Show More
@@ -5,9 +5,6 b'' | |||
|
5 | 5 | IPython 8.0 |
|
6 | 6 | ----------- |
|
7 | 7 | |
|
8 | IPython 8.0 is still in alpha/beta stage. Please help us improve those release notes | |
|
9 | by sending PRs that modify docs/source/whatsnew/version8.rst | |
|
10 | ||
|
11 | 8 | IPython 8.0 is bringing a large number of new features and improvements to both the |
|
12 | 9 | user of the terminal and of the kernel via Jupyter. The removal of compatibility |
|
13 | 10 | with older version of Python is also the opportunity to do a couple of |
@@ -15,24 +12,24 b' performance improvement in particular with respect to startup time.' | |||
|
15 | 12 | The 8.x branch started diverging from its predecessor around IPython 7.12 |
|
16 | 13 | (January 2020). |
|
17 | 14 | |
|
18 |
This release contains 250+ |
|
|
15 | This release contains 250+ pull requests, in addition to many of the features | |
|
19 | 16 | and backports that have made it to the 7.x branch. All PRs that went into this |
|
20 | 17 | released are properly tagged with the 8.0 milestone if you wish to have a more |
|
21 | 18 | in depth look at the changes. |
|
22 | 19 | |
|
23 |
Please fell free to send pull |
|
|
20 | Please fell free to send pull requests to updates those notes after release, | |
|
24 | 21 | I have likely forgotten a few things reviewing 250+ PRs. |
|
25 | 22 | |
|
26 | 23 | Dependencies changes/downstream packaging |
|
27 | 24 | ----------------------------------------- |
|
28 | 25 | |
|
29 |
|
|
|
30 |
and follow PEP 517 |
|
|
26 | Most of our building steps have been changed to be (mostly) declarative | |
|
27 | and follow PEP 517. We are trying to completely remove ``setup.py`` (:ghpull:`13238`) and are | |
|
31 | 28 | looking for help to do so. |
|
32 | 29 | |
|
33 |
- |
|
|
30 | - minimum supported ``traitlets`` version is now 5+ | |
|
34 | 31 | - we now require ``stack_data`` |
|
35 |
- |
|
|
32 | - minimal Python is now 3.8 | |
|
36 | 33 | - ``nose`` is not a testing requirement anymore |
|
37 | 34 | - ``pytest`` replaces nose. |
|
38 | 35 | - ``iptest``/``iptest3`` cli entrypoints do not exists anymore. |
@@ -44,24 +41,24 b' Deprecation and removal' | |||
|
44 | 41 | ----------------------- |
|
45 | 42 | |
|
46 | 43 | We removed almost all features, arguments, functions, and modules that were |
|
47 | marked as deprecated between IPython 1.0 and 5.0. As reminder 5.0 was released | |
|
48 |
in 2016, and 1.0 in 2013. Last release of the 5 branch was 5.10.0, in |
|
|
44 | marked as deprecated between IPython 1.0 and 5.0. As a reminder, 5.0 was released | |
|
45 | in 2016, and 1.0 in 2013. Last release of the 5 branch was 5.10.0, in May 2020. | |
|
49 | 46 | The few remaining deprecated features we left have better deprecation warnings |
|
50 | 47 | or have been turned into explicit errors for better error messages. |
|
51 | 48 | |
|
52 | 49 | I will use this occasion to add the following requests to anyone emitting a |
|
53 | 50 | deprecation warning: |
|
54 | 51 | |
|
55 |
- Please |
|
|
52 | - Please use at least ``stacklevel=2`` so that the warning is emitted into the | |
|
56 | 53 | caller context, and not the callee one. |
|
57 | 54 | - Please add **since which version** something is deprecated. |
|
58 | 55 | |
|
59 |
As a side note it is much easier to |
|
|
60 |
numbers than ``try/except`` when |
|
|
56 | As a side note, it is much easier to conditionally compare version | |
|
57 | numbers rather than using ``try/except`` when functionality changes with a version. | |
|
61 | 58 | |
|
62 | 59 | I won't list all the removed features here, but modules like ``IPython.kernel``, |
|
63 | which was just a shim module around ``ipykernel`` for the past 8 years have been | |
|
64 | remove, and so many other similar things that pre-date the name **Jupyter** | |
|
60 | which was just a shim module around ``ipykernel`` for the past 8 years, have been | |
|
61 | removed, and so many other similar things that pre-date the name **Jupyter** | |
|
