]> granicus.if.org Git - python/log
python
21 years agoSpeed-up the joiner call by avoiding Py_BuildValue().
Raymond Hettinger [Fri, 27 Feb 2004 10:30:49 +0000 (10:30 +0000)]
Speed-up the joiner call by avoiding Py_BuildValue().

21 years agomade cPickle fall back to the copy_reg/reduce protocol,
Christian Tismer [Thu, 26 Feb 2004 16:21:45 +0000 (16:21 +0000)]
made cPickle fall back to the copy_reg/reduce protocol,
if a function cannot be stored as global.
This is for compatibility with pickle.py .

21 years agoIgnore sizehint argument. Fixes SF #844561.
Marc-André Lemburg [Thu, 26 Feb 2004 15:22:17 +0000 (15:22 +0000)]
Ignore sizehint argument. Fixes SF #844561.

21 years agoOops, didn't mean to commit the removal of float_compare!
Michael W. Hudson [Thu, 26 Feb 2004 13:16:03 +0000 (13:16 +0000)]
Oops, didn't mean to commit the removal of float_compare!

21 years agoPass a variable that actually exists to PyFPE_END_PROTECT in
Michael W. Hudson [Thu, 26 Feb 2004 12:33:09 +0000 (12:33 +0000)]
Pass a variable that actually exists to PyFPE_END_PROTECT in
float_richcompare.  Reported on c.l.py by Helmut Jarausch.

21 years agoFixes SF bug # 778421
Gregory P. Smith [Thu, 26 Feb 2004 10:07:14 +0000 (10:07 +0000)]
Fixes SF bug # 778421

 * Fixed a bug in the compatibility interface set_location() method
   where it would not properly search to the next nearest key when
   used on BTree databases.  [SF bug id 788421]
 * Fixed a bug in the compatibility interface set_location() method
   where it could crash when looking up keys in a hash or recno
   format database due to an incorrect free().

21 years agoMake _spawn_posix be ready for EINTR. waitpid(2) can be interrupted
Hye-Shik Chang [Tue, 24 Feb 2004 23:54:17 +0000 (23:54 +0000)]
Make _spawn_posix be ready for EINTR. waitpid(2) can be interrupted
by SIGCHLD or sth because no signal is masked before. This fixes
an optimized installation problem on FreeBSD libpthread.

21 years agokLsUnknownType and kLSUnknownCreator were ints in stead of OSTypes.
Jack Jansen [Tue, 24 Feb 2004 21:49:10 +0000 (21:49 +0000)]
kLsUnknownType and kLSUnknownCreator were ints in stead of OSTypes.
Reported by Bob Ippolito.

21 years agoSetup file to allow the QuickTime for 2.4 to be compiled
Jack Jansen [Tue, 24 Feb 2004 21:25:31 +0000 (21:25 +0000)]
Setup file to allow the QuickTime for 2.4 to be compiled
for MacPython 2.3 (and, hopefully, python 2.3 for windows too).

21 years agofix typo in reference to RFC 3464 DSN MIME type
Fred Drake [Tue, 24 Feb 2004 20:58:10 +0000 (20:58 +0000)]
fix typo in reference to RFC 3464 DSN MIME type

21 years agoFix two bugs in the new do_open() implementation for HTTPHandler.
Jeremy Hylton [Tue, 24 Feb 2004 19:40:35 +0000 (19:40 +0000)]
Fix two bugs in the new do_open() implementation for HTTPHandler.

Invoke the standard error handlers for non-200 responses.

Always supply a "Connection: close" header to prevent the server from
leaving the connection open.  Downstream users of the socket may
attempt recv()/read() with no arguments, which would block if the
connection were kept open.

21 years agofix English usage error reported by Ken Fuchs
Fred Drake [Tue, 24 Feb 2004 16:13:36 +0000 (16:13 +0000)]
fix English usage error reported by Ken Fuchs

21 years agoReflow long line.
Jeremy Hylton [Mon, 23 Feb 2004 17:27:57 +0000 (17:27 +0000)]
Reflow long line.

