]> granicus.if.org Git - python/log
python
9 years agoMerge 3.5 (test_eintr)
Victor Stinner [Mon, 12 Oct 2015 21:37:31 +0000 (23:37 +0200)]
Merge 3.5 (test_eintr)

9 years agoIssue #25277: Use a longer sleep in test_eintr to reduce the risk of race
Victor Stinner [Mon, 12 Oct 2015 21:37:02 +0000 (23:37 +0200)]
Issue #25277: Use a longer sleep in test_eintr to reduce the risk of race
condition in test_eintr.

9 years agoIssue #25353: Optimize unicode escape and raw unicode escape encoders to use
Victor Stinner [Mon, 12 Oct 2015 20:36:57 +0000 (22:36 +0200)]
Issue #25353: Optimize unicode escape and raw unicode escape encoders to use
the new _PyBytesWriter API.

9 years agoIssue #24164: Fix test_pyclbr
Victor Stinner [Mon, 12 Oct 2015 12:38:24 +0000 (14:38 +0200)]
Issue #24164: Fix test_pyclbr

Ignore pickle.partial symbol which comes from functools.partial.

9 years agoFix compilation error in _PyBytesWriter_WriteBytes() on Windows
Victor Stinner [Mon, 12 Oct 2015 11:57:47 +0000 (13:57 +0200)]
Fix compilation error in _PyBytesWriter_WriteBytes() on Windows

9 years agoWriter APIs: use empty string singletons
Victor Stinner [Mon, 12 Oct 2015 11:29:43 +0000 (13:29 +0200)]
Writer APIs: use empty string singletons

Modify _PyBytesWriter_Finish() and _PyUnicodeWriter_Finish() to return the
empty bytes/Unicode string if the string is empty.

9 years agoRelax _PyBytesWriter API
Victor Stinner [Mon, 12 Oct 2015 11:12:54 +0000 (13:12 +0200)]
Relax _PyBytesWriter API

Don't require _PyBytesWriter pointer to be a "char *". Same change for
_PyBytesWriter_WriteBytes() parameter.

For example, binascii uses "unsigned char*".

9 years agomerge 3.5
Benjamin Peterson [Mon, 12 Oct 2015 06:03:47 +0000 (23:03 -0700)]
merge 3.5

9 years agomerge 3.4
Benjamin Peterson [Mon, 12 Oct 2015 06:03:41 +0000 (23:03 -0700)]
merge 3.4

9 years agoactually link to the version attributes documentation
Benjamin Peterson [Mon, 12 Oct 2015 06:03:22 +0000 (23:03 -0700)]
actually link to the version attributes documentation

9 years agoMinor fixup. maxlen is already known.
Raymond Hettinger [Mon, 12 Oct 2015 05:52:54 +0000 (22:52 -0700)]
Minor fixup.  maxlen is already known.

9 years agoRefactor the deque trim logic to eliminate the two separate trim functions.
Raymond Hettinger [Mon, 12 Oct 2015 05:34:48 +0000 (22:34 -0700)]
Refactor the deque trim logic to eliminate the two separate trim functions.

9 years agoMerge with 3.5
Terry Jan Reedy [Mon, 12 Oct 2015 02:08:02 +0000 (22:08 -0400)]
Merge with 3.5

9 years agoMerge with 3.4
Terry Jan Reedy [Mon, 12 Oct 2015 02:07:48 +0000 (22:07 -0400)]
Merge with 3.4

9 years agoIssue #22726: Re-activate config dialog help button with some content about
Terry Jan Reedy [Mon, 12 Oct 2015 02:07:31 +0000 (22:07 -0400)]
Issue #22726: Re-activate config dialog help button with some content about
the other buttons and the new IDLE Dark theme.

