | 1106 | Plus, they are an extensible system so why not give the users one more |
|
1094 | 1107 | tool which is fairly flexible by nature? Finally, having the @magic optional |
|
1095 | 1108 | syntax allows a user to always be able to access the shell's control system, |
|
1096 | 1109 | regardless of name collisions with defined variables or system commands. |
|
1097 | 1110 | \layout Standard |
|
1098 | 1111 | |
|
1099 | 1112 | But we need to move all magic functionality into a protected namespace, |
|
1100 | 1113 | instead of the current messy name-mangling tricks (which don't scale well). |
|
1101 | 1114 | |
|
1102 | 1115 | \layout Section |
|
1103 | 1116 | |
|
1104 | 1117 | Future improvements |
|
1105 | 1118 | \layout Itemize |
|
1106 | 1119 | |
|
1107 | 1120 | When from <mod> import * is used, first check the existing namespace and |
|
1108 | 1121 | at least issue a warning on screen if names are overwritten. |
|
1109 | 1122 | \layout Itemize |
|
1110 | 1123 | |
|
1111 | 1124 | Auto indent? Done, for users with readline support. |
|
1112 | 1125 | \layout Subsection |
|
1113 | 1126 | |
|
1114 | 1127 | Better completion a la zsh |
|
1115 | 1128 | \layout Standard |
|
1116 | 1129 | |
|
1117 | 1130 | This was suggested by Arnd: |
|
1118 | 1131 | \layout Standard |
|
1119 | 1132 | |
|
1120 | 1133 | > >\SpecialChar ~ |
|
1121 | 1134 | \SpecialChar ~ |
|
1122 | 1135 | \SpecialChar ~ |
|
1123 | 1136 | More general, this might lead to something like |
|
1124 | 1137 | \layout Standard |
|
1125 | 1138 | |
|
1126 | 1139 | > >\SpecialChar ~ |
|
1127 | 1140 | \SpecialChar ~ |
|
1128 | 1141 | \SpecialChar ~ |
|
1129 | 1142 | command specific completion ? |
|
1130 | 1143 | \layout Standard |
|
1131 | 1144 | |
|
1132 | 1145 | > |
|
1133 | 1146 | \layout Standard |
|
1134 | 1147 | |
|
1135 | 1148 | > I'm not sure what you mean here. |
|
1136 | 1149 | \layout Standard |
|
1137 | 1150 | |
|
1138 | 1151 | \SpecialChar ~ |
|
1139 | 1152 | |
|
1140 | 1153 | \layout Standard |
|
1141 | 1154 | |
|
1142 | 1155 | Sorry, that was not understandable, indeed ... |
|
1143 | 1156 | \layout Standard |
|
1144 | 1157 | |
|
1145 | 1158 | I thought of something like |
|
1146 | 1159 | \layout Standard |
|
1147 | 1160 | |
|
1148 | 1161 | \SpecialChar ~ |
|
1149 | 1162 | - cd and then use TAB to go through the list of directories |
|
1150 | 1163 | \layout Standard |
|
1151 | 1164 | |
|
1152 | 1165 | \SpecialChar ~ |
|
1153 | 1166 | - ls and then TAB to consider all files and directories |
|
1154 | 1167 | \layout Standard |
|
1155 | 1168 | |
|
1156 | 1169 | \SpecialChar ~ |
|
1157 | 1170 | - cat and TAB: only files (no directories ...) |
|
1158 | 1171 | \layout Standard |
|
1159 | 1172 | |
|
1160 | 1173 | \SpecialChar ~ |
|
1161 | 1174 | |
|
1162 | 1175 | \layout Standard |
|
1163 | 1176 | |
|
1164 | 1177 | For zsh things like this are established by defining in .zshrc |
|
1165 | 1178 | \layout Standard |
|
1166 | 1179 | |
|
1167 | 1180 | \SpecialChar ~ |
|
1168 | 1181 | |
|
1169 | 1182 | \layout Standard |
|
1170 | 1183 | |
|
1171 | 1184 | compctl -g '*.dvi' xdvi |
|
1172 | 1185 | \layout Standard |
|
1173 | 1186 | |
|
1174 | 1187 | compctl -g '*.dvi' dvips |
|
1175 | 1188 | \layout Standard |
|
1176 | 1189 | |
|
1177 | 1190 | compctl -g '*.tex' latex |
|
1178 | 1191 | \layout Standard |
|
1179 | 1192 | |
|
1180 | 1193 | compctl -g '*.tex' tex |
|
1181 | 1194 | \layout Standard |
|
1182 | 1195 | |
|
1183 | 1196 | ... |
|
1184 | 1197 | \layout Section |
|
1185 | 1198 | |
|
1186 | 1199 | Outline of steps |
|
1187 | 1200 | \layout Standard |
|
1188 | 1201 | |
|
1189 | 1202 | Here's a rough outline of the order in which to start implementing the various |
|
