]> granicus.if.org Git - python/log
python
21 years ago2.4a0 Package Manager shouldn't attempt to use the 2.3 database.
Jack Jansen [Thu, 27 Nov 2003 22:55:39 +0000 (22:55 +0000)]
2.4a0 Package Manager shouldn't attempt to use the 2.3 database.
Things will definitely change before 2.4, but for now use a slightly
different URL.

21 years agothe "idle" script has moved from Lib/idlelib to Tools/scripts.
Jack Jansen [Thu, 27 Nov 2003 22:54:28 +0000 (22:54 +0000)]
the "idle" script has moved from Lib/idlelib to Tools/scripts.

21 years agoPatch #849350: Update to document bool return values. Backported to 2.3.
Martin v. Löwis [Thu, 27 Nov 2003 19:48:03 +0000 (19:48 +0000)]
Patch #849350: Update to document bool return values. Backported to 2.3.

21 years agoPatch #849595: Add socket.shutdown() constants.
Martin v. Löwis [Thu, 27 Nov 2003 19:40:22 +0000 (19:40 +0000)]
Patch #849595: Add socket.shutdown() constants.

21 years ago- clean up generated HTML
Fred Drake [Wed, 26 Nov 2003 20:55:49 +0000 (20:55 +0000)]
- clean up generated HTML
- make the output more XHTML friendly

21 years agoFix typo and mark-up; shorten text
Andrew M. Kuchling [Wed, 26 Nov 2003 18:05:26 +0000 (18:05 +0000)]
Fix typo and mark-up; shorten text

21 years agoFix typos
Andrew M. Kuchling [Wed, 26 Nov 2003 18:03:48 +0000 (18:03 +0000)]
Fix typos

21 years agoNits from a review of the documentation update.
Raymond Hettinger [Wed, 26 Nov 2003 17:52:45 +0000 (17:52 +0000)]
Nits from a review of the documentation update.

21 years agoAdd version changed doc for addition of fillchar to ljust/rjust/center
Neal Norwitz [Wed, 26 Nov 2003 14:54:56 +0000 (14:54 +0000)]
Add version changed doc for addition of fillchar to ljust/rjust/center

21 years agoAdd optional fillchar argument to ljust(), rjust(), and center() string methods.
Raymond Hettinger [Wed, 26 Nov 2003 08:21:35 +0000 (08:21 +0000)]
Add optional fillchar argument to ljust(), rjust(), and center() string methods.

21 years agoAs discussed on python-dev, banish apply(), buffer(), coerce(), and
Raymond Hettinger [Tue, 25 Nov 2003 21:48:21 +0000 (21:48 +0000)]
As discussed on python-dev, banish apply(), buffer(), coerce(), and
intern() to a separate region in the docs.

21 years agoExpose dict_contains() and PyDict_Contains() with is about 10% faster
Raymond Hettinger [Tue, 25 Nov 2003 21:12:14 +0000 (21:12 +0000)]
Expose dict_contains() and PyDict_Contains() with is about 10% faster
than PySequence_Contains() and more clearly applicable to dicts.

Apply the new function in setobject.c where __contains__ checking is
ubiquitous.

21 years agomake the generated HTML more XHTML friendly
Fred Drake [Tue, 25 Nov 2003 16:21:00 +0000 (16:21 +0000)]
make the generated HTML more XHTML friendly

21 years agoFix a typo introduced at 1.21
Kurt B. Kaiser [Tue, 25 Nov 2003 05:01:00 +0000 (05:01 +0000)]
Fix a typo introduced at 1.21

M IOBinding.py

Backported to 23-maint

21 years agoFactor out more duplicate code.
Raymond Hettinger [Mon, 24 Nov 2003 22:18:49 +0000 (22:18 +0000)]
Factor out more duplicate code.

21 years agoNote the addition of set() and frozenset().
Raymond Hettinger [Mon, 24 Nov 2003 07:14:54 +0000 (07:14 +0000)]
Note the addition of set() and frozenset().

21 years agoKeybindings with the Shift modifier now work correctly. So do bindings
Kurt B. Kaiser [Mon, 24 Nov 2003 05:26:16 +0000 (05:26 +0000)]
Keybindings with the Shift modifier now work correctly.  So do bindings
which use the Space key.  Limit unmodified user keybindings to the
function keys.
Python Bug 775353, IDLEfork Bugs 755647, 761557

Improve error handling during startup if there's no Tkinter.