9 years agoMerge from 3.5
Steve Dower [Mon, 12 Oct 2015 01:07:07 +0000 (18:07 -0700)]
Merge from 3.5

9 years agoKeeps all-users launcher checkbox visible when the option cannot be changed.
Steve Dower [Mon, 12 Oct 2015 01:06:55 +0000 (18:06 -0700)]
Keeps all-users launcher checkbox visible when the option cannot be changed.

9 years agoMerge from 3.5
Steve Dower [Mon, 12 Oct 2015 01:05:27 +0000 (18:05 -0700)]
Merge from 3.5

9 years agoOnly detects features from previous version when a bundle is found.
Steve Dower [Mon, 12 Oct 2015 01:05:11 +0000 (18:05 -0700)]
Only detects features from previous version when a bundle is found.
Otherwise, stray registry entries would cause issues.
Also fixes an accelerator collision and improves UAC icons when upgrading.

9 years agoIssue #25143: Improves installer error messages for unsupported platforms.
Steve Dower [Sun, 11 Oct 2015 23:40:52 +0000 (16:40 -0700)]
Issue #25143: Improves installer error messages for unsupported platforms.

9 years agoIssue #25143: Improves installer error messages for unsupported platforms.
Steve Dower [Sun, 11 Oct 2015 23:40:41 +0000 (16:40 -0700)]
Issue #25143: Improves installer error messages for unsupported platforms.

9 years agoIssue #25163: Display correct directory in installer when using non-default settings.
Steve Dower [Sun, 11 Oct 2015 22:37:36 +0000 (15:37 -0700)]
Issue #25163: Display correct directory in installer when using non-default settings.

9 years agoIssue #25163: Display correct directory in installer when using non-default settings.
Steve Dower [Sun, 11 Oct 2015 22:37:22 +0000 (15:37 -0700)]
Issue #25163: Display correct directory in installer when using non-default settings.

9 years agoIssue #25361: Disables use of SSE2 instructions in Windows 32-bit build
Steve Dower [Sun, 11 Oct 2015 22:16:21 +0000 (15:16 -0700)]
Issue #25361: Disables use of SSE2 instructions in Windows 32-bit build

9 years agoIssue #25361: Disables use of SSE2 instructions in Windows 32-bit build
Steve Dower [Sun, 11 Oct 2015 22:15:52 +0000 (15:15 -0700)]
Issue #25361: Disables use of SSE2 instructions in Windows 32-bit build

9 years agoHoist the deque->maxlen lookup out of the inner-loop.
Raymond Hettinger [Sun, 11 Oct 2015 16:43:50 +0000 (09:43 -0700)]
Hoist the deque->maxlen lookup out of the inner-loop.

9 years agoIssue #24164: Fixed test_descr: __getnewargs_ex__ now is supported in protocols 2...
Serhiy Storchaka [Sun, 11 Oct 2015 14:52:09 +0000 (17:52 +0300)]
Issue #24164: Fixed test_descr: __getnewargs_ex__ now is supported in protocols 2 and 3.

9 years agoNo longer skip tests for classes with __getnewargs_ex__. The copy module
Serhiy Storchaka [Sun, 11 Oct 2015 14:48:51 +0000 (17:48 +0300)]
No longer skip tests for classes with __getnewargs_ex__.  The copy module
already supports reduce protocol 4 (issue #20289).

9 years agoNo longer skip tests for classes with __getnewargs_ex__. The copy module
Serhiy Storchaka [Sun, 11 Oct 2015 14:48:28 +0000 (17:48 +0300)]
No longer skip tests for classes with __getnewargs_ex__.  The copy module
already supports reduce protocol 4 (issue #20289).

9 years agoCleanup test_descr: remove C8 that is the same as C3.
Serhiy Storchaka [Sun, 11 Oct 2015 14:43:38 +0000 (17:43 +0300)]
Cleanup test_descr: remove C8 that is the same as C3.

9 years agoCleanup test_descr: remove C8 that is the same as C3.
Serhiy Storchaka [Sun, 11 Oct 2015 14:43:12 +0000 (17:43 +0300)]
Cleanup test_descr: remove C8 that is the same as C3.