21 years agoFix wrong character mapping in koi8_u: SF bug #902501.
Marc-André Lemburg [Mon, 23 Feb 2004 09:00:43 +0000 (09:00 +0000)]
Fix wrong character mapping in koi8_u: SF bug #902501.

21 years agoAdded close() (which flushes) to BufferingHandler and tidied MemoryHandler.close...
Vinay Sajip [Sat, 21 Feb 2004 22:14:34 +0000 (22:14 +0000)]
Added close() (which flushes) to BufferingHandler and tidied MemoryHandler.close() [SF #901330]

21 years agoHandler close() functions call flush() [SF #901330]
Vinay Sajip [Sat, 21 Feb 2004 22:12:32 +0000 (22:12 +0000)]
Handler close() functions call flush() [SF #901330]

21 years agoadding passing test. testing for g(*Nothing()) where Nothing is a user-defined iterator.
Samuele Pedroni [Sat, 21 Feb 2004 21:03:30 +0000 (21:03 +0000)]
adding passing test. testing for g(*Nothing()) where Nothing is a user-defined iterator.

21 years agoUse the right wininstXX.exe, depending on
Thomas Heller [Fri, 20 Feb 2004 19:38:50 +0000 (19:38 +0000)]
Use the right wininstXX.exe, depending on
msvccompiler.get_build_version().

Distributions without a pre-install-script didn't work any longer, we
must at least provide the terminating NUL character.

21 years agowininst-6.exe and wininst-7.1.exe are in CVS, so that they can be
Thomas Heller [Fri, 20 Feb 2004 18:33:38 +0000 (18:33 +0000)]
wininst-6.exe and wininst-7.1.exe are in CVS, so that they can be
included in Python distributions for systems other than Windows.
Windows installers can be build on non-Windows systems as long as they
only include pure python module distributions.

21 years agowininst.exe is no longer used - we now need wininst-6.exe or wininst-7.1.exe.
Thomas Heller [Fri, 20 Feb 2004 18:26:55 +0000 (18:26 +0000)]
wininst.exe is no longer used - we now need wininst-6.exe or wininst-7.1.exe.

21 years agoSolution and project file to build wininstXX.exe with MSVC7.1 (Visual
Thomas Heller [Fri, 20 Feb 2004 18:23:47 +0000 (18:23 +0000)]
Solution and project file to build wininstXX.exe with MSVC7.1 (Visual
Studio .NET 2003).

The output files are named wininst-7.1.exe and wininst-7.1_d.exe.

21 years agoTo avoid problems with conflicting dlls, the windows installers built
Thomas Heller [Fri, 20 Feb 2004 18:05:13 +0000 (18:05 +0000)]
To avoid problems with conflicting dlls, the windows installers built
by bdist_wininst *must* use the same runtime libary as the Python
version.

Actually this means the Python version where the installer is run, not
the one which is used to build it.  Must think about that - for now I
assume MSVC6 is used up to Python 2.3, and MSVC7.1 is used starting at
Python 2.4.

So the filename for wininst.exe is now wininst-6.exe for the Release
version and wininst-6_d.exe for the Debug version, when built with
MSVC6.

21 years agoRecompiled the binary wininst.exe.
Thomas Heller [Fri, 20 Feb 2004 14:44:32 +0000 (14:44 +0000)]
Recompiled the binary wininst.exe.

Patch #892660 from Mark Hammond, for distutils bdist_wininst command.

install.c: support for a 'pre-install-script', run before anything has
been installed. Provides a 'message_box' module function for use by
either the pre-install or post-install scripts.

bdist_wininst.py: support for pre-install script. Typo (build->built),
fixes so that --target-version can still work, even when the
distribution has extension modules - in this case, we insist on
--skip-build, as we still can't actually build other versions.