65 | 62 | itself. |
|
66 | 63 | |
|
67 | 64 | We no longer need to add ``IPyhton.extensions`` to the PYTHONPATH because that is being |
@@ -74,15 +71,15 b' other packages and no longer need to be inside IPython.' | |||
|
74 | 71 | Documentation |
|
75 | 72 | ------------- |
|
76 | 73 | |
|
77 |
|
|
|
78 |
the experimental `VΓ©lin <https://pypi.org/project/velin/>`_ project |
|
|
74 | The majority of our docstrings have now been reformatted and automatically fixed by | |
|
75 | the experimental `VΓ©lin <https://pypi.org/project/velin/>`_ project to conform | |
|
79 | 76 | to numpydoc. |
|
80 | 77 | |
|
81 | 78 | Type annotations |
|
82 | 79 | ---------------- |
|
83 | 80 | |
|
84 | 81 | While IPython itself is highly dynamic and can't be completely typed, many of |
|
85 |
the function now have type annotation, and part of the codebase |
|
|
82 | the functions now have type annotations, and part of the codebase is now checked | |
|
86 | 83 | by mypy. |
|
87 | 84 | |
|
88 | 85 | |
@@ -92,9 +89,9 b' Featured changes' | |||
|
92 | 89 | Here is a features list of changes in IPython 8.0. This is of course non-exhaustive. |
|
93 | 90 | Please note as well that many features have been added in the 7.x branch as well |
|
94 | 91 | (and hence why you want to read the 7.x what's new notes), in particular |
|
95 |
features contributed by QuantStack (with respect to debugger protocol |
|
|
96 | Python), as well as many debugger features that I was please to implement as | |
|
97 |
part of my work at QuanSight and |
|
|
92 | features contributed by QuantStack (with respect to debugger protocol and Xeus | |
|
93 | Python), as well as many debugger features that I was pleased to implement as | |
|
94 | part of my work at QuanSight and sponsored by DE Shaw. | |
|
98 | 95 | |
|
99 | 96 | Traceback improvements |
|
100 | 97 | ~~~~~~~~~~~~~~~~~~~~~~ |
@@ -137,9 +134,8 b' The error traceback is now correctly formatted, showing the cell number in which' | |||
|
137 | 134 | |
|
138 | 135 | ZeroDivisionError: division by zero |
|
139 | 136 | |
|
140 | The Second on is the integration of the ``stack_data`` package; | |
|
141 | which provide smarter informations in traceback; in particular it will highlight | |
|
142 | the AST node where an error occurs which can help to quickly narrow down errors. | |
|
137 | The ``stack_data`` package has been integrated, which provides smarter information in the traceback; | |
|
138 | in particular it will highlight the AST node where an error occurs which can help to quickly narrow down errors. | |
|
143 | 139 | |
|
144 | 140 | For example in the following snippet:: |
|
145 | 141 | |
@@ -154,8 +150,8 b' For example in the following snippet::' | |||
|
154 | 150 | ) + foo(2) |
|
155 | 151 | |
|
156 | 152 | |
|
157 |
|
|
|
158 |
IPython 8.0 is capable of telling you |
|
|
153 | calling ``bar()`` would raise an ``IndexError`` on the return line of ``foo``, | |
|
154 | and IPython 8.0 is capable of telling you where the index error occurs:: | |
|
159 | 155 | |
|
160 | 156 | |
|
161 | 157 | IndexError |
@@ -178,11 +174,10 b' IPython 8.0 is capable of telling you, where the index error occurs::' | |||
|
178 | 174 | ----> 3 return x[0][i][0] |
|
179 | 175 | ^^^^^^^ |
|
180 | 176 | |
|
181 |
|
|
|
182 | terminal and notebooks. | |
|
177 | The corresponding locations marked here with ``^`` will show up highlighted in | |
|
178 | the terminal and notebooks. | |
|
183 | 179 | |
|
184 | The Third, which is the most discreet but can have a high impact on | |
|
185 | productivity, a colon ``::`` and line number is appended after a filename in | |
|
180 | Finally, a colon ``::`` and line number is appended after a filename in | |
|
186 | 181 | traceback:: |
|
187 | 182 | |
|
188 | 183 | |
@@ -196,8 +191,9 b' traceback::' | |||
|
196 | 191 | 1 def f(): |
|
197 | 192 | ----> 2 1/0 |
|
198 | 193 | |
|
199 |
Many terminal and editor have integrations |
|
|
200 |
relevant file/line when this syntax is used |
|
|