M NEWS.txt
M PyShell.py
M config-keys.def
M configHandler.py
M keybindingDialog.py

Backport candidate.

21 years agoStop GCC warning about int literal that's so long that it becomes an
Guido van Rossum [Mon, 24 Nov 2003 04:13:13 +0000 (04:13 +0000)]
Stop GCC warning about int literal that's so long that it becomes an
unsigned int (on a 32-bit machine), by adding an explicit 'u' to the
literal (a prime used to improve the hash function for frozenset).

21 years agoSilence GCC warning when asserts are turned off.
Guido van Rossum [Mon, 24 Nov 2003 04:02:13 +0000 (04:02 +0000)]
Silence GCC warning when asserts are turned off.

21 years agoUpdate NEWS.txt to include some items missed earlier. Update the
Kurt B. Kaiser [Mon, 24 Nov 2003 03:23:16 +0000 (03:23 +0000)]
Update NEWS.txt to include some items missed earlier.  Update the
IDLE version to 1.1a0.
Modified Files:
NEWS.txt idlever.py

21 years ago* Checkin remaining documentation
Raymond Hettinger [Mon, 24 Nov 2003 02:57:33 +0000 (02:57 +0000)]
* Checkin remaining documentation
* Add more tests
* Refactor and neaten the code a bit.
* Rename union_update() to update().
* Improve the algorithms (making them a closer to sets.py).

21 years agotest_guess_all_types(): Use a more robust test for checking that
Barry Warsaw [Sun, 23 Nov 2003 16:21:55 +0000 (16:21 +0000)]
test_guess_all_types(): Use a more robust test for checking that
guess_all_extensions() returns (at least) what we expect.  As Jeff
Epler suggests in

http://mail.python.org/pipermail/python-dev/2003-September/038264.html

We use a set to test the results.  This fixes the test when
test_urllib2 is run before test_mimetypes.

21 years ago* Simplify hash function and add test to show effectiveness of the hash
Raymond Hettinger [Sun, 23 Nov 2003 02:49:05 +0000 (02:49 +0000)]
* Simplify hash function and add test to show effectiveness of the hash
  function.

* Add a better test for deepcopying.

* Add tests to show the __init__() function works like it does for list
  and tuple.  Add related test.

* Have shallow copies of frozensets return self.  Add related test.

* Have frozenset(f) return f if f is already a frozenset. Add related test.

* Beefed-up some existing tests.

21 years ago- When method objects have an attribute that can be satisfied either
Guido van Rossum [Sat, 22 Nov 2003 23:55:50 +0000 (23:55 +0000)]
- When method objects have an attribute that can be satisfied either
  by the function object or by the method object, the function
  object's attribute usually wins.  Christian Tismer pointed out that
  that this is really a mistake, because this only happens for special
  methods (like __reduce__) where the method object's version is
  really more appropriate than the function's attribute.  So from now
  on, all method attributes will have precedence over function
  attributes with the same name.

21 years agoExtend temporary hashability to remove() and discard().
Raymond Hettinger [Sat, 22 Nov 2003 03:55:23 +0000 (03:55 +0000)]
Extend temporary hashability to remove() and discard().
Brings the functionality back in line with sets.py.

21 years agoMore words: gave more motivation, and added cautions about the special
Tim Peters [Fri, 21 Nov 2003 22:20:57 +0000 (22:20 +0000)]
More words:  gave more motivation, and added cautions about the special
dangers of trying to iterate over weak dicts.

21 years agoAdd a missing import. Closes SF # 816344.
Barry Warsaw [Fri, 21 Nov 2003 20:28:15 +0000 (20:28 +0000)]
Add a missing import.  Closes SF # 816344.

21 years agoAllow temporary hashability for the __contains__ test.
Raymond Hettinger [Fri, 21 Nov 2003 18:36:54 +0000 (18:36 +0000)]
Allow temporary hashability for the __contains__ test.
(Requested by Alex Martelli.)

21 years agoissubset() and issuperset() to work with general iterables
Raymond Hettinger [Fri, 21 Nov 2003 07:56:36 +0000 (07:56 +0000)]
issubset() and issuperset() to work with general iterables

21 years agoThree minor performance improvements:
Raymond Hettinger [Thu, 20 Nov 2003 22:54:33 +0000 (22:54 +0000)]
Three minor performance improvements:

* Improve the hash function to increase the chance that distinct sets will
  have distinct xor'd hash totals.