21 years agoPatch #892660 from Mark Hammond, for distutils bdist_wininst command.
Thomas Heller [Fri, 20 Feb 2004 14:43:21 +0000 (14:43 +0000)]
Patch #892660 from Mark Hammond, for distutils bdist_wininst command.

install.c: support for a 'pre-install-script', run before anything has
been installed. Provides a 'message_box' module function for use by
either the pre-install or post-install scripts.

bdist_wininst.py: support for pre-install script. Typo (build->built),
fixes so that --target-version can still work, even when the
distribution has extension modules - in this case, we insist on
--skip-build, as we still can't actually build other versions.

21 years agoSocket handler closed prior to end of test.
Vinay Sajip [Fri, 20 Feb 2004 13:19:16 +0000 (13:19 +0000)]
Socket handler closed prior to end of test.

21 years agoCopyright year & version number/version date changes.
Vinay Sajip [Fri, 20 Feb 2004 13:18:36 +0000 (13:18 +0000)]
Copyright year & version number/version date changes.
Exception traceback text is now cached.
Closing a handler now removes it from the internal _handlers list.
Handlers now chain to Handler.close() from their close() methods.
Exception info can be passed as a tuple in exc_info.
shutdown() is registered to be called at application exit.

21 years agoCopyright year change.
Vinay Sajip [Fri, 20 Feb 2004 13:17:27 +0000 (13:17 +0000)]
Copyright year change.
Corrections to comments.
Tracebacks can now be sent via SocketHandler.
SocketHandler now uses exponential backoff strategy.
Handlers now chain to Handler.close() from their close() methods.

21 years agoCopyright year change.
Vinay Sajip [Fri, 20 Feb 2004 13:16:36 +0000 (13:16 +0000)]
Copyright year change.
Corrections to comments.
Added RESET_ERROR definition.

21 years ago- "Mac OS" should always include the space
Fred Drake [Thu, 19 Feb 2004 23:17:46 +0000 (23:17 +0000)]
- "Mac OS" should always include the space
- comment out documentation for a non-existant feature (--spec-file)
  that the comments indicate isn't clearly needed
- lots of minor markup adjustments to get a more consistent
  presentation

21 years ago- according to Apple's publication style guide, yes, "Mac people" use
Fred Drake [Thu, 19 Feb 2004 23:03:29 +0000 (23:03 +0000)]
- according to Apple's publication style guide, yes, "Mac people" use
  the term Installer (always capitalized, however)
- generalize the text about the term "installer" in a fairly
  reasonable way

21 years agouse API documentation style for the descriptions of the additional
Fred Drake [Thu, 19 Feb 2004 22:28:15 +0000 (22:28 +0000)]
use API documentation style for the descriptions of the additional
functions made available in the post-installation script run by the
Windows installer

21 years agominor markup improvements
Fred Drake [Thu, 19 Feb 2004 22:16:05 +0000 (22:16 +0000)]
minor markup improvements

21 years ago"Fix" (for certain configurations of the planets, including
Michael W. Hudson [Thu, 19 Feb 2004 19:35:22 +0000 (19:35 +0000)]
"Fix" (for certain configurations of the planets, including
recent gcc on Linux/x86)

[ 899109 ] 1==float('nan')

by implementing rich comparisons for floats.

Seems to make comparisons involving NaNs somewhat less surprising
when the underlying C compiler actually implements C99 semantics.

21 years agoFix two small bugs: (1) on Windows, pathname munging didn't work
Guido van Rossum [Thu, 19 Feb 2004 19:16:50 +0000 (19:16 +0000)]
Fix two small bugs: (1) on Windows, pathname munging didn't work
right; (2) write_results_file() didn't return a tuple of two ints when
it couldn't create the file.  Will backport.

21 years agoKeep the list.pop() optimization while restoring the many possibility
Raymond Hettinger [Thu, 19 Feb 2004 06:12:06 +0000 (06:12 +0000)]
Keep the list.pop() optimization while restoring the many possibility
for types other than PyInt being accepted for the optional argument.
(Spotted by Neal Norwitz.)