9 years agoCleanup test_descr: remove C8 that is the same as C3.
Serhiy Storchaka [Sun, 11 Oct 2015 14:42:39 +0000 (17:42 +0300)]
Cleanup test_descr: remove C8 that is the same as C3.

9 years agoIssue #25357: Add an optional newline paramer to binascii.b2a_base64().
Victor Stinner [Sun, 11 Oct 2015 09:01:02 +0000 (11:01 +0200)]
Issue #25357: Add an optional newline paramer to binascii.b2a_base64().
base64.b64encode() uses it to avoid a memory copy.

9 years agoMerge 3.5 (test_coroutines)
Victor Stinner [Sun, 11 Oct 2015 08:53:50 +0000 (10:53 +0200)]
Merge 3.5 (test_coroutines)

9 years agoClose #25367: Fix test_coroutines()
Victor Stinner [Sun, 11 Oct 2015 08:53:15 +0000 (10:53 +0200)]
Close #25367: Fix test_coroutines()

Fix usage of support.import_module('asyncio'): store the result in an 'asyncio'
variable.

9 years agoNull merge 3.5
Victor Stinner [Sun, 11 Oct 2015 08:40:27 +0000 (10:40 +0200)]
Null merge 3.5

9 years agolose #25373: Fix regrtest --slow with interrupted test
Victor Stinner [Sun, 11 Oct 2015 08:39:56 +0000 (10:39 +0200)]
lose #25373: Fix regrtest --slow with interrupted test

9 years agoClose #25373: Fix regrtest --slow with interrupted test
Victor Stinner [Sun, 11 Oct 2015 08:37:25 +0000 (10:37 +0200)]
Close #25373: Fix regrtest --slow with interrupted test

* Fix accumulate_result(): don't use time on interrupted and failed test
* Add unit test for interrupted test
* Add unit test on --slow with interrupted test, with and without
  multiprocessing

9 years agoMerge 3.5 (test_coroutines, issue #25367)
Victor Stinner [Sun, 11 Oct 2015 08:10:49 +0000 (10:10 +0200)]
Merge 3.5 (test_coroutines, issue #25367)

9 years agoClose #25367: Fix test_coroutines with no thread support
Victor Stinner [Sun, 11 Oct 2015 08:10:31 +0000 (10:10 +0200)]
Close #25367: Fix test_coroutines with no thread support

Skip test_asyncio_1() when the asyncio module cannot be imported because
CPython is compiled with no thread support.

9 years agotest_regrtest: catch stderr in test_nowindows()
Victor Stinner [Sun, 11 Oct 2015 08:04:26 +0000 (10:04 +0200)]
test_regrtest: catch stderr in test_nowindows()

Check also that the deprecation warning is emited.

9 years agoClose #25369: Fix test_regrtest without thread support
Victor Stinner [Sun, 11 Oct 2015 08:03:28 +0000 (10:03 +0200)]
Close #25369: Fix test_regrtest without thread support

9 years agoClose #24784: Fix compilation without thread support
Victor Stinner [Sun, 11 Oct 2015 07:54:42 +0000 (09:54 +0200)]
Close #24784: Fix compilation without thread support

Add "#ifdef WITH_THREAD" around cals to:

* PyGILState_Check()
* _PyImport_AcquireLock()
* _PyImport_ReleaseLock()

9 years agoClose #24784: Fix compilation without thread support
Victor Stinner [Sun, 11 Oct 2015 07:54:42 +0000 (09:54 +0200)]
Close #24784: Fix compilation without thread support

Add "#ifdef WITH_THREAD" around cals to:

* PyGILState_Check()
* _PyImport_AcquireLock()
* _PyImport_ReleaseLock()

9 years agoClose #25368: Fix test_eintr when Python is compiled without thread support
Victor Stinner [Sun, 11 Oct 2015 07:47:17 +0000 (09:47 +0200)]
Close #25368: Fix test_eintr when Python is compiled without thread support