1190 | 1203 | parts of the redesign. |
|
1191 | 1204 | The first 'test of success' should be a clean pychecker run (not the mess |
|
1192 | 1205 | we get right now). |
|
1193 | 1206 | \layout Itemize |
|
1194 | 1207 | |
|
1195 | 1208 | Make Logger and Magic not be mixins but attributes of the main class. |
|
1196 | 1209 | |
|
1197 | 1210 | \begin_deeper |
|
1198 | 1211 | \layout Itemize |
|
1199 | 1212 | |
|
1200 | 1213 | Magic should have a pointer back to the main instance (even if this creates |
|
1201 | 1214 | a recursive structure) so it can control it with minimal message-passing |
|
1202 | 1215 | machinery. |
|
1203 | 1216 | |
|
1204 | 1217 | \layout Itemize |
|
1205 | 1218 | |
|
1206 | 1219 | Logger can be a standalone object, simply with a nice, clean interface. |
|
1207 | 1220 | \layout Itemize |
|
1208 | 1221 | |
|
1209 | 1222 | Logger currently handles part of the prompt caching, but other parts of |
|
1210 | 1223 | that are in the prompts class itself. |
|
1211 | 1224 | Clean up. |
|
1212 | 1225 | \end_deeper |
|
1213 | 1226 | \layout Itemize |
|
1214 | 1227 | |
|
1215 | 1228 | Change to python-based config system. |
|
1216 | 1229 | \layout Itemize |
|
1217 | 1230 | |
|
1218 | 1231 | Move make_IPython() into the main shell class, as part of the constructor. |
|
1219 | 1232 | Do this |
|
1220 | 1233 | \emph on |
|
1221 | 1234 | after |
|
1222 | 1235 | \emph default |
|
1223 | 1236 | the config system has been changed, debugging will be a lot easier then. |
|
1224 | 1237 | \layout Itemize |
|
1225 | 1238 | |
|
1226 | 1239 | Merge the embeddable class and the normal one into one. |
|
1227 | 1240 | After all, the standard ipython script |
|
1228 | 1241 | \emph on |
|
1229 | 1242 | is |
|
1230 | 1243 | \emph default |
|
1231 | 1244 | a python program with IPython embedded in it. |
|
1232 | 1245 | There's no need for two separate classes ( |
|
1233 | 1246 | \emph on |
|
1234 | 1247 | maybe |
|
1235 | 1248 | \emph default |
|
1236 | 1249 | keep the old one around for the sake of backwards compatibility). |
|
1237 | 1250 | \layout Section |
|
1238 | 1251 | |
|
1239 | 1252 | Ville Vainio's suggestions |
|
1240 | 1253 | \layout Standard |
|
1241 | 1254 | |
|
1242 | 1255 | Some notes sent in by Ville Vainio |
|
1243 | 1256 | \family typewriter |
|
1244 | 1257 | <vivainio@kolumbus.fi> |
|
1245 | 1258 | \family default |
|
1246 | 1259 | on Tue, 29 Jun 2004. |
|
1247 | 1260 | Keep here for reference, some of it replicates things already said above. |
|
1248 | 1261 | \layout Standard |
|
1249 | 1262 | |
|
1250 | 1263 | Current ipython seems to "special case" lots of stuff - aliases, magics |
|
1251 | 1264 | etc. |
|
1252 | 1265 | It would seem to yield itself to a simpler and more extensible architecture, |
|
1253 | 1266 | consisting of multple dictionaries, where just the order of search is determine |
|
1254 | 1267 | d by profile/prefix. |
|
1255 | 1268 | All the functionality would just be "pushed" to ipython core, i.e. |
|
1256 | 1269 | the objects that represent the functionality are instantiated on "plugins" |
|
1257 | 1270 | and they are registered with ipython core. |
|
1258 | 1271 | i.e. |
|
1259 | 1272 | \layout Standard |
|
1260 | 1273 | |
|
1261 | 1274 | def magic_f(options, args): pass |
|
1262 | 1275 | \layout Standard |
|
1263 | 1276 | |
|
1264 | 1277 | m = MyMagic(magic_f) m.arghandler = stockhandlers.OptParseArgHandler m.options |
|
1265 | 1278 | = .... |
|
1266 | 1279 | # optparse options, for easy passing to magic_f and help display |
|
1267 | 1280 | \layout Standard |
|
1268 | 1281 | |
|
1269 | 1282 | # note that arghandler takes a peek at the instance, sees options, and proceeds |
|
1270 | 1283 | # accordingly. |
|