21 years agoRemove gcc warning from using "main". Use METH_NOARGS instead of METH_VARARGS
Neal Norwitz [Thu, 19 Feb 2004 02:44:22 +0000 (02:44 +0000)]
Remove gcc warning from using "main".  Use METH_NOARGS instead of METH_VARARGS

21 years agoGet test to work when run from regrtest (add test_main), remove all CRs (^M)s
Neal Norwitz [Thu, 19 Feb 2004 02:37:29 +0000 (02:37 +0000)]
Get test to work when run from regrtest (add test_main), remove all CRs (^M)s

21 years agoImplementation of patch 869468
David Ascher [Wed, 18 Feb 2004 05:59:53 +0000 (05:59 +0000)]
Implementation of patch 869468

Allow the user to create Tkinter.Tcl objects which are
just like Tkinter.Tk objects except that they do not
initialize Tk. This is useful in circumstances where the
script is being run on machines that do not have an X
server running -- in those cases, Tk initialization fails,
even if no window is ever created.

Includes documentation change and tests.

Tested on Linux, Solaris and Windows.

Reviewed by Martin von Loewis.

21 years agocommentary about how bad ConfigParser is doesn't help here, and the
Fred Drake [Tue, 17 Feb 2004 22:35:19 +0000 (22:35 +0000)]
commentary about how bad ConfigParser is doesn't help here, and the
suggested approach to dealing with it isn't a good one; we need a
better general purpose config reader, not a distutils-specific reader

21 years agoOops. Return -1 to distinguish error from empty dict.
Jeremy Hylton [Tue, 17 Feb 2004 20:10:11 +0000 (20:10 +0000)]
Oops.  Return -1 to distinguish error from empty dict.

This change probably isn't work a bug fix.  It's unlikely that anyone
was calling this method without passing it a real dict.

21 years agoDouble the speed of list.pop() which was spending most of its time parsing
Raymond Hettinger [Tue, 17 Feb 2004 11:36:16 +0000 (11:36 +0000)]
Double the speed of list.pop() which was spending most of its time parsing
arguments.

21 years agoMention the optimization of list.extend().
Raymond Hettinger [Tue, 17 Feb 2004 10:46:32 +0000 (10:46 +0000)]
Mention the optimization of list.extend().

21 years agomarkup correction
Fred Drake [Tue, 17 Feb 2004 04:17:36 +0000 (04:17 +0000)]
markup correction

21 years agoMake socketmodule compile again on a modern Linux (that supports Bluetooth).
Anthony Baxter [Mon, 16 Feb 2004 05:35:28 +0000 (05:35 +0000)]
Make socketmodule compile again on a modern Linux (that supports Bluetooth).
The Bluetooth code was obviously never tested on Linux.

21 years agoFix docstrings to mention the correct function
Neal Norwitz [Mon, 16 Feb 2004 01:26:34 +0000 (01:26 +0000)]
Fix docstrings to mention the correct function

21 years agoPatch #892673: Replace /usr/local/bin/python with
Martin v. Löwis [Sun, 15 Feb 2004 21:27:03 +0000 (21:27 +0000)]
Patch #892673: Replace /usr/local/bin/python with
/usr/bin/env python'%{binsuffix}
Backported to 2.3.

21 years agoPatch #711838: Allow non-anonymous ftp urls in urllib2.
Martin v. Löwis [Sun, 15 Feb 2004 21:19:18 +0000 (21:19 +0000)]
Patch #711838: Allow non-anonymous ftp urls in urllib2.
Backported to 2.3.

21 years agoPatch #893566: Document that tp_dealloc may be called from any thread.
Martin v. Löwis [Sun, 15 Feb 2004 21:01:17 +0000 (21:01 +0000)]
Patch #893566: Document that tp_dealloc may be called from any thread.

21 years agoPatch #817379: Allow for absolute ftp paths.
Martin v. Löwis [Sun, 15 Feb 2004 20:51:39 +0000 (20:51 +0000)]
Patch #817379: Allow for absolute ftp paths.
Backported to 2.3.