* Use PyDict_Merge where possible (it is faster than an equivalent iter/set
  pair).

* Don't rebuild dictionaries where the input already has one.

21 years agoAdd two items
Andrew M. Kuchling [Thu, 20 Nov 2003 22:22:19 +0000 (22:22 +0000)]
Add two items

21 years agotest_applesingle is an expected skip almost anywhere.
Guido van Rossum [Thu, 20 Nov 2003 22:11:29 +0000 (22:11 +0000)]
test_applesingle is an expected skip almost anywhere.

21 years agoSF bug 839548: Bug in type's GC handling causes segfaults.
Tim Peters [Thu, 20 Nov 2003 21:21:46 +0000 (21:21 +0000)]
SF bug 839548:  Bug in type's GC handling causes segfaults.
Also SF patch 843455.

This is a critical bugfix.
I'll backport to 2.3 maint, but not beyond that.  The bugs this fixes
have been there since weakrefs were introduced.

21 years agotest_applesingle is an expected skip on Win32
Raymond Hettinger [Thu, 20 Nov 2003 19:02:02 +0000 (19:02 +0000)]
test_applesingle is an expected skip on Win32

21 years agoAdded some help to OSX/Dist/README.txt, plus all the information
Jack Jansen [Thu, 20 Nov 2003 13:50:28 +0000 (13:50 +0000)]
Added some help to OSX/Dist/README.txt, plus all the information
from the OS9 readme that is still relevant.
Got rid of Distributions/readme.txt.

21 years agoFix typo fix.
Walter Dörwald [Thu, 20 Nov 2003 13:38:01 +0000 (13:38 +0000)]
Fix typo fix.

21 years agoTalk about old code: removed a reference to THINK_C.
Jack Jansen [Thu, 20 Nov 2003 13:33:48 +0000 (13:33 +0000)]
Talk about old code: removed a reference to THINK_C.

21 years agoGot rid of macglue.h, replacing it by pymactoolbox.h where relevant.
Jack Jansen [Thu, 20 Nov 2003 13:31:00 +0000 (13:31 +0000)]
Got rid of macglue.h, replacing it by pymactoolbox.h where relevant.
Cleaned up various things in the toolbox modules.

21 years agoGot rid of macglue.h, moved the little bit that remains relevant
Jack Jansen [Thu, 20 Nov 2003 13:28:19 +0000 (13:28 +0000)]
Got rid of macglue.h, moved the little bit that remains relevant
to pymactoolbox.h (where it should have been in the first place).

21 years agoNo longer used.
Jack Jansen [Thu, 20 Nov 2003 13:27:33 +0000 (13:27 +0000)]
No longer used.

21 years agoGetting rid of all the code inside #ifdef macintosh too.
Jack Jansen [Thu, 20 Nov 2003 01:44:59 +0000 (01:44 +0000)]
Getting rid of all the code inside #ifdef macintosh too.

21 years agoGetting rid of code dependent on GUSI or the MetroWerks compiler.
Jack Jansen [Wed, 19 Nov 2003 22:52:23 +0000 (22:52 +0000)]
Getting rid of code dependent on GUSI or the MetroWerks compiler.

21 years agoGetting rid of one more TARGET_API_MAC_OSX.
Jack Jansen [Wed, 19 Nov 2003 22:43:57 +0000 (22:43 +0000)]
Getting rid of one more TARGET_API_MAC_OSX.

21 years agoPatch #831747: Add skip_accept_encoding parameter to putrequest.
Martin v. Löwis [Wed, 19 Nov 2003 19:51:55 +0000 (19:51 +0000)]
Patch #831747: Add skip_accept_encoding parameter to putrequest.

21 years agoGetting rid of code conditional on TARGET_API_MAC_*.
Jack Jansen [Wed, 19 Nov 2003 16:34:04 +0000 (16:34 +0000)]
Getting rid of code conditional on TARGET_API_MAC_*.

