]> granicus.if.org Git - python/log
python
12 years agoFix cross compiling issue in setup.py, ensure that lib_dirs and inc_dirs are
Christian Heimes [Wed, 12 Dec 2012 11:41:50 +0000 (12:41 +0100)]
Fix cross compiling issue in setup.py, ensure that lib_dirs and inc_dirs are
defined in cross compiling mode, too.

12 years agoFix cross compiling issue in setup.py, ensure that lib_dirs and inc_dirs are
Christian Heimes [Wed, 12 Dec 2012 11:41:00 +0000 (12:41 +0100)]
Fix cross compiling issue in setup.py, ensure that lib_dirs and inc_dirs are
defined in cross compiling mode, too.

12 years agoCode style fixup: No need for double ((parenthesis)) and use {} on an if else.
Gregory P. Smith [Tue, 11 Dec 2012 04:22:55 +0000 (20:22 -0800)]
Code style fixup: No need for double ((parenthesis)) and use {} on an if else.

12 years agoCode style fixup: No need for double ((parenthesis)) and use {} on an if else.
Gregory P. Smith [Tue, 11 Dec 2012 04:22:31 +0000 (20:22 -0800)]
Code style fixup: No need for double ((parenthesis)) and use {} on an if else.

12 years agoCode style fixup: No need for double ((parenthesis)) and use {} on an if else.
Gregory P. Smith [Tue, 11 Dec 2012 04:20:20 +0000 (20:20 -0800)]
Code style fixup: No need for double ((parenthesis)) and use {} on an if else.

12 years agoFix the internals of our hash functions to used unsigned values during hash
Gregory P. Smith [Tue, 11 Dec 2012 03:51:29 +0000 (19:51 -0800)]
Fix the internals of our hash functions to used unsigned values during hash
computation as the overflow behavior of signed integers is undefined.

NOTE: This change is smaller compared to 3.2 as much of this cleanup had
already been done.  I added the comment that my change in 3.2 added so that the
code would match up.  Otherwise this just adds or synchronizes appropriate UL
designations on some constants to be pedantic.

In practice we require compiling everything with -fwrapv which forces overflow
to be defined as twos compliment but this keeps the code cleaner for checkers
or in the case where someone has compiled it without -fwrapv or their
compiler's equivalent.  We could work to get rid of the -fwrapv requirement
in 3.4 but that requires more planning.

Found by Clang trunk's Undefined Behavior Sanitizer (UBSan).

Cleanup only - no functionality or hash values change.

12 years agonull merge, no change needed in 3.3.
Gregory P. Smith [Tue, 11 Dec 2012 02:34:29 +0000 (18:34 -0800)]
null merge, no change needed in 3.3.

12 years agoKeep y a Py_hash_t instead of Py_uhash_t as it is compared with == -1 and the
Gregory P. Smith [Tue, 11 Dec 2012 02:34:09 +0000 (18:34 -0800)]
Keep y a Py_hash_t instead of Py_uhash_t as it is compared with == -1 and the
compiler logic will do the right thing with just x as a Py_uhash_t.  This
matches what was already done in the 3.3 version.

cleanup only - no functionality or hash values change.

12 years agoFix the internals of our hash functions to used unsigned values during hash
Gregory P. Smith [Tue, 11 Dec 2012 02:32:53 +0000 (18:32 -0800)]
Fix the internals of our hash functions to used unsigned values during hash
computation as the overflow behavior of signed integers is undefined.

NOTE: This change is smaller compared to 3.2 as much of this cleanup had
already been done.  I added the comment that my change in 3.2 added so that the
code would match up.  Otherwise this just adds or synchronizes appropriate UL
designations on some constants to be pedantic.

In practice we require compiling everything with -fwrapv which forces overflow
to be defined as twos compliment but this keeps the code cleaner for checkers
or in the case where someone has compiled it without -fwrapv or their
compiler's equivalent.