21 years agoFix typo.
Neil Schemenauer [Sun, 15 Feb 2004 16:43:20 +0000 (16:43 +0000)]
Fix typo.

21 years ago* Moved the responsibility for emptying the previous list from list_fill
Raymond Hettinger [Sun, 15 Feb 2004 04:06:39 +0000 (04:06 +0000)]
* Moved the responsibility for emptying the previous list from list_fill
  to list_init.

* Replaced the code in list_extend with the superior code from list_fill.

* Eliminated list_fill.

Results:

* list.extend() no longer creates an intermediate tuple except to handle
  the special case of x.extend(x).  The saves memory and time.

* list.extend(x) runs
    about the same x is a list or tuple,
    a little faster when x is an iterable not defining __len__, and
    twice as fast when x is an iterable defining __len__.

* the code is about 15 lines shorter and no longer duplicates
  functionality.

21 years agoRefactor list_extend() and list_fill() for gains in code size, memory
Raymond Hettinger [Sun, 15 Feb 2004 03:57:00 +0000 (03:57 +0000)]
Refactor list_extend() and list_fill() for gains in code size, memory
utilization, and speed:

* Moved the responsibility for emptying the previous list from list_fill
  to list_init.

* Replaced the code in list_extend with the superior code from list_fill.

* Eliminated list_fill.

Results:

* list.extend() no longer creates an intermediate tuple except to handle
  the special case of x.extend(x).  The saves memory and time.

* list.extend(x) runs
    5 to 10% faster when x is a list or tuple
    15% faster when x is an iterable not defining __len__
    twice as fast when x is an iterable defining __len__

* the code is about 15 lines shorter and no longer duplicates
  functionality.

21 years agoFine tune the speed/space trade-off for overallocating small lists.
Raymond Hettinger [Sat, 14 Feb 2004 18:34:46 +0000 (18:34 +0000)]
Fine tune the speed/space trade-off for overallocating small lists.

The Py2.3 approach overallocated small lists by up to 8 elements.
The last checkin would limited this to one but slowed down (by 20 to 30%)
the creation of small lists between 3 to 8 elements.

This tune-up balances the two, limiting overallocation to 3 elements
(significantly reducing space consumption from Py2.3) and running faster
than the previous checkin.

The first part of the growth pattern (0, 4, 8, 16) neatly meshes with
allocators that trigger data movement only when crossing a power of two
boundary.  Also, then even numbers mesh well with common data alignments.

21 years agoFix missing return value. Spotted by Neal Norwitz
Raymond Hettinger [Sat, 14 Feb 2004 03:07:21 +0000 (03:07 +0000)]
Fix missing return value.  Spotted by Neal Norwitz

21 years ago- Fixing annoying warnings.
Gustavo Niemeyer [Sat, 14 Feb 2004 00:31:13 +0000 (00:31 +0000)]
- Fixing annoying warnings.

21 years ago- Fixed #853061: allow BZ2Compressor.compress() to receive an empty string
Gustavo Niemeyer [Sat, 14 Feb 2004 00:02:45 +0000 (00:02 +0000)]
- Fixed #853061: allow BZ2Compressor.compress() to receive an empty string
  as parameter.

21 years agoLists are measured in elements not bytes.
Raymond Hettinger [Fri, 13 Feb 2004 21:50:27 +0000 (21:50 +0000)]
Lists are measured in elements not bytes.

21 years agofurther testing indicates that the simplified version of the test
Fred Drake [Fri, 13 Feb 2004 19:21:57 +0000 (19:21 +0000)]
further testing indicates that the simplified version of the test
(re-using an existing test object class) no longer triggered the
original segfault when the fix was backed out; restoring the local
test object class to make the test effective

the assignment of the ref created at the end does not affect the test,
since the segfault happended before weakref.ref() returned; removing
the assignment

21 years ago* Note list optimizations
Raymond Hettinger [Fri, 13 Feb 2004 19:00:07 +0000 (19:00 +0000)]
* Note list optimizations
* Move an example out of a comment.