194 | Many terminals and editors have integrations enabling you to directly jump to the | |
|
195 | relevant file/line when this syntax is used, so this small addition may have a high | |
|
196 | impact on productivity. | |
|
201 | 197 | |
|
202 | 198 | |
|
203 | 199 | Autosuggestons |
@@ -274,7 +270,7 b' Show pinfo information in ipdb using "?" and "??"' | |||
|
274 | 270 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
|
275 | 271 | |
|
276 | 272 | In IPDB, it is now possible to show the information about an object using "?" |
|
277 | and "??", in much the same way it can be done when using the IPython prompt:: | |
|
273 | and "??", in much the same way that it can be done when using the IPython prompt:: | |
|
278 | 274 | |
|
279 | 275 | ipdb> partial? |
|
280 | 276 | Init signature: partial(self, /, *args, **kwargs) |
@@ -291,14 +287,14 b' Previously, ``pinfo`` or ``pinfo2`` command had to be used for this purpose.' | |||
|
291 | 287 | Autoreload 3 feature |
|
292 | 288 | ~~~~~~~~~~~~~~~~~~~~ |
|
293 | 289 | |
|
294 |
Example: When an IPython session is r |
|
|
295 | you will now have the option '3' to select which means the following: | |
|
290 | Example: When an IPython session is run with the 'autoreload' extension loaded, | |
|
291 | you will now have the option '3' to select, which means the following: | |
|
296 | 292 | |
|
297 | 293 | 1. replicate all functionality from option 2 |
|
298 | 294 | 2. autoload all new funcs/classes/enums/globals from the module when they are added |
|
299 | 295 | 3. autoload all newly imported funcs/classes/enums/globals from external modules |
|
300 | 296 | |
|
301 | Try ``%autoreload 3`` in an IPython session after running ``%load_ext autoreload`` | |
|
297 | Try ``%autoreload 3`` in an IPython session after running ``%load_ext autoreload``. | |
|
302 | 298 | |
|
303 | 299 | For more information please see the following unit test : ``extensions/tests/test_autoreload.py:test_autoload_newly_added_objects`` |
|
304 | 300 | |
@@ -309,7 +305,7 b' If ``black`` is installed in the same environment as IPython, terminal IPython' | |||
|
309 | 305 | will now *by default* reformat the code in the CLI when possible. You can |
|
310 | 306 | disable this with ``--TerminalInteractiveShell.autoformatter=None``. |
|
311 | 307 | |
|
312 | This feature was present in 7.x but disabled by default. | |
|
308 | This feature was present in 7.x, but disabled by default. | |
|
313 | 309 | |
|
314 | 310 | |
|
315 | 311 | History Range Glob feature |
@@ -336,10 +332,10 b' then the glob pattern would be used (globbing *all* history) *and the range woul' | |||
|
336 | 332 | |
|
337 | 333 | With this enhancement, if a user specifies both a range and a glob pattern, then the glob pattern will be applied to the specified range of history. |
|
338 | 334 | |
|
339 |
Don't start a multi |
|
|
335 | Don't start a multi-line cell with sunken parenthesis | |
|
340 | 336 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
|
341 | 337 | |
|
342 | From now on IPython will not ask for the next line of input when given a single | |
|
338 | From now on, IPython will not ask for the next line of input when given a single | |
|
343 | 339 | line with more closing than opening brackets. For example, this means that if |
|
344 | 340 | you (mis)type ``]]`` instead of ``[]``, a ``SyntaxError`` will show up, instead of |
|
345 | 341 | the ``...:`` prompt continuation. |
@@ -394,89 +390,85 b' Using them this way will make them take the history of the current session up' | |||
|
394 | 390 | to the point of the magic call (such that the magic itself will not be |
|
395 | 391 | included). |
|
396 | 392 | |
|
397 |
Therefore it is now possible to save the whole history to a file using |
|
|
393 | Therefore it is now possible to save the whole history to a file using | |
|
398 | 394 | ``%save <filename>``, load and edit it using ``%load`` (makes for a nice usage |
|
399 | 395 | when followed with :kbd:`F2`), send it to `dpaste.org <http://dpast.org>`_ using |
|