Found by Clang trunk's Undefined Behavior Sanitizer (UBSan).

Cleanup only - no functionality or hash values change.

12 years agoFix the internals of our hash functions to used unsigned values during hash
Gregory P. Smith [Tue, 11 Dec 2012 02:15:46 +0000 (18:15 -0800)]
Fix the internals of our hash functions to used unsigned values during hash
computation as the overflow behavior of signed integers is undefined.

In practice we require compiling everything with -fwrapv which forces overflow
to be defined as twos compliment but this keeps the code cleaner for checkers
or in the case where someone has compiled it without -fwrapv or their
compiler's equivalent.

Found by Clang trunk's Undefined Behavior Sanitizer (UBSan).

Cleanup only - no functionality or hash values change.

12 years agoUsing 'long double' to force this structure to be worst case aligned is no
Gregory P. Smith [Tue, 11 Dec 2012 02:05:05 +0000 (18:05 -0800)]
Using 'long double' to force this structure to be worst case aligned is no
longer required as of Python 2.5+ when the gc_refs changed from an int (4
bytes) to a Py_ssize_t (8 bytes) as the minimum size is 16 bytes.

The use of a 'long double' triggered a warning by Clang trunk's
Undefined-Behavior Sanitizer as on many platforms a long double requires
16-byte alignment but the Python memory allocator only guarantees 8 byte
alignment.

So our code would allocate and use these structures with technically improper
alignment.  Though it didn't matter since the 'dummy' field is never used.
This silences that warning.

Spelunking into code history, the double was added in 2001 to force better
alignment on some platforms and changed to a long double in 2002 to appease
Tru64.  That issue should no loner be present since the upgrade from int to
Py_ssize_t where the minimum structure size increased to 16 (unless anyone
knows of a platform where ssize_t is 4 bytes?) or 24 bytes depending on if the
build uses 4 or 8 byte pointers.

We can probably get rid of the double and this union hack all together today.
That is a slightly more invasive change that can be left for later.

A more correct non-hacky alternative if any alignment issues are still found
would be to use a compiler specific alignment declaration on the structure and
determine which value to use at configure time.

12 years ago1 << 31 is invalid for signed integers, fix it by making 1 unsigned.
Gregory P. Smith [Tue, 11 Dec 2012 01:45:16 +0000 (17:45 -0800)]
1 << 31 is invalid for signed integers, fix it by making 1 unsigned.

Found by Clang trunk's Undefined-Behavior Sanitizer.  [more to come]

12 years ago1 << 31 is invalid for signed integers, fix it by making 1 unsigned.
Gregory P. Smith [Tue, 11 Dec 2012 01:45:03 +0000 (17:45 -0800)]
1 << 31 is invalid for signed integers, fix it by making 1 unsigned.

Found by Clang trunk's Undefined-Behavior Sanitizer.  [more to come]

12 years ago1 << 31 is invalid for signed integers, fix it by making 1 unsigned.
Gregory P. Smith [Tue, 11 Dec 2012 01:44:44 +0000 (17:44 -0800)]
1 << 31 is invalid for signed integers, fix it by making 1 unsigned.

Found by Clang trunk's Undefined-Behavior Sanitizer.  [more to come]

12 years ago#15872: Some more Windows related tuning to shutil.rmtree tests
Hynek Schlawack [Mon, 10 Dec 2012 15:35:16 +0000 (16:35 +0100)]
#15872: Some more Windows related tuning to shutil.rmtree tests

Turns out, the snakebite bots have also their peculiarities.

I'm really not proud of this stream of commits. :(

12 years ago#15872: Some more Windows related tuning to shutil.rmtree tests
Hynek Schlawack [Mon, 10 Dec 2012 15:33:41 +0000 (16:33 +0100)]
#15872: Some more Windows related tuning to shutil.rmtree tests

Turns out, the snakebite bots behave also their peculiarities.