21 years agoOptimize list.pop() for the common special case of popping off the end.
Raymond Hettinger [Fri, 13 Feb 2004 18:36:31 +0000 (18:36 +0000)]
Optimize list.pop() for the common special case of popping off the end.
More than doubles its speed.

21 years ago* Optimized list appends and pops by making fewer calls the underlying system
Raymond Hettinger [Fri, 13 Feb 2004 11:36:39 +0000 (11:36 +0000)]
* Optimized list appends and pops by making fewer calls the underlying system
  realloc().  This is achieved by tracking the overallocation size in a new
  field and using that information to skip calls to realloc() whenever
  possible.

* Simplified and tightened the amount of overallocation.  For larger lists,
  this overallocates by 1/8th (compared to the previous scheme which ranged
  between 1/4th to 1/32nd over-allocation).  For smaller lists (n<6), the
  maximum overallocation is one byte (formerly it could be upto eight bytes).
  This saves memory in applications with large numbers of small lists.

* Eliminated the NRESIZE macro in favor of a new, static list_resize function
  that encapsulates the resizing logic.  Coverting this back to macro would
  give a small (under 1%) speed-up.  This was too small to warrant the loss
  of readability, maintainability, and de-coupling.

* Some functions using NRESIZE had grown unnecessarily complex in their
  efforts to bend to the macro's calling pattern.  With the new list_resize
  function in place, those other functions could be simplified.  That is
  being saved for a separate patch.

* The ob_item==NULL check could be eliminated from the new list_resize
  function.  This would entail finding each piece of code that sets ob_item
  to NULL and adding a new line to invalidate the overallocation tracking
  field.  Rather than impose a new requirement on other pieces of list code,
  it was preferred to leave the NULL check in place and retain the benefits
  of decoupling, maintainability and information hiding (only PyList_New()
  and list_sort() need to know about the new field).  This approach also
  reduces the odds of breaking an extension module.

(Collaborative effort by Raymond Hettinger, Hye-Shik Chang, Tim Peters,
 and Armin Rigo.)

21 years agoWe have cp932 support in standard distribution now. And there's now
Hye-Shik Chang [Fri, 13 Feb 2004 07:14:13 +0000 (07:14 +0000)]
We have cp932 support in standard distribution now. And there's now
no major codepages unsupported by Python, so remove the eg. case.

21 years agouse existing test object instead of defining a new class
Fred Drake [Thu, 12 Feb 2004 19:30:17 +0000 (19:30 +0000)]
use existing test object instead of defining a new class

21 years agominor markup improvements
Fred Drake [Thu, 12 Feb 2004 18:13:12 +0000 (18:13 +0000)]
minor markup improvements

21 years agoReplace backticks with repr() or "%r"
Walter Dörwald [Thu, 12 Feb 2004 17:35:32 +0000 (17:35 +0000)]
Replace backticks with repr() or "%r"

From SF patch #852334.

21 years agoThis is my patch #876198 plus a NEWS entry and a header frob.
Michael W. Hudson [Thu, 12 Feb 2004 15:28:27 +0000 (15:28 +0000)]
This is my patch #876198 plus a NEWS entry and a header frob.

Remove the ability to use (from C) arbitrary objects supporting the
read buffer interface as the co_code member of code objects.

21 years agofix minor markup error: \code{for} --> \keyword{for}
Fred Drake [Thu, 12 Feb 2004 14:35:18 +0000 (14:35 +0000)]
fix minor markup error: \code{for} --> \keyword{for}

21 years agoSF 895560: minor typo
Raymond Hettinger [Thu, 12 Feb 2004 09:50:42 +0000 (09:50 +0000)]
SF 895560:  minor typo

21 years agoneed to initialize ob_type slot at run-time, at least on cygwin
Skip Montanaro [Tue, 10 Feb 2004 20:27:40 +0000 (20:27 +0000)]
need to initialize ob_type slot at run-time, at least on cygwin

21 years agoadd support for \e to the {alltt} environment
Fred Drake [Tue, 10 Feb 2004 18:30:22 +0000 (18:30 +0000)]
add support for \e to the {alltt} environment

21 years agoSF #894428, fix typo
Neal Norwitz [Tue, 10 Feb 2004 18:07:16 +0000 (18:07 +0000)]
SF #894428, fix typo

21 years agoremove support for missing ANSI C header files (limits.h, stddef.h, etc).
Skip Montanaro [Tue, 10 Feb 2004 16:50:21 +0000 (16:50 +0000)]
remove support for missing ANSI C header files (limits.h, stddef.h, etc).