9 years agomerge 3.5 (#25375)
Benjamin Peterson [Sun, 11 Oct 2015 06:24:53 +0000 (23:24 -0700)]
merge 3.5 (#25375)

9 years agomerge 3.4
Benjamin Peterson [Sun, 11 Oct 2015 06:24:39 +0000 (23:24 -0700)]
merge 3.4

9 years agodon't mention Python 2.2 (closes #25375)
Benjamin Peterson [Sun, 11 Oct 2015 06:23:55 +0000 (23:23 -0700)]
don't mention Python 2.2 (closes #25375)

9 years agoMinor tweak. Make the maxlen comparisons a little more clear and consistent.
Raymond Hettinger [Sun, 11 Oct 2015 03:56:02 +0000 (23:56 -0400)]
Minor tweak.  Make the maxlen comparisons a little more clear and consistent.

9 years agomerge 3.5 (#25362)
Benjamin Peterson [Sun, 11 Oct 2015 02:36:51 +0000 (19:36 -0700)]
merge 3.5 (#25362)

9 years agomerge 3.4 (#25362)
Benjamin Peterson [Sun, 11 Oct 2015 02:36:40 +0000 (19:36 -0700)]
merge 3.4 (#25362)

9 years agouse the with statement for locking the internal condition (closes #25362)
Benjamin Peterson [Sun, 11 Oct 2015 02:34:46 +0000 (19:34 -0700)]
use the with statement for locking the internal condition (closes #25362)

Patch by Nir Soffer.

9 years agomerge 3.5 (#25371)
Benjamin Peterson [Sun, 11 Oct 2015 02:32:41 +0000 (19:32 -0700)]
merge 3.5 (#25371)

9 years agomerge 3.4 (#25371)
Benjamin Peterson [Sun, 11 Oct 2015 02:32:33 +0000 (19:32 -0700)]
merge 3.4 (#25371)

9 years agoadd a missing comma (closes #25371)
Benjamin Peterson [Sun, 11 Oct 2015 02:32:20 +0000 (19:32 -0700)]
add a missing comma (closes #25371)

9 years agoIssue #24164: Objects that need calling ``__new__`` with keyword arguments,
Serhiy Storchaka [Sat, 10 Oct 2015 19:42:18 +0000 (22:42 +0300)]
Issue #24164: Objects that need calling ``__new__`` with keyword arguments,
can now be pickled using pickle protocols older than protocol version 4.

9 years agoNull merge
Serhiy Storchaka [Sat, 10 Oct 2015 17:27:52 +0000 (20:27 +0300)]
Null merge

9 years agoNull merge
Serhiy Storchaka [Sat, 10 Oct 2015 17:27:28 +0000 (20:27 +0300)]
Null merge

9 years agoCorrect Misc/NEWS.
Serhiy Storchaka [Sat, 10 Oct 2015 17:26:16 +0000 (20:26 +0300)]
Correct Misc/NEWS.

9 years agoIssue #25365: test_pickle now works in threads disabled builds.
Serhiy Storchaka [Sat, 10 Oct 2015 17:11:13 +0000 (20:11 +0300)]
Issue #25365: test_pickle now works in threads disabled builds.

9 years agoIssue #25365: test_pickle now works in threads disabled builds.
Serhiy Storchaka [Sat, 10 Oct 2015 17:10:54 +0000 (20:10 +0300)]
Issue #25365: test_pickle now works in threads disabled builds.

9 years agoIssue #25365: test_pickle now works in threads disabled builds.
Serhiy Storchaka [Sat, 10 Oct 2015 17:10:07 +0000 (20:10 +0300)]
Issue #25365: test_pickle now works in threads disabled builds.

9 years agoIssue #25364: zipfile now works in threads disabled builds.
Serhiy Storchaka [Sat, 10 Oct 2015 16:44:23 +0000 (19:44 +0300)]
Issue #25364: zipfile now works in threads disabled builds.