I'm really not proud of this stream of commits. :(

12 years ago#15872: Some more Windows related tuning to shutil.rmtree tests
Hynek Schlawack [Mon, 10 Dec 2012 15:29:57 +0000 (16:29 +0100)]
#15872: Some more Windows related tuning to shutil.rmtree tests

Turns out, the snakebite bots behave also their peculiarities.

I'm really not proud of this stream of commits. :(

12 years ago#15872: Be flexible with appending *.* in shutil.rmtree test case
Hynek Schlawack [Mon, 10 Dec 2012 11:05:45 +0000 (12:05 +0100)]
#15872: Be flexible with appending *.* in shutil.rmtree test case

The Windows buildbots seem to be unable to agree whether they need them or not.

12 years ago#15872: Be flexible with appending *.* in shutil.rmtree test case
Hynek Schlawack [Mon, 10 Dec 2012 11:02:26 +0000 (12:02 +0100)]
#15872: Be flexible with appending *.* in shutil.rmtree test case

The Windows buildbots seem to be unable to agree whether they need them or not.

12 years ago#15872: Be flexible with appending *.* in shutil.rmtree test case
Hynek Schlawack [Mon, 10 Dec 2012 11:01:28 +0000 (12:01 +0100)]
#15872: Be flexible with appending *.* in shutil.rmtree test case

The Windows buildbots seem to be unable to agree whether they need them or not.

12 years ago#15872: More shutil test fixes for Windows
Hynek Schlawack [Mon, 10 Dec 2012 10:12:57 +0000 (11:12 +0100)]
#15872: More shutil test fixes for Windows

This one is different from 3.2 and 3.3.  Windows ceased using *.* since 3.4
apparently.

12 years agoFix Issue15701 : add .headers attribute to urllib.error.HTTPError
Senthil Kumaran [Mon, 10 Dec 2012 10:09:35 +0000 (02:09 -0800)]
Fix Issue15701 : add .headers attribute to urllib.error.HTTPError

12 years ago#15872: More shutil test fixes for Windows
Hynek Schlawack [Mon, 10 Dec 2012 10:08:59 +0000 (11:08 +0100)]
#15872: More shutil test fixes for Windows

12 years ago#15872: More shutil test fixes for Windows
Hynek Schlawack [Mon, 10 Dec 2012 10:08:09 +0000 (11:08 +0100)]
#15872: More shutil test fixes for Windows

12 years ago#15872: Fix shutil.rmtree error tests for Windows
Hynek Schlawack [Mon, 10 Dec 2012 09:10:40 +0000 (10:10 +0100)]
#15872: Fix shutil.rmtree error tests for Windows

12 years ago#15872: Fix shutil.rmtree error tests for Windows
Hynek Schlawack [Mon, 10 Dec 2012 09:08:41 +0000 (10:08 +0100)]
#15872: Fix shutil.rmtree error tests for Windows

12 years ago#15872: Fix shutil.rmtree error tests for Windows
Hynek Schlawack [Mon, 10 Dec 2012 09:07:11 +0000 (10:07 +0100)]
#15872: Fix shutil.rmtree error tests for Windows

12 years ago#15872: Fix 3.3 regression introduced by the new fd-based shutil.rmtree
Hynek Schlawack [Mon, 10 Dec 2012 08:15:23 +0000 (09:15 +0100)]
#15872: Fix 3.3 regression introduced by the new fd-based shutil.rmtree

It caused rmtree to not ignore certain errors when ignore_errors was set.

Patch by Alessandro Moura and Serhiy Storchaka.

12 years ago#15872: Fix 3.3 regression introduced by the new fd-based shutil.rmtree
Hynek Schlawack [Mon, 10 Dec 2012 08:11:25 +0000 (09:11 +0100)]
#15872: Fix 3.3 regression introduced by the new fd-based shutil.rmtree

It caused rmtree to not ignore certain errors when ignore_errors was set.

Patch by Alessandro Moura and Serhiy Storchaka.