21 years agoPyDoc_STR is always defined nowadays (and has been for quite some time:-)
Jack Jansen [Wed, 19 Nov 2003 16:18:02 +0000 (16:18 +0000)]
PyDoc_STR is always defined nowadays (and has been for quite some time:-)

21 years agoGetting rid of WITHOUT_FRAMEWORKS and ACCESSOR_CALLS_ARE_FUNCTIONS:
Jack Jansen [Wed, 19 Nov 2003 16:13:35 +0000 (16:13 +0000)]
Getting rid of WITHOUT_FRAMEWORKS and ACCESSOR_CALLS_ARE_FUNCTIONS:
MacOS9isms.

21 years agoGet rid of MacOS9 support. Paths are still hard-coded, that'll be fixed
Jack Jansen [Wed, 19 Nov 2003 16:12:08 +0000 (16:12 +0000)]
Get rid of MacOS9 support. Paths are still hard-coded, that'll be fixed
later.

21 years agoRemove deprecation of sets.Set.update().
Raymond Hettinger [Wed, 19 Nov 2003 15:52:14 +0000 (15:52 +0000)]
Remove deprecation of sets.Set.update().

21 years agoWITHOUT_FRAMEWORKS conditional code bites the dust: this was for
Jack Jansen [Wed, 19 Nov 2003 15:32:46 +0000 (15:32 +0000)]
WITHOUT_FRAMEWORKS conditional code bites the dust: this was for
pre-carbon MacOS9 support.

21 years agoGone: all this functionality is now in the Carbon.File and Folder modules.
Jack Jansen [Wed, 19 Nov 2003 15:30:41 +0000 (15:30 +0000)]
Gone: all this functionality is now in the Carbon.File and Folder modules.

21 years agoGetting rid of support for the ancient Apple MPW compiler.
Jack Jansen [Wed, 19 Nov 2003 15:24:47 +0000 (15:24 +0000)]
Getting rid of support for the ancient Apple MPW compiler.

21 years agoMacOS9 support is gone.
Jack Jansen [Wed, 19 Nov 2003 14:55:33 +0000 (14:55 +0000)]
MacOS9 support is gone.

21 years agoRemoving the obvious OS9-only documents§
Jack Jansen [Wed, 19 Nov 2003 14:54:25 +0000 (14:54 +0000)]
Removing the obvious OS9-only documents§

21 years agoGetting rid of support for MacOS9 and earlier. This is the first step,
Jack Jansen [Wed, 19 Nov 2003 14:34:18 +0000 (14:34 +0000)]
Getting rid of support for MacOS9 and earlier. This is the first step,
and the biggest in size, but probably the easiest. Hunting through the
source code comes next.

21 years agoTemporary fix for buildon on both Panther and Jaguar.
Jack Jansen [Wed, 19 Nov 2003 13:56:40 +0000 (13:56 +0000)]
Temporary fix for buildon on both Panther and Jaguar.

21 years agoMoved various files over from the release23-maint branch.
Jack Jansen [Wed, 19 Nov 2003 13:53:55 +0000 (13:53 +0000)]
Moved various files over from the release23-maint branch.

21 years agoUpped version
Jack Jansen [Wed, 19 Nov 2003 13:50:21 +0000 (13:50 +0000)]
Upped version

21 years agoUpped version numbers, and converted to UTF-16.
Jack Jansen [Wed, 19 Nov 2003 13:49:28 +0000 (13:49 +0000)]
Upped version numbers, and converted to UTF-16.

21 years agoUpped.
Jack Jansen [Wed, 19 Nov 2003 13:48:13 +0000 (13:48 +0000)]
Upped.

21 years agoThis file is utf-16, not utf-8 (or ascii).
Jack Jansen [Wed, 19 Nov 2003 13:47:22 +0000 (13:47 +0000)]
This file is utf-16, not utf-8 (or ascii).

21 years agoForward port of various fixes that were initially only done on the
Jack Jansen [Wed, 19 Nov 2003 13:45:26 +0000 (13:45 +0000)]
Forward port of various fixes that were initially only done on the
release23-maint branch:
- Remember the scroll position when rebuilding the browser (as we do far
too often). Fixes #824430.
- Allow for the documentation to be inside PythonIDE as well as in
the Python.app inside the framework (the original location for 2.3).
- Updated version numbers
- In PythonIDE, add the Tools/IDE directory as the second entry in
sys.path in stead of as the first, leaving PythonIDE.app/Contents/Resources
as the first one.
- The code for setting the working directory to $HOME was both incorrect
and in the wrong place, fixed.
- On OSX the default location for IDE scripts is now
$HOME/Library/Python/IDE-Scripts.