9 years agoIssue #25364: zipfile now works in threads disabled builds.
Serhiy Storchaka [Sat, 10 Oct 2015 16:43:32 +0000 (19:43 +0300)]
Issue #25364: zipfile now works in threads disabled builds.

9 years agoIssue #25322: Merge contextlib.suppress test fix from 3.5
Martin Panter [Sat, 10 Oct 2015 11:06:05 +0000 (11:06 +0000)]
Issue #25322: Merge contextlib.suppress test fix from 3.5

9 years agoIssue #25322: Merge contextlib.suppress test fix from 3.4 into 3.5
Martin Panter [Sat, 10 Oct 2015 11:05:47 +0000 (11:05 +0000)]
Issue #25322: Merge contextlib.suppress test fix from 3.4 into 3.5

9 years agoIssue #25322: Fix test for nested contextlib.suppress
Martin Panter [Sat, 10 Oct 2015 11:04:44 +0000 (11:04 +0000)]
Issue #25322: Fix test for nested contextlib.suppress

9 years agoIssue #25161: Merge full stops from 3.5
Martin Panter [Sat, 10 Oct 2015 10:45:00 +0000 (10:45 +0000)]
Issue #25161: Merge full stops from 3.5

9 years agoIssue #25161: Merge full stops from 3.4 into 3.5
Martin Panter [Sat, 10 Oct 2015 10:44:25 +0000 (10:44 +0000)]
Issue #25161: Merge full stops from 3.4 into 3.5

9 years agoIssue #25161: Add full stops in documentation; patch by Takase Arihiro
Martin Panter [Sat, 10 Oct 2015 10:36:22 +0000 (10:36 +0000)]
Issue #25161: Add full stops in documentation; patch by Takase Arihiro

9 years agoIssue #22413: Merge StringIO doc from 3.5
Martin Panter [Sat, 10 Oct 2015 10:20:25 +0000 (10:20 +0000)]
Issue #22413: Merge StringIO doc from 3.5

9 years agoIssue #22413: Remove comment made out of date by Argument Clinic
Martin Panter [Sat, 10 Oct 2015 10:17:57 +0000 (10:17 +0000)]
Issue #22413: Remove comment made out of date by Argument Clinic

9 years agoIssue #22413: Merge StringIO doc from 3.4 into 3.5
Martin Panter [Sat, 10 Oct 2015 10:15:21 +0000 (10:15 +0000)]
Issue #22413: Merge StringIO doc from 3.4 into 3.5

9 years agoIssue #24402: Merge potential test fix from 3.5
Martin Panter [Sat, 10 Oct 2015 05:29:19 +0000 (05:29 +0000)]
Issue #24402: Merge potential test fix from 3.5

9 years agoIssue #24402: Merge potential test fix from 3.4 into 3.5
Martin Panter [Sat, 10 Oct 2015 05:28:17 +0000 (05:28 +0000)]
Issue #24402: Merge potential test fix from 3.4 into 3.5

9 years agoIssue #24402: Factor out PtyTests.run_child() in input() tests
Martin Panter [Sat, 10 Oct 2015 05:27:15 +0000 (05:27 +0000)]
Issue #24402: Factor out PtyTests.run_child() in input() tests

This reuses existing code to hopefully make the new test_input_no_stdout_
fileno() test work. It is hanging Free BSD 9 and OS X Tiger buildbots, and I
don't know why.

9 years agoIssue #22413: Document newline effect on StringIO initializer and getvalue
Martin Panter [Sat, 10 Oct 2015 03:01:20 +0000 (03:01 +0000)]
Issue #22413: Document newline effect on StringIO initializer and getvalue

Also add to comment in the C code.