12 years ago#15872: Add tests for a 3.3 regression in the new fd-based shutil.rmtree
Hynek Schlawack [Mon, 10 Dec 2012 08:00:09 +0000 (09:00 +0100)]
#15872: Add tests for a 3.3 regression in the new fd-based shutil.rmtree

It cause shutil.rmtree not ignore all errors. Also add a test ensuring that
rmtree fails when being called on a symlink. Patch by Serhiy Storchaka.

12 years agoIssue #16629: Merge IDLE test fix from 3.3.
Chris Jerdonek [Mon, 10 Dec 2012 02:22:21 +0000 (18:22 -0800)]
Issue #16629: Merge IDLE test fix from 3.3.

12 years agoIssue #16629: Merge IDLE test fix from 3.2.
Chris Jerdonek [Mon, 10 Dec 2012 02:19:54 +0000 (18:19 -0800)]
Issue #16629: Merge IDLE test fix from 3.2.

12 years agoIssue #16629: Fix IDLE idlelib.CallTips test. Patch by Roger Serwy.
Chris Jerdonek [Mon, 10 Dec 2012 02:17:27 +0000 (18:17 -0800)]
Issue #16629: Fix IDLE idlelib.CallTips test.  Patch by Roger Serwy.

This commit updates a test broken by the change made for issue #14783.

12 years agoIssue #16582: use int exit code in tkinter._exit
Andrew Svetlov [Sun, 9 Dec 2012 22:03:55 +0000 (00:03 +0200)]
Issue #16582: use int exit code in tkinter._exit

12 years agoIssue #16582: use int exit code in tkinter._exit
Andrew Svetlov [Sun, 9 Dec 2012 22:03:39 +0000 (00:03 +0200)]
Issue #16582: use int exit code in tkinter._exit

12 years agoIssue #16582: use int exit code in tkinter._exit
Andrew Svetlov [Sun, 9 Dec 2012 22:02:31 +0000 (00:02 +0200)]
Issue #16582: use int exit code in tkinter._exit

12 years agomerge from 3.3 - Document reason attribute for urllib.error.HTTPError
Senthil Kumaran [Sun, 9 Dec 2012 21:53:15 +0000 (13:53 -0800)]
merge from 3.3 - Document reason attribute for urllib.error.HTTPError

12 years agomerge from 3.2 - reason attribute for urllib.error.HTTPError
Senthil Kumaran [Sun, 9 Dec 2012 21:52:31 +0000 (13:52 -0800)]
merge from 3.2 - reason attribute for urllib.error.HTTPError

12 years agoFix issue13211 - Document the reason attribute for urllib.error.HTTPError
Senthil Kumaran [Sun, 9 Dec 2012 21:51:05 +0000 (13:51 -0800)]
Fix issue13211 - Document the reason attribute for urllib.error.HTTPError

12 years agoMake test of poll() use unbuffered IO
Richard Oudkerk [Sun, 9 Dec 2012 16:05:22 +0000 (16:05 +0000)]
Make test of poll() use unbuffered IO

12 years agoIssue #16616: Enable test in test_poll which was (accidentally?) disabled
Richard Oudkerk [Sun, 9 Dec 2012 16:05:20 +0000 (16:05 +0000)]
Issue #16616: Enable test in test_poll which was (accidentally?) disabled

12 years agomerge 3.3
Benjamin Peterson [Sun, 9 Dec 2012 15:17:39 +0000 (10:17 -0500)]
merge 3.3

12 years agomerge 3.2
Benjamin Peterson [Sun, 9 Dec 2012 15:16:46 +0000 (10:16 -0500)]
merge 3.2

12 years agomerge 3.1
Benjamin Peterson [Sun, 9 Dec 2012 15:14:42 +0000 (10:14 -0500)]
merge 3.1

12 years agoIssue #16248: Disable code execution from the user's home directory by tkinter when...
Antoine Pitrou [Sun, 9 Dec 2012 13:47:23 +0000 (14:47 +0100)]
Issue #16248: Disable code execution from the user's home directory by tkinter when the -E flag is passed to Python.
Patch by Zachary Ware.