21 years agoFix indentation error in testGetServByName and rewrite loop to avoid clumsy
Skip Montanaro [Tue, 10 Feb 2004 15:51:15 +0000 (15:51 +0000)]
Fix indentation error in testGetServByName and rewrite loop to avoid clumsy
sentinel variable

21 years agoMake reversed() transparent with respect to length.
Raymond Hettinger [Tue, 10 Feb 2004 09:33:39 +0000 (09:33 +0000)]
Make reversed() transparent with respect to length.

21 years agoGive itertools.repeat() a length method.
Raymond Hettinger [Tue, 10 Feb 2004 09:25:40 +0000 (09:25 +0000)]
Give itertools.repeat() a length method.

21 years agothe \file macro is no longer a good example of where \e can't be used
Fred Drake [Mon, 9 Feb 2004 21:00:29 +0000 (21:00 +0000)]
the \file macro is no longer a good example of where \e can't be used

21 years agosupport \e in \file and \filenq; this is useful for Windows paths
Fred Drake [Mon, 9 Feb 2004 20:58:08 +0000 (20:58 +0000)]
support \e in \file and \filenq; this is useful for Windows paths

21 years agoMinor edits
Andrew M. Kuchling [Mon, 9 Feb 2004 13:23:34 +0000 (13:23 +0000)]
Minor edits

21 years agoDocumentation for PyDescr_NewClassMethod was missing - here's at least
Thomas Heller [Mon, 9 Feb 2004 10:47:11 +0000 (10:47 +0000)]
Documentation for PyDescr_NewClassMethod was missing - here's at least
the function prototype.

Already backported to release23-maint.

21 years agoSF patch #892821: example for urllib2 has SyntaxError
Raymond Hettinger [Sun, 8 Feb 2004 20:25:01 +0000 (20:25 +0000)]
SF patch #892821: example for urllib2 has SyntaxError
(Contributed by George Yoshida.)

21 years agoFix misspelled name.
Raymond Hettinger [Sun, 8 Feb 2004 20:18:26 +0000 (20:18 +0000)]
Fix misspelled name.

21 years agoFix typo
Raymond Hettinger [Sun, 8 Feb 2004 20:05:40 +0000 (20:05 +0000)]
Fix typo

21 years agoSF patch #884022: dynamic execution profiling vs opcode prediction
Raymond Hettinger [Sun, 8 Feb 2004 19:59:27 +0000 (19:59 +0000)]
SF patch #884022:  dynamic execution profiling vs opcode prediction
(Contributed by Andrew I MacIntyre.)

disables opcode prediction when dynamic execution
profiling is in effect, so the profiling counters at
the top of the main interpreter loop in eval_frame()
are updated for each opcode.

21 years agoSF patch #880552: Fix typo in usage message(prechm.py)
Raymond Hettinger [Sun, 8 Feb 2004 19:24:18 +0000 (19:24 +0000)]
SF patch #880552:  Fix typo in usage message(prechm.py)
(Contributed by George Yoshida.)

* Also convert tabs to spaces.

21 years agoSF patch #875689: >100k alloc wasted on startup
Raymond Hettinger [Sun, 8 Feb 2004 18:54:37 +0000 (18:54 +0000)]
SF patch #875689: >100k alloc wasted on startup
(Contributed by Mike Pall.)

Make sure fill_free_list() is called only once rather than 106 times
when pre-allocating small ints.