400 | 396 | ``%pastebin``, or view the whole thing syntax-highlighted with a single |
|
401 | 397 | ``%pycat``. |
|
402 | 398 | |
|
403 | 399 | |
|
404 |
Windows tim |
|
|
405 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
|
406 |
Timing for example with ``%%time`` |
|
|
407 | This is at the end the same as W-All. | |
|
408 | To be a bit tighter to linux one could change to ``time.process_time`` instead. | |
|
409 | Thus for example one would no longer count periods of sleep and further. | |
|
410 | ||
|
400 | Windows timing implementation: Switch to process_time | |
|
401 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
|
402 | Timing on Windows, for example with ``%%time``, was changed from being based on ``time.perf_counter`` | |
|
403 | (which counted time even when the process was sleeping) to being based on ``time.process_time`` instead | |
|
404 | (which only counts CPU time). This brings it closer to the behavior on Linux. See :ghpull:`12984`. | |
|
411 | 405 | |
|
412 | 406 | Miscellaneous |
|
413 | 407 | ~~~~~~~~~~~~~ |
|
414 | - Non-text formatters are not disabled in terminal which should simplify | |
|
415 | writing extension displaying images or other mimetypes supporting terminals. | |
|
408 | - Non-text formatters are not disabled in the terminal, which should simplify | |
|
409 | writing extensions displaying images or other mimetypes in supporting terminals. | |
|
416 | 410 | :ghpull:`12315` |
|
417 | - | |
|
418 | 411 | - It is now possible to automatically insert matching brackets in Terminal IPython using the |
|
419 | 412 | ``TerminalInteractiveShell.auto_match=True`` option. :ghpull:`12586` |
|
420 | - We are thinking of deprecating the current ``%%javascript`` magic in favor of a better replacement. See :ghpull:`13376` | |
|
421 | - ``%time`` uses ``process_time`` instead of ``perf_counter``, see :ghpull:`12984` | |
|
413 | - We are thinking of deprecating the current ``%%javascript`` magic in favor of a better replacement. See :ghpull:`13376`. | |
|
422 | 414 | - ``~`` is now expanded when part of a path in most magics :ghpull:`13385` |
|
423 | - ``%/%%timeit`` magic now adds comma every thousands to make reading long number easier :ghpull:`13379` | |
|
415 | - ``%/%%timeit`` magic now adds a comma every thousands to make reading a long number easier :ghpull:`13379` | |
|
424 | 416 | - ``"info"`` messages can now be customised to hide some fields :ghpull:`13343` |
|
425 | 417 | - ``collections.UserList`` now pretty-prints :ghpull:`13320` |
|
426 |
- The debugger now ha |
|
|
418 | - The debugger now has a persistent history, which should make it less | |
|
427 | 419 | annoying to retype commands :ghpull:`13246` |
|
428 |
- ``!pip`` ``!conda`` ``!cd`` or ``!ls`` are likely doing the wrong thing |
|
|
429 |
now warn users if they use |
|
|
430 |
- |
|
|
420 | - ``!pip`` ``!conda`` ``!cd`` or ``!ls`` are likely doing the wrong thing. We | |
|
421 | now warn users if they use one of those commands. :ghpull:`12954` | |
|
422 | - Make ``%precision`` work for ``numpy.float64`` type :ghpull:`12902` | |
|
431 | 423 | |
|
432 | 424 | Re-added support for XDG config directories |
|
433 | 425 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
|
434 | 426 | |
|
435 |
XDG support through the years |
|
|
436 |
identical location in all platforms |
|
|
437 | failure a couple of years ago IPython was modified to automatically migrate XDG | |
|
438 |
config files back into ``~/.ipython`` |
|
|
439 |
|
|
|
427 | XDG support through the years comes and goes. There is a tension between having | |
|
428 | an identical location for configuration in all platforms versus having simple instructions. | |
|
429 | After initial failures a couple of years ago, IPython was modified to automatically migrate XDG | |
|
430 | config files back into ``~/.ipython``. That migration code has now been removed. | |
|
431 | IPython now checks the XDG locations, so if you _manually_ move your config | |
|