12 years agoIssue #16248: Disable code execution from the user's home directory by tkinter when...
Antoine Pitrou [Sun, 9 Dec 2012 13:46:46 +0000 (14:46 +0100)]
Issue #16248: Disable code execution from the user's home directory by tkinter when the -E flag is passed to Python.
Patch by Zachary Ware.

12 years agoIssue #16248: Disable code execution from the user's home directory by tkinter when...
Antoine Pitrou [Sun, 9 Dec 2012 13:46:18 +0000 (14:46 +0100)]
Issue #16248: Disable code execution from the user's home directory by tkinter when the -E flag is passed to Python.
Patch by Zachary Ware.

12 years agoIssue #16248: Disable code execution from the user's home directory by tkinter when...
Antoine Pitrou [Sun, 9 Dec 2012 13:46:18 +0000 (14:46 +0100)]
Issue #16248: Disable code execution from the user's home directory by tkinter when the -E flag is passed to Python.
Patch by Zachary Ware.

12 years agoIssue #13390: New function :func:`sys.getallocatedblocks()` returns the number of...
Antoine Pitrou [Sun, 9 Dec 2012 13:28:26 +0000 (14:28 +0100)]
Issue #13390: New function :func:`sys.getallocatedblocks()` returns the number of memory blocks currently allocated.
Also, the ``-R`` option to regrtest uses this function to guard against memory allocation leaks.

12 years agoMerge from 3.3 (issue #15209)
Nick Coghlan [Sun, 9 Dec 2012 06:22:17 +0000 (16:22 +1000)]
Merge from 3.3 (issue #15209)

12 years agoIssue #15209: Fix typo and some additional wording tweaks
Nick Coghlan [Sun, 9 Dec 2012 06:21:46 +0000 (16:21 +1000)]
Issue #15209: Fix typo and some additional wording tweaks

12 years agoMerge fixes for #13614, #13512 and #7719 from 3.3
Éric Araujo [Sun, 9 Dec 2012 03:57:08 +0000 (22:57 -0500)]
Merge fixes for #13614, #13512 and #7719 from 3.3

12 years agoMerge fixes for #13614, #13512 and #7719 from 3.2
Éric Araujo [Sun, 9 Dec 2012 03:53:43 +0000 (22:53 -0500)]
Merge fixes for #13614, #13512 and #7719 from 3.2

12 years agoBranch merge
Éric Araujo [Sun, 9 Dec 2012 03:47:03 +0000 (22:47 -0500)]
Branch merge

12 years agoFix setup.py register failure with invalid rst in description (#13614).
Éric Araujo [Sun, 9 Dec 2012 03:41:11 +0000 (22:41 -0500)]
Fix setup.py register failure with invalid rst in description (#13614).

Original patch by Julien Courteau and Pierre Paul Lefebvre.

12 years agoFix a few markup/grammar nits
Éric Araujo [Sat, 8 Dec 2012 23:35:31 +0000 (18:35 -0500)]
Fix a few markup/grammar nits

12 years agoIssue #16602: When a weakref's target was part of a long deallocation chain, the...
Antoine Pitrou [Sat, 8 Dec 2012 20:18:50 +0000 (21:18 +0100)]
Issue #16602: When a weakref's target was part of a long deallocation chain, the object could remain reachable through its weakref even though its refcount had dropped to zero.

Thanks to Eugene Toder for diagnosing and reporting the issue.

12 years agoIssue #16602: When a weakref's target was part of a long deallocation chain, the...
Antoine Pitrou [Sat, 8 Dec 2012 20:17:03 +0000 (21:17 +0100)]
Issue #16602: When a weakref's target was part of a long deallocation chain, the object could remain reachable through its weakref even though its refcount had dropped to zero.