21 years agoSF bug #892854: typo in textwrap doc page
Raymond Hettinger [Sun, 8 Feb 2004 18:09:32 +0000 (18:09 +0000)]
SF bug #892854:  typo in textwrap doc page
(Reported by Drew Perttula.)

21 years agoSF bug #892492: Multiple close() for asyncore.dispatcher.
Raymond Hettinger [Sun, 8 Feb 2004 11:32:50 +0000 (11:32 +0000)]
SF bug #892492:  Multiple close() for asyncore.dispatcher.
(Contributed by Alexey Klimkin.)

Don't keep the file descriptor after the channel is deleted.

21 years agoRevert improvement to list.append() checked in before it was ready.
Raymond Hettinger [Sun, 8 Feb 2004 11:08:52 +0000 (11:08 +0000)]
Revert improvement to list.append() checked in before it was ready.

21 years agoLet reversed() work with itself.
Raymond Hettinger [Sun, 8 Feb 2004 10:49:42 +0000 (10:49 +0000)]
Let reversed() work with itself.

21 years agoFixed a bug in object.__reduce_ex__ (reduce_2) when using protocol
Jim Fulton [Sun, 8 Feb 2004 04:21:26 +0000 (04:21 +0000)]
Fixed a bug in object.__reduce_ex__ (reduce_2) when using protocol
  2.  Failure to clear the error when attempts to get the __getstate__
  attribute fail caused intermittent errors and odd behavior.

21 years agoMake deque.rotate() smarter. Beef-up related tests.
Raymond Hettinger [Sun, 8 Feb 2004 04:05:26 +0000 (04:05 +0000)]
Make deque.rotate() smarter.  Beef-up related tests.

21 years agoPatch #868499, adds -T option for code coverage. The implementation is a
Barry Warsaw [Sat, 7 Feb 2004 22:43:03 +0000 (22:43 +0000)]
Patch #868499, adds -T option for code coverage.  The implementation is a
fairly simpleminded adaptation of Zope3's test.py -T flag.

I also changed some booleans to use True/False where appropriate.

21 years ago* Incorporate Skip's suggestions for documentation (explain the word deque
Raymond Hettinger [Sat, 7 Feb 2004 21:13:00 +0000 (21:13 +0000)]
* Incorporate Skip's suggestions for documentation  (explain the word deque
  comes from and show the differences from lists).
* Add a rotate() method.

21 years agono longer support linux1 platform
Skip Montanaro [Sat, 7 Feb 2004 20:03:34 +0000 (20:03 +0000)]
no longer support linux1 platform

21 years agono longer support sunos4 platform
Skip Montanaro [Sat, 7 Feb 2004 20:01:11 +0000 (20:01 +0000)]
no longer support sunos4 platform

21 years agoRemove support for --without-universal-newlines (see PEP 11).
Skip Montanaro [Sat, 7 Feb 2004 13:53:46 +0000 (13:53 +0000)]
Remove support for --without-universal-newlines (see PEP 11).

21 years agoRemove support for systems defining Py_PTHREAD_D[467] in
Skip Montanaro [Sat, 7 Feb 2004 13:00:18 +0000 (13:00 +0000)]
Remove support for systems defining Py_PTHREAD_D[467] in
Python/thread_pthread.h.

21 years agoRemove HAVE_STRPTIME - no longer necessary with the pure Python version of
Skip Montanaro [Sat, 7 Feb 2004 12:55:46 +0000 (12:55 +0000)]
Remove HAVE_STRPTIME - no longer necessary with the pure Python version of
time.strptime().

21 years agoa couple other sunos4 support items removed
Skip Montanaro [Sat, 7 Feb 2004 12:50:19 +0000 (12:50 +0000)]
a couple other sunos4 support items removed

21 years agoUse collection.deque() instead of a list for a FIFO queue.
Raymond Hettinger [Sat, 7 Feb 2004 03:19:10 +0000 (03:19 +0000)]
Use collection.deque() instead of a list for a FIFO queue.