440 | 432 | files to your preferred location, IPython will not move them back. |
|
441 | 433 | |
|
442 | 434 | |
|
443 | Numfocus Small Developer Grant | |
|
444 |
------------------------- |
|
|
435 | Preparing for Python 3.10 | |
|
436 | ------------------------- | |
|
445 | 437 | |
|
446 |
To prepare for Python 3.10 we have |
|
|
447 |
any dependency that is not |
|
|
448 |
test suite to pytest |
|
|
449 |
``iptest`` command is now gone |
|
|
438 | To prepare for Python 3.10, we have started working on removing reliance and | |
|
439 | any dependency that is not compatible with Python 3.10. This includes migrating our | |
|
440 | test suite to pytest and starting to remove nose. This also means that the | |
|
441 | ``iptest`` command is now gone and all testing is via pytest. | |
|
450 | 442 | |
|
451 | 443 | This was in large part thanks to the NumFOCUS Small Developer grant, which enabled us to |
|
452 | 444 | allocate \$4000 to hire `Nikita Kniazev (@Kojoley) <https://github.com/Kojoley>`_, |
|
453 | 445 | who did a fantastic job at updating our code base, migrating to pytest, pushing |
|
454 | 446 | our coverage, and fixing a large number of bugs. I highly recommend contacting |
|
455 | them if you need help with C++ and Python projects | |
|
447 | them if you need help with C++ and Python projects. | |
|
456 | 448 | |
|
457 | 449 | You can find all relevant issues and PRs with the SDG 2021 tag `<https://github.com/ipython/ipython/issues?q=label%3A%22Numfocus+SDG+2021%22+>`__ |
|
458 | 450 | |
|
459 |
Removing support for |
|
|
460 | --------------------------------- | |
|
451 | Removing support for older Python versions | |
|
452 | ------------------------------------------ | |
|
461 | 453 | |
|
462 | 454 | |
|
463 |
We are |
|
|
464 |
efficient ``pathlib`` |
|
|
455 | We are removing support for Python up through 3.7, allowing internal code to use the more | |
|
456 | efficient ``pathlib`` and to make better use of type annotations. | |
|
465 | 457 | |
|
466 | 458 | .. image:: ../_images/8.0/pathlib_pathlib_everywhere.jpg |
|
467 | 459 | :alt: "Meme image of Toy Story with Woody and Buzz, with the text 'pathlib, pathlib everywhere'" |
|
468 | 460 | |
|
469 | 461 | |
|
470 |
We ha |
|
|
462 | We had about 34 PRs only to update some logic to update some functions from managing strings to | |
|
471 | 463 | using Pathlib. |
|
472 | 464 | |
|
473 | The completer has also seen significant updates and make use of newer Jedi API | |
|
465 | The completer has also seen significant updates and now makes use of newer Jedi APIs, | |
|
474 | 466 | offering faster and more reliable tab completion. |
|
475 | 467 | |
|
476 | 468 | Misc Statistics |
|
477 | 469 | --------------- |
|
478 | 470 | |
|
479 | Here are some numbers: | |
|
471 | Here are some numbers:: | |
|
480 | 472 |
|
|
481 | 473 | 7.x: 296 files, 12561 blank lines, 20282 comments, 35142 line of code. |
|
482 | 474 | 8.0: 252 files, 12053 blank lines, 19232 comments, 34505 line of code. |
@@ -484,8 +476,8 b' Here are some numbers:' | |||
|
484 | 476 |
|
|
485 | 477 |
|
|
486 | 478 |
|
|
487 | We have commits from 162 authors, who contributed 1916 commits in 23 month, excluding merges to not bias toward | |
|
488 | maintainers pushing buttons.:: | |
|
479 | We have commits from 162 authors, who contributed 1916 commits in 23 month, excluding merges (to not bias toward | |
|
480 | maintainers pushing buttons).:: | |
|
489 | 481 | |
|
490 | 482 | $ git shortlog -s --no-merges 7.x...master | sort -nr |
|
491 | 483 | 535 Matthias Bussonnier |
@@ -649,7 +641,7 b' maintainers pushing buttons.::' | |||
|
649 | 641 | 1 Albert Zhang |
|
650 | 642 | 1 Adam Johnson |
|
651 | 643 | |
|
652 | This does not of course represent non-code contributions. | |
|
644 | This does not, of course, represent non-code contributions, for which we are also grateful. | |
|
653 | 645 | |
|
654 | 646 | |
|
655 | 647 | API Changes using Frappuccino |
General Comments 0
You need to be logged in to leave comments.
Login now