Thanks to Eugene Toder for diagnosing and reporting the issue.

12 years agoIssue #16602: When a weakref's target was part of a long deallocation chain, the...
Antoine Pitrou [Sat, 8 Dec 2012 20:15:26 +0000 (21:15 +0100)]
Issue #16602: When a weakref's target was part of a long deallocation chain, the object could remain reachable through its weakref even though its refcount had dropped to zero.

Thanks to Eugene Toder for diagnosing and reporting the issue.

12 years agoCreate ~/.pypirc securely (#13512).
Éric Araujo [Sat, 8 Dec 2012 19:51:47 +0000 (14:51 -0500)]
Create ~/.pypirc securely (#13512).

There was a window between the write and the chmod where the user’s
password would be exposed, depending on default permissions.  Philip
Jenvey’s patch fixes it.

12 years agoIgnore .nfs* files in distutils (#7719).
Éric Araujo [Sat, 8 Dec 2012 19:21:51 +0000 (14:21 -0500)]
Ignore .nfs* files in distutils (#7719).

These files are created by some NFS clients a file is edited and removed
concurrently (see added link in doc for more info).  If such a file is
removed between distutils calls listdir and copy, it will get confused.
Other special files are ignored in sdist (namely VCS directories), but
this has to be filtered out earlier.

12 years agoDrop double newlines printed in some file iteration examples.
Andrew Svetlov [Sat, 8 Dec 2012 15:59:58 +0000 (17:59 +0200)]
Drop double newlines printed in some file iteration examples.

Patch by Steven Kryskalla.

12 years agoDrop double newlines printed in some file iteration examples.
Andrew Svetlov [Sat, 8 Dec 2012 15:59:23 +0000 (17:59 +0200)]
Drop double newlines printed in some file iteration examples.

Patch by Steven Kryskalla.

12 years agoDrop double newlines printed in some file iteration examples.
Andrew Svetlov [Sat, 8 Dec 2012 15:59:03 +0000 (17:59 +0200)]
Drop double newlines printed in some file iteration examples.

Patch by Steven Kryskalla.

12 years agoMerge from 3.3 (issue #16267)
Nick Coghlan [Sat, 8 Dec 2012 12:57:21 +0000 (22:57 +1000)]
Merge from 3.3 (issue #16267)

12 years agoClose issue #16267: better docs for @abstractmethod composition
Nick Coghlan [Sat, 8 Dec 2012 12:56:02 +0000 (22:56 +1000)]
Close issue #16267: better docs for @abstractmethod composition

12 years agoMerge from 3.3 (Issue #15209)
Nick Coghlan [Sat, 8 Dec 2012 12:24:23 +0000 (22:24 +1000)]
Merge from 3.3 (Issue #15209)

12 years agoIssue #15209: Clarify exception chaining description
Nick Coghlan [Sat, 8 Dec 2012 11:39:24 +0000 (21:39 +1000)]
Issue #15209: Clarify exception chaining description

- not allowed when implicitly re-raised the current exception
- last exception raised is always displayed last
- attempt to make it clearer when/if cause and context are shown

12 years agoMerge debug output removal with 3.3.
Ezio Melotti [Sat, 8 Dec 2012 10:30:58 +0000 (12:30 +0200)]
Merge debug output removal with 3.3.

12 years agoMerge debug output removal with 3.2.
Ezio Melotti [Sat, 8 Dec 2012 10:30:44 +0000 (12:30 +0200)]
Merge debug output removal with 3.2.

12 years agoRemove debug output from example.
Ezio Melotti [Sat, 8 Dec 2012 10:29:40 +0000 (12:29 +0200)]
Remove debug output from example.

12 years agoIssue #16628: Fix a memory leak in ctypes.resize().
Antoine Pitrou [Sat, 8 Dec 2012 10:07:46 +0000 (11:07 +0100)]
Issue #16628: Fix a memory leak in ctypes.resize().