21 years agoPorted from 23maint branch: add icons to BuildApplet.
Jack Jansen [Wed, 19 Nov 2003 13:37:22 +0000 (13:37 +0000)]
Ported from 23maint branch: add icons to BuildApplet.

21 years agoFix typos.
Walter Dörwald [Wed, 19 Nov 2003 13:35:49 +0000 (13:35 +0000)]
Fix typos.

21 years agoAdded a site
Jack Jansen [Wed, 19 Nov 2003 13:33:47 +0000 (13:33 +0000)]
Added a site

21 years agoUpped version
Jack Jansen [Wed, 19 Nov 2003 13:33:19 +0000 (13:33 +0000)]
Upped version

21 years ago- After an exception, run.py was not setting the exception vector. Noam
Kurt B. Kaiser [Wed, 19 Nov 2003 04:52:32 +0000 (04:52 +0000)]
- After an exception, run.py was not setting the exception vector. Noam
  Raphael suggested correcting this so pdb's postmortem pm() would work.
  IDLEfork Patch 844675

Modified: NEWS.txt  run.py

21 years ago__init__(): The docstring was incorrect regarding how header wrapping
Barry Warsaw [Wed, 19 Nov 2003 02:23:01 +0000 (02:23 +0000)]
__init__(): The docstring was incorrect regarding how header wrapping
gets done when maxheaderlen <> 0.  The header really gets wrapped via
the email.Header.Header class, which has a more sophisticated
algorithm than just splitting on semi-colons.

21 years agoGenerator's constructor: The documentation was incorrect regarding how
Barry Warsaw [Wed, 19 Nov 2003 02:22:36 +0000 (02:22 +0000)]
Generator's constructor: The documentation was incorrect regarding how
header wrapping gets done when maxheaderlen <> 0.  The header really
gets wrapped via the email.Header.Header class, which has a more
sophisticated algorithm than just splitting on semi-colons.

21 years agoModified version by Bob Ippolito. It passes the just-added test_applesingle
Jack Jansen [Tue, 18 Nov 2003 23:09:19 +0000 (23:09 +0000)]
Modified version by Bob Ippolito. It passes the just-added test_applesingle
after some minor mods. Fixes #803498, but should NOT be backported because
the original problem seems to be unreproducable.

21 years agoFix for [ 765456 ]: testAFakeZlib failed on platforms that use a
Just van Rossum [Tue, 18 Nov 2003 23:00:55 +0000 (23:00 +0000)]
Fix for [ 765456 ]: testAFakeZlib failed on platforms that use a
statically linked zlib module, but since the problem it tests can't
exist on these systems, simply skip it then. Will backport.

21 years agoFix for [ 782686 ]: new files used \r as the default line separator
Just van Rossum [Tue, 18 Nov 2003 22:41:16 +0000 (22:41 +0000)]
Fix for [ 782686 ]: new files used \r as the default line separator
instead of os.linesep.

21 years agoTest the applesingle decoder.
Jack Jansen [Tue, 18 Nov 2003 22:36:12 +0000 (22:36 +0000)]
Test the applesingle decoder.

21 years agoPatch #836434: Use dlopen/dlsym on AIX if available. Also disable
Martin v. Löwis [Tue, 18 Nov 2003 20:00:44 +0000 (20:00 +0000)]
Patch #836434: Use dlopen/dlsym on AIX if available. Also disable
_XOPEN_SOURCE on AIX 4.

21 years agoPatch #841807: Check whether a versioned libpython.so symlink is needed
Martin v. Löwis [Tue, 18 Nov 2003 19:54:20 +0000 (19:54 +0000)]
Patch #841807: Check whether a versioned libpython.so symlink is needed
in altbininstall. Backported to 2.3.

21 years agoPatch #843088: Fix typos. Backported to 2.3.
Martin v. Löwis [Tue, 18 Nov 2003 19:48:57 +0000 (19:48 +0000)]
Patch #843088: Fix typos. Backported to 2.3.

21 years agoPatch #794400: Let PYTHONSTARTUP influence the compiler flags.
Martin v. Löwis [Tue, 18 Nov 2003 19:46:25 +0000 (19:46 +0000)]
Patch #794400: Let PYTHONSTARTUP influence the compiler flags.

