##// END OF EJS Templates
exchangev2: fetch file revisions...
exchangev2: fetch file revisions Now that the server has an API for fetching file data, we can call into it to fetch file revisions. The implementation is relatively straightforward: we examine the manifests that we fetched and find all new file revisions referenced by them. We build up a mapping from file path to file nodes to manifest node. (The mapping to first manifest node allows us to map back to first changelog node/revision, which is used for the linkrev.) Once that map is built up, we iterate over it in a deterministic manner and fetch and store file data. The code is very similar to manifest fetching. So similar that we could probably extract the common bits into a generic function. With file data retrieval implemented, `hg clone` and `hg pull` are effectively feature complete, at least as far as the completeness of data transfer for essential repository data (changesets, manifests, files, phases, and bookmarks). We're still missing support for obsolescence markers, the hgtags fnodes cache, and the branchmap cache. But these are non-essential for the moment (and will be implemented later). This is a good point to assess the state of exchangev2 in terms of performance. I ran a local `hg clone` for the mozilla-unified repository using both version 1 and version 2 of the wire protocols and exchange methods. This is effectively comparing the performance of the wire protocol overhead and "getbundle" versus domain-specific commands. Wire protocol version 2 doesn't have compression implemented yet. So I tested version 1 with `server.compressionengines=none` to remove compression overhead from the equation. server before: user 220.420+0.000 sys 14.420+0.000 after: user 321.980+0.000 sys 18.990+0.000 client before: real 561.650 secs (user 497.670+0.000 sys 28.160+0.000) after: real 1226.260 secs (user 944.240+0.000 sys 354.150+0.000) We have substantial regressions on both client and server. This is obviously not desirable. I'm aware of some reasons: * Lack of hgtagsfnodes transfer (contributes significant CPU to client). * Lack of branch cache transfer (contributes significant CPU to client). * Little to no profiling / optimization performed on wire protocol version 2 code. * There appears to be a memory leak on the client and that is likely causing swapping on my machine. * Using multiple threads on the client may be counter-productive because Python. * We're not compressing on the server. * We're tracking file nodes on the client via manifest diffing rather than using linkrev shortcuts on the server. I'm pretty confident that most of these issues are addressable. But even if we can't get wire protocol version 2 on performance parity with "getbundle," I still think it is important to have the set of low level data-specific retrieval commands that we have implemented so far. This is because the existence of such commands allows flexibility in how clients access server data. Differential Revision: https://phab.mercurial-scm.org/D4491

File last commit:

r37195:68ee6182 default
r39676:039bf1ed default
Show More
advice.py
207 lines | 7.3 KiB | text/x-python | PythonLexer
##############################################################################
#
# Copyright (c) 2003 Zope Foundation and Contributors.
# All Rights Reserved.
#
# This software is subject to the provisions of the Zope Public License,
# Version 2.1 (ZPL). A copy of the ZPL should accompany this distribution.
# THIS SOFTWARE IS PROVIDED "AS IS" AND ANY AND ALL EXPRESS OR IMPLIED
# WARRANTIES ARE DISCLAIMED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
# WARRANTIES OF TITLE, MERCHANTABILITY, AGAINST INFRINGEMENT, AND FITNESS
# FOR A PARTICULAR PURPOSE.
#
##############################################################################
"""Class advice.
This module was adapted from 'protocols.advice', part of the Python
Enterprise Application Kit (PEAK). Please notify the PEAK authors
(pje@telecommunity.com and tsarna@sarna.org) if bugs are found or
Zope-specific changes are required, so that the PEAK version of this module
can be kept in sync.
PEAK is a Python application framework that interoperates with (but does
not require) Zope 3 and Twisted. It provides tools for manipulating UML
models, object-relational persistence, aspect-oriented programming, and more.
Visit the PEAK home page at http://peak.telecommunity.com for more information.
"""
from __future__ import absolute_import
from types import FunctionType
try:
from types import ClassType
except ImportError:
__python3 = True
else:
__python3 = False
import sys
def getFrameInfo(frame):
"""Return (kind,module,locals,globals) for a frame
'kind' is one of "exec", "module", "class", "function call", or "unknown".