12 years agoIssue #16628: Fix a memory leak in ctypes.resize().
Antoine Pitrou [Sat, 8 Dec 2012 10:07:16 +0000 (11:07 +0100)]
Issue #16628: Fix a memory leak in ctypes.resize().

12 years agoIssue #16628: Fix a memory leak in ctypes.resize().
Antoine Pitrou [Sat, 8 Dec 2012 10:05:50 +0000 (11:05 +0100)]
Issue #16628: Fix a memory leak in ctypes.resize().

12 years agoadd fixer for reload() -> imp.reload() (closes #11797)\n\nPatch by Laurie Clark-Micha...
Benjamin Peterson [Sat, 8 Dec 2012 03:44:10 +0000 (22:44 -0500)]
add fixer for reload() -> imp.reload() (closes #11797)\n\nPatch by Laurie Clark-Michalek and Berker Peksag

12 years agoIssue #16495: remove extraneous NULL encoding check from bytes_decode().
Chris Jerdonek [Fri, 7 Dec 2012 23:51:53 +0000 (15:51 -0800)]
Issue #16495: remove extraneous NULL encoding check from bytes_decode().

The NULL encoding check in bytes_decode() was unnecessary because this case
is already taken care of by the call to _Py_normalize_encoding() inside
PyUnicode_Decode().

12 years agoImprove OrderedDict equality test.
Raymond Hettinger [Fri, 7 Dec 2012 18:18:22 +0000 (10:18 -0800)]
Improve OrderedDict equality test.

12 years agoassert than we never try to deal with True, False, or None as a name
Benjamin Peterson [Thu, 6 Dec 2012 22:49:58 +0000 (17:49 -0500)]
assert than we never try to deal with True, False, or None as a name

12 years agocreate NameConstant AST class for None, True, and False literals (closes #16619)
Benjamin Peterson [Thu, 6 Dec 2012 22:41:04 +0000 (17:41 -0500)]
create NameConstant AST class for None, True, and False literals (closes #16619)

12 years agoSpecify which I/O ABC methods have implementations in the docs.
Andrew Svetlov [Thu, 6 Dec 2012 10:21:22 +0000 (12:21 +0200)]
Specify which I/O ABC methods have implementations in the docs.

12 years agoSpecify which I/O ABC methods have implementations in the docs.
Andrew Svetlov [Thu, 6 Dec 2012 10:21:12 +0000 (12:21 +0200)]
Specify which I/O ABC methods have implementations in the docs.

12 years agoSpecify which I/O ABC methods have implementations in the docs.
Andrew Svetlov [Thu, 6 Dec 2012 10:20:56 +0000 (12:20 +0200)]
Specify which I/O ABC methods have implementations in the docs.

12 years agoUpdate comment: SAVE_EXC_STATE and SWAP_EXC_STATE macroses are saave_exc_state and...
Andrew Svetlov [Wed, 5 Dec 2012 15:59:29 +0000 (17:59 +0200)]
Update comment: SAVE_EXC_STATE and SWAP_EXC_STATE macroses are saave_exc_state and swap_exc_state functions now.

12 years agoUpdate comment: SAVE_EXC_STATE and SWAP_EXC_STATE macroses are saave_exc_state and...
Andrew Svetlov [Wed, 5 Dec 2012 15:59:10 +0000 (17:59 +0200)]
Update comment: SAVE_EXC_STATE and SWAP_EXC_STATE macroses are saave_exc_state and swap_exc_state functions now.

12 years agoMerge: skip pdb test for #13120 if threading is not available.
Andrew Svetlov [Wed, 5 Dec 2012 13:07:10 +0000 (15:07 +0200)]
Merge: skip pdb test for #13120 if threading is not available.

12 years agoMerge: skip pdb test for #13120 if threading is not available.
Andrew Svetlov [Wed, 5 Dec 2012 13:06:54 +0000 (15:06 +0200)]
Merge: skip pdb test for #13120 if threading is not available.