9 years agoIssue #24402: Merge input() fix from 3.5
Martin Panter [Sat, 10 Oct 2015 02:09:41 +0000 (02:09 +0000)]
Issue #24402: Merge input() fix from 3.5

9 years agoIssue #24402: Merge input() fix from 3.4 into 3.5
Martin Panter [Sat, 10 Oct 2015 01:55:23 +0000 (01:55 +0000)]
Issue #24402: Merge input() fix from 3.4 into 3.5

9 years agoIssue #24402: Fix input() when stdout.fileno() fails; diagnosed by Eryksun
Martin Panter [Sat, 10 Oct 2015 01:25:38 +0000 (01:25 +0000)]
Issue #24402: Fix input() when stdout.fileno() fails; diagnosed by Eryksun

Also factored out some test cases into a new PtyTests class.

9 years agoCloses #25344: Merged fix from 3.5.
Vinay Sajip [Fri, 9 Oct 2015 23:54:18 +0000 (00:54 +0100)]
Closes #25344: Merged fix from 3.5.

9 years agoCloses #25344: Merged fix from 3.4.
Vinay Sajip [Fri, 9 Oct 2015 23:53:37 +0000 (00:53 +0100)]
Closes #25344: Merged fix from 3.4.

9 years agoCloses #25344: Added cookbook recipe to show buffering of logging events.
Vinay Sajip [Fri, 9 Oct 2015 23:52:35 +0000 (00:52 +0100)]
Closes #25344: Added cookbook recipe to show buffering of logging events.

9 years agoMerge for issue #25099
Brett Cannon [Fri, 9 Oct 2015 22:10:10 +0000 (15:10 -0700)]
Merge for issue #25099

9 years agoIssue #25099: Skip relevant tests in test_compileall when an entry on
Brett Cannon [Fri, 9 Oct 2015 22:09:43 +0000 (15:09 -0700)]
Issue #25099: Skip relevant tests in test_compileall when an entry on
sys.path has an unwritable __pycache__ directory.

This typically comes up when someone runs the test suite from an
administrative install of Python on Windows where the user does not
have write permissions to the stdlib's directory.

Thanks to Zachary Ware and Matthias Klose for reporting bugs related
to this issue.
(grafted from 34bbd537b3e688dfbb6498e9083445a6a72fc4b1)

9 years agoIssue #25099: Skip relevant tests in test_compileall when an entry on
Brett Cannon [Fri, 9 Oct 2015 22:09:43 +0000 (15:09 -0700)]
Issue #25099: Skip relevant tests in test_compileall when an entry on
sys.path has an unwritable __pycache__ directory.

This typically comes up when someone runs the test suite from an
administrative install of Python on Windows where the user does not
have write permissions to the stdlib's directory.

Thanks to Zachary Ware and Matthias Klose for reporting bugs related
to this issue.

9 years agoIssue #25349: Add fast path for b'%c' % int
Victor Stinner [Fri, 9 Oct 2015 20:50:36 +0000 (22:50 +0200)]
Issue #25349: Add fast path for b'%c' % int

Optimize also %% formater.

9 years agoIssue #25349: Optimize bytes % int
Victor Stinner [Fri, 9 Oct 2015 20:43:24 +0000 (22:43 +0200)]
Issue #25349: Optimize bytes % int

Optimize bytes.__mod__(args) for integere formats: %d (%i, %u), %o, %x and %X.
_PyBytesWriter is now used to format directly the integer into the writer
buffer, instead of using a temporary bytes object.

Formatting is between 30% and 50% faster on a microbenchmark.

9 years agoMerge #25328: add missing raise keyword in decode_data+SMTPUTF8 check.
R David Murray [Fri, 9 Oct 2015 14:20:58 +0000 (10:20 -0400)]
Merge #25328: add missing raise keyword in decode_data+SMTPUTF8 check.

9 years ago#25328: add missing raise keyword in decode_data+SMTPUTF8 check.
R David Murray [Fri, 9 Oct 2015 14:19:33 +0000 (10:19 -0400)]
#25328: add missing raise keyword in decode_data+SMTPUTF8 check.