"""
f_locals = frame.f_locals
f_globals = frame.f_globals
sameNamespace = f_locals is f_globals
hasModule = '__module__' in f_locals
hasName = '__name__' in f_globals
sameName = hasModule and hasName
sameName = sameName and f_globals['__name__']==f_locals['__module__']
module = hasName and sys.modules.get(f_globals['__name__']) or None
namespaceIsModule = module and module.__dict__ is f_globals
if not namespaceIsModule:
# some kind of funky exec
kind = "exec"
elif sameNamespace and not hasModule:
kind = "module"
elif sameName and not sameNamespace:
kind = "class"
elif not sameNamespace:
kind = "function call"
else: # pragma: no cover
# How can you have f_locals is f_globals, and have '__module__' set?
# This is probably module-level code, but with a '__module__' variable.
kind = "unknown"
return kind, module, f_locals, f_globals
def addClassAdvisor(callback, depth=2):
"""Set up 'callback' to be passed the containing class upon creation
This function is designed to be called by an "advising" function executed
in a class suite. The "advising" function supplies a callback that it
wishes to have executed when the containing class is created. The
callback will be given one argument: the newly created containing class.
The return value of the callback will be used in place of the class, so
the callback should return the input if it does not wish to replace the
class.
The optional 'depth' argument to this function determines the number of
frames between this function and the targeted class suite. 'depth'
defaults to 2, since this skips this function's frame and one calling
function frame. If you use this function from a function called directly
in the class suite, the default will be correct, otherwise you will need
to determine the correct depth yourself.
This function works by installing a special class factory function in
place of the '__metaclass__' of the containing class. Therefore, only
callbacks *after* the last '__metaclass__' assignment in the containing
class will be executed. Be sure that classes using "advising" functions
declare any '__metaclass__' *first*, to ensure all callbacks are run."""
# This entire approach is invalid under Py3K. Don't even try to fix
# the coverage for this block there. :(
if __python3: # pragma: no cover
raise TypeError('Class advice impossible in Python3')
frame = sys._getframe(depth)
kind, module, caller_locals, caller_globals = getFrameInfo(frame)
# This causes a problem when zope interfaces are used from doctest.
# In these cases, kind == "exec".
#
#if kind != "class":
# raise SyntaxError(
# "Advice must be in the body of a class statement"
# )
previousMetaclass = caller_locals.get('__metaclass__')
if __python3: # pragma: no cover
defaultMetaclass = caller_globals.get('__metaclass__', type)
else:
defaultMetaclass = caller_globals.get('__metaclass__', ClassType)
def advise(name, bases, cdict):
if '__metaclass__' in cdict:
del cdict['__metaclass__']
if previousMetaclass is None:
if bases:
# find best metaclass or use global __metaclass__ if no bases
meta = determineMetaclass(bases)
else:
meta = defaultMetaclass
elif isClassAdvisor(previousMetaclass):
# special case: we can't compute the "true" metaclass here,
# so we need to invoke the previous metaclass and let it
# figure it out for us (and apply its own advice in the process)
meta = previousMetaclass
else:
meta = determineMetaclass(bases, previousMetaclass)
newClass = meta(name,bases,cdict)
# this lets the callback replace the class completely, if it wants to
return callback(newClass)
# introspection data only, not used by inner function
advise.previousMetaclass = previousMetaclass
advise.callback = callback
# install the advisor
caller_locals['__metaclass__'] = advise
def isClassAdvisor(ob):
"""True if 'ob' is a class advisor function"""
return isinstance(ob,FunctionType) and hasattr(ob,'previousMetaclass')
def determineMetaclass(bases, explicit_mc=None):
"""Determine metaclass from 1+ bases and optional explicit __metaclass__"""
meta = [getattr(b,'__class__',type(b)) for b in bases]
if explicit_mc is not None:
# The explicit metaclass needs to be verified for compatibility
# as well, and allowed to resolve the incompatible bases, if any
meta.append(explicit_mc)
if len(meta)==1:
# easy case
return meta[0]
candidates = minimalBases(meta) # minimal set of metaclasses
if not candidates: # pragma: no cover
# they're all "classic" classes
assert(not __python3) # This should not happen under Python 3
return ClassType
elif len(candidates)>1:
# We could auto-combine, but for now we won't...
raise TypeError("Incompatible metatypes",bases)
# Just one, return it
return candidates[0]
def minimalBases(classes):
"""Reduce a list of base classes to its ordered minimum equivalent"""
if not __python3: # pragma: no cover
classes = [c for c in classes if c is not ClassType]
candidates = []
for m in classes:
for n in classes:
if issubclass(n,m) and m is not n:
break
else:
# m has no subclasses in 'classes'
if m in candidates:
candidates.remove(m) # ensure that we're later in the list
candidates.append(m)
return candidates