12 years agoSkip pdb test for #13120 if threading is not available.
Andrew Svetlov [Wed, 5 Dec 2012 13:06:23 +0000 (15:06 +0200)]
Skip pdb test for #13120 if threading is not available.

12 years agoMerge 3.3
Andrew Svetlov [Wed, 5 Dec 2012 09:12:50 +0000 (11:12 +0200)]
Merge 3.3

12 years agoFix typo.
Andrew Svetlov [Wed, 5 Dec 2012 09:12:14 +0000 (11:12 +0200)]
Fix typo.

12 years agoIssue #15627: This is simply an update to the name of a new method recently added
Eric Snow [Wed, 5 Dec 2012 07:43:43 +0000 (23:43 -0800)]
Issue #15627: This is simply an update to the name of a new method recently added
to importlib.abc.SourceLoader.

12 years agoMerge issue #13120: Allow to call pdb.set_trace() from thread.
Andrew Svetlov [Tue, 4 Dec 2012 19:10:20 +0000 (21:10 +0200)]
Merge issue #13120: Allow to call pdb.set_trace() from thread.

Patch by Ilya Sandler.

12 years agoIssue #13120: Allow to call pdb.set_trace() from thread.
Andrew Svetlov [Tue, 4 Dec 2012 19:08:28 +0000 (21:08 +0200)]
Issue #13120: Allow to call pdb.set_trace() from thread.

Patch by Ilya Sandler.

12 years agoIssue #16444: test more bytes in support.TESTFN_UNDECODABLE to support more Windows...
Victor Stinner [Tue, 4 Dec 2012 10:55:04 +0000 (11:55 +0100)]
Issue #16444: test more bytes in support.TESTFN_UNDECODABLE to support more Windows code pages

12 years agoMerge issue #13120: Allow to call pdb.set_trace() from thread.
Andrew Svetlov [Tue, 4 Dec 2012 19:10:55 +0000 (21:10 +0200)]
Merge issue #13120: Allow to call pdb.set_trace() from thread.

Patch by Ilya Sandler.

12 years agoCleanup unicodeobject.c
Victor Stinner [Tue, 4 Dec 2012 08:30:24 +0000 (09:30 +0100)]
Cleanup unicodeobject.c

 * Remove micro-optization:
   (errors == "surrogateescape" || strcmp(errors, "surrogateescape") == 0).
   Only use strcmp()
 * Initialize 'arg' members in unicode_format_arg() to help the compiler to
   diagnose real bugs and also make the code simpler to read

12 years agoIssue #16455: On FreeBSD and Solaris, if the locale is C, the
Victor Stinner [Tue, 4 Dec 2012 00:34:47 +0000 (01:34 +0100)]
Issue #16455: On FreeBSD and Solaris, if the locale is C, the
ASCII/surrogateescape codec is now used, instead of the locale encoding, to
decode the command line arguments. This change fixes inconsistencies with
os.fsencode() and os.fsdecode() because these operating systems announces an
ASCII locale encoding, whereas the ISO-8859-1 encoding is used in practice.

12 years agoFix test splitting in previous commit.
Antoine Pitrou [Mon, 3 Dec 2012 20:09:27 +0000 (21:09 +0100)]
Fix test splitting in previous commit.

12 years agoFix test splitting in previous commit.
Antoine Pitrou [Mon, 3 Dec 2012 20:09:08 +0000 (21:09 +0100)]
Fix test splitting in previous commit.

12 years agoFix test splitting in previous commit.
Antoine Pitrou [Mon, 3 Dec 2012 20:08:43 +0000 (21:08 +0100)]
Fix test splitting in previous commit.

12 years agoSplit the bigmem re test in two separate tests with different memory requirements.
Antoine Pitrou [Mon, 3 Dec 2012 19:56:27 +0000 (20:56 +0100)]
Split the bigmem re test in two separate tests with different memory requirements.