This is a relatively benign bug, since having both be true was correctly
rejected at in SMTPServer even before this patch.

Patch by Xiang Zhang.

9 years agoOptimize error handlers of ASCII and Latin1 encoders when the replacement
Victor Stinner [Fri, 9 Oct 2015 11:10:05 +0000 (13:10 +0200)]
Optimize error handlers of ASCII and Latin1 encoders when the replacement
string is pure ASCII: use _PyBytesWriter_WriteBytes(), don't check individual
character.

Cleanup unicode_encode_ucs1():

* Rename repunicode to rep
* Clear rep object on error
* Factorize code between bytes and unicode path

9 years agoAdd _PyBytesWriter_WriteBytes() to factorize the code
Victor Stinner [Fri, 9 Oct 2015 10:57:22 +0000 (12:57 +0200)]
Add _PyBytesWriter_WriteBytes() to factorize the code

9 years ago_PyBytesWriter: simplify code to avoid "prealloc" parameters
Victor Stinner [Fri, 9 Oct 2015 10:38:53 +0000 (12:38 +0200)]
_PyBytesWriter: simplify code to avoid "prealloc" parameters

Substract preallocate bytes from min_size before calling
_PyBytesWriter_Prepare().

9 years ago_PyBytesWriter: rename size attribute to min_size
Victor Stinner [Fri, 9 Oct 2015 10:37:03 +0000 (12:37 +0200)]
_PyBytesWriter: rename size attribute to min_size

9 years agoIssue #25349: Optimize bytes % args using the new private _PyBytesWriter API
Victor Stinner [Fri, 9 Oct 2015 09:48:06 +0000 (11:48 +0200)]
Issue #25349: Optimize bytes % args using the new private _PyBytesWriter API

* Thanks to the _PyBytesWriter API, output smaller than 512 bytes are allocated
  on the stack and so avoid calling _PyBytes_Resize(). Because of that, change
  the default buffer size to fmtcnt instead of fmtcnt+100.
* Rely on _PyBytesWriter algorithm to overallocate the buffer instead of using
  a custom code. For example, _PyBytesWriter uses a different overallocation
  factor (25% or 50%) depending on the platform to get best performances.
* Disable overallocation for the last write.
* Replace C loops to fill characters with memset()
* Add also many comments to _PyBytes_Format()
* Remove unused FORMATBUFLEN constant
* Avoid the creation of a temporary bytes object when formatting a floating
  point number (when no custom formatting option is used)
* Fix also reference leaks on error handling
* Use Py_MEMCPY() to copy bytes between two formatters (%)

9 years agoHoist constant expression out of the inner loop.
Raymond Hettinger [Fri, 9 Oct 2015 05:34:08 +0000 (01:34 -0400)]
Hoist constant expression out of the inner loop.

9 years agoMake comparison more consistent
Raymond Hettinger [Fri, 9 Oct 2015 04:42:47 +0000 (00:42 -0400)]
Make comparison more consistent

9 years agoIssue #25298: Add lock and rlock weakref tests (Contributed by Nir Soffer).
Raymond Hettinger [Fri, 9 Oct 2015 04:03:51 +0000 (00:03 -0400)]
Issue #25298:  Add lock and rlock weakref tests (Contributed by Nir Soffer).

9 years agoIssue #25318: cleanup code _PyBytesWriter
Victor Stinner [Fri, 9 Oct 2015 01:38:24 +0000 (03:38 +0200)]
Issue #25318: cleanup code _PyBytesWriter

Rename "stack buffer" to "small buffer".

Add also an assertion in _PyBytesWriter_GetPos().

9 years agoIssue #25318: Fix backslashreplace()
Victor Stinner [Fri, 9 Oct 2015 01:37:11 +0000 (03:37 +0200)]
Issue #25318: Fix backslashreplace()

Fix code to estimate the needed space.