1271 | 1284 | Various arg handlers can ask for arbitrary options. |
|
1272 | 1285 | # some handler might optionally glob the filenames, search data folders |
|
1273 | 1286 | for filenames etc. |
|
1274 | 1287 | \layout Standard |
|
1275 | 1288 | |
|
1276 | 1289 | ipythonregistry.register(category = "magic", name = "mymagic", obj = m) |
|
1277 | 1290 | \layout Standard |
|
1278 | 1291 | |
|
1279 | 1292 | I bet most of the current functionality could easily be added to such a |
|
1280 | 1293 | registry by just instantiating e.g. |
|
1281 | 1294 | "Magic" class and registering all the functions with some sensible default |
|
1282 | 1295 | args. |
|
1283 | 1296 | Supporting legacy stuff in general would be easy - just implement new handlers |
|
1284 | 1297 | (arg and otherwise) for new stuff, and have the old handlers around forever |
|
1285 | 1298 | / as long as is deemed appropriate. |
|
1286 | 1299 | The 'python' namespace (locals() + globals()) should be special, of course. |
|
1287 | 1300 | \layout Standard |
|
1288 | 1301 | |
|
1289 | 1302 | It should be easy to have arbitrary number of "categories" (like 'magic', |
|
1290 | 1303 | 'shellcommand','projectspecific_myproject', 'projectspecific_otherproject'). |
|
1291 | 1304 | It would only influence the order in which the completions are suggested, |
|
1292 | 1305 | and in case of name collision which one is selected. |
|
1293 | 1306 | Also, I think all completions should be shown, even the ones in "later" |
|
1294 | 1307 | categories in the case of a match in an "earlier" category. |
|
1295 | 1308 | \layout Standard |
|
1296 | 1309 | |
|
1297 | 1310 | The "functionality object" might also have a callable object 'expandarg', |
|
1298 | 1311 | and ipython would run it (with the arg index) when tab completion is attempted |
|
1299 | 1312 | after typing the function name. |
|
1300 | 1313 | It would return the possible completions for that particular command... |
|
1301 | 1314 | or None to "revert to default file completions". |
|
1302 | 1315 | Such functionality could be useful in making ipython an "operating console" |
|
1303 | 1316 | of a sort. |
|
1304 | 1317 | I'm talking about: |
|
1305 | 1318 | \layout Standard |
|
1306 | 1319 | |
|
1307 | 1320 | >> lscat reactor # list commands in category - reactor is "project specific" |
|
1308 | 1321 | category |
|
1309 | 1322 | \layout Standard |
|
1310 | 1323 | |
|
1311 | 1324 | r_operate |
|
1312 | 1325 | \layout Standard |
|
1313 | 1326 | |
|
1314 | 1327 | >> r_operate <tab> start shutdown notify_meltdown evacuate |
|
1315 | 1328 | \layout Standard |
|
1316 | 1329 | |
|
1317 | 1330 | >> r_operate shutdown <tab> |
|
1318 | 1331 | \layout Standard |
|
1319 | 1332 | |
|
1320 | 1333 | 1 2 5 6 # note that 3 and 4 are already shut down |
|
1321 | 1334 | \layout Standard |
|
1322 | 1335 | |
|
1323 | 1336 | >> r_operate shutdown 2 |
|
1324 | 1337 | \layout Standard |
|
1325 | 1338 | |
|
1326 | 1339 | Shutting down.. |
|
1327 | 1340 | ok. |
|
1328 | 1341 | \layout Standard |
|
1329 | 1342 | |
|
1330 | 1343 | >> r_operate start <tab> |
|
1331 | 1344 | \layout Standard |
|
1332 | 1345 | |
|
1333 | 1346 | 2 3 4 # 2 was shut down, can be started now |
|
1334 | 1347 | \layout Standard |
|
1335 | 1348 | |
|
1336 | 1349 | >> r_operate start 2 |
|
1337 | 1350 | \layout Standard |
|
1338 | 1351 | |
|
1339 | 1352 | Starting.... |
|
1340 | 1353 | ok. |
|
1341 | 1354 | \layout Standard |
|
1342 | 1355 | |
|
1343 | 1356 | I'm talking about having a super-configurable man-machine language here! |
|
1344 | 1357 | Like cmd.Cmd on steroids, as a free addition to ipython! |
|
1345 | 1358 | \the_end |
General Comments 0
You need to be logged in to leave comments.
Login now