Lars Gustäbel [Mon, 14 May 2012 11:18:16 +0000 (13:18 +0200)]
Issue #13815: Resurrect the ExFileObject class.
After a discussion in the tracker, the decision was made to keep the
ExFileObject class after all as a subclass of io.BufferedReader instead of
removing it completely.
Brian Curtin [Sun, 13 May 2012 21:16:09 +0000 (16:16 -0500)]
changeset: 76969:0cbe1099226d
branch: vs2010
tag: tip
user: Brian Curtin <brian@python.org>
date: Sun May 13 16:15:11 2012 -0500
summary: Changes to allow Profile Guided Optimization builds to succeed on VS2010
Brett Cannon [Sun, 13 May 2012 17:04:21 +0000 (13:04 -0400)]
Issue #13959: Document imp.find_module/load_module as deprecated.
The code itself does not raise a DeprecationWarning as the functions
are technically fine, it's just a bad API. Unfortunately experience
has shown that the terrible API has been exposed in various places,
necessitating that it stick around probably until py4k comes around
since it is such a shift to move over to importlib.find_loader().
Brett Cannon [Sat, 12 May 2012 21:43:17 +0000 (17:43 -0400)]
Issue #13959: Introduce importlib.find_loader().
The long-term goal is to deprecate imp.find_module() in favour of this
API, but it will take some time as some APIs explicitly return/use what
imp.find_module() returns.
Stefan Krah [Sat, 12 May 2012 21:11:51 +0000 (23:11 +0200)]
Issue #14779: Do not use get_config_var('SIZEOF_VOID_P') on OS X 64-/32-bit
universal: it returns a meaningless result. Use sys.maxsize instead of
platform.architecture as a fallback. Patch by Ned Deily.
Brett Cannon [Fri, 11 May 2012 18:48:41 +0000 (14:48 -0400)]
Issue #13959: Have
importlib.abc.FileLoader.load_module()/get_filename() and
importlib.machinery.ExtensionFileLoader.load_module() have their
single argument be optional as the loader's constructor has all the
ncessary information.
This allows for the deprecation of
imp.load_source()/load_compile()/load_package().
Brett Cannon [Fri, 11 May 2012 16:58:42 +0000 (12:58 -0400)]
Issue #13959: Deprecate imp.get_suffixes() for new attributes on
importlib.machinery that provide the suffix details for import.
The attributes were not put on imp so as to compartmentalize
everything importlib needs for setting up imports in
importlib.machinery.
This also led to an indirect deprecation of inspect.getmoduleinfo() as
it directly returned imp.get_suffix's returned tuple which no longer
makes sense.
Richard Oudkerk [Thu, 10 May 2012 15:11:12 +0000 (16:11 +0100)]
Issue #14753: Make multiprocessing treat negative timeouts as it did in 3.2
In Python 3.2 and earlier, Process.join() and Connection.poll()
treated negative timeouts as zero timeouts. Earlier versions from
the 3.3 line of development treat them as infinite timeouts.
Victor Stinner [Mon, 7 May 2012 21:50:05 +0000 (23:50 +0200)]
Backout ab500b297900: the check for integer overflow is wrong
Issue #14716: Change integer overflow check in unicode_writer_prepare()
to compute the limit at compile time instead of runtime. Patch writen by Serhiy
Storchaka.
Victor Stinner [Mon, 7 May 2012 11:02:44 +0000 (13:02 +0200)]
Issue #14716: Change integer overflow check in unicode_writer_prepare()
to compute the limit at compile time instead of runtime. Patch writen by Serhiy
Storchaka.
Victor Stinner [Mon, 7 May 2012 10:47:02 +0000 (12:47 +0200)]
Close #14716: str.format() now uses the new "unicode writer" API instead of the
PyAccu API. For example, it makes str.format() from 25% to 30% faster on Linux.
Mark Dickinson [Mon, 7 May 2012 10:20:50 +0000 (11:20 +0100)]
Issue #14700: Fix two broken and undefined-behaviour-inducing overflow checks in old-style string formatting. Thanks Serhiy Storchaka for report and original patch.