21 years agoImplement straightforward suggestions from gcc warnings (remove unused
Guido van Rossum [Tue, 18 Nov 2003 19:27:19 +0000 (19:27 +0000)]
Implement straightforward suggestions from gcc warnings (remove unused
variable, add extra braces).

21 years agoDocumentation for set objects.
Raymond Hettinger [Tue, 18 Nov 2003 17:50:34 +0000 (17:50 +0000)]
Documentation for set objects.

21 years agoUse PySequence_Contains() instead of direct access macro.
Raymond Hettinger [Tue, 18 Nov 2003 14:15:31 +0000 (14:15 +0000)]
Use PySequence_Contains() instead of direct access macro.

21 years agoVarious fixups (most suggested by Armin Rigo).
Raymond Hettinger [Mon, 17 Nov 2003 16:42:33 +0000 (16:42 +0000)]
Various fixups (most suggested by Armin Rigo).

21 years agoFix output spacing typo
Raymond Hettinger [Sun, 16 Nov 2003 16:36:58 +0000 (16:36 +0000)]
Fix output spacing typo

21 years ago* Migrate set() and frozenset() from the sandbox.
Raymond Hettinger [Sun, 16 Nov 2003 16:17:49 +0000 (16:17 +0000)]
* Migrate set() and frozenset() from the sandbox.
* Install the unittests, docs, newsitem, include file, and makefile update.
* Exercise the new functions whereever sets.py was being used.

Includes the docs for libfuncs.tex.  Separate docs for the types are
forthcoming.

21 years agoFix typo
Raymond Hettinger [Sun, 16 Nov 2003 13:44:19 +0000 (13:44 +0000)]
Fix typo

21 years agoChange ValueErrors to TypeErrors and add PyList_Check() assertions.
Raymond Hettinger [Sat, 15 Nov 2003 12:40:28 +0000 (12:40 +0000)]
Change ValueErrors to TypeErrors and add PyList_Check() assertions.

21 years agoVerify heappop argument is a list.
Raymond Hettinger [Sat, 15 Nov 2003 12:33:01 +0000 (12:33 +0000)]
Verify heappop argument is a list.

21 years agoMention patch #841977: modulefinder didn't find extension modules in packages
Thomas Heller [Fri, 14 Nov 2003 10:33:20 +0000 (10:33 +0000)]
Mention patch #841977: modulefinder didn't find extension modules in packages

Backported to release-maint23

21 years agoSF #841977 - modulefinder fails to find extension modules in packages
Thomas Heller [Fri, 14 Nov 2003 10:28:42 +0000 (10:28 +0000)]
SF #841977 - modulefinder fails to find extension modules in packages

The find_all_submodules() method in modulefinder only
looks for *.py, *.pyc, and *.pyo files.  Python
extension modules are only found if they are referenced
in import statements somewhere.

This patch uses the actual list from imp.get_suffixes().

Backported myself.

21 years agoupdate_refs(): assert that incoming refcounts aren't 0. The comment
Tim Peters [Fri, 14 Nov 2003 00:01:17 +0000 (00:01 +0000)]
update_refs():  assert that incoming refcounts aren't 0.  The comment
for this function has always claimed that was true, but it wasn't
verified before.  For the latest batch of "double deallocation" bugs
(stemming from weakref callbacks invoked by way of subtype_dealloc),
this assert would have triggered (instead of waiting for
_Py_ForgetReference to die with a segfault later).

21 years agosubtype_dealloc(): Simplified overly contorted retracking logic. With
Tim Peters [Thu, 13 Nov 2003 22:50:00 +0000 (22:50 +0000)]
subtype_dealloc():  Simplified overly contorted retracking logic.  With
this change, I think subtype_dealloc is actually a smidgen less obscure
than it was in 2.3 -- we got rid of a negation in an "if" <wink>.

21 years agosubtype_dealloc(): A more complete fix for critical bug 840829 +
Tim Peters [Thu, 13 Nov 2003 21:59:32 +0000 (21:59 +0000)]
subtype_dealloc():  A more complete fix for critical bug 840829 +
expanded the test case with a piece that needs the more-complete fix.

I'll backport this to 2.3 maint.

21 years agoVarious edits
Andrew M. Kuchling [Thu, 13 Nov 2003 21:33:26 +0000 (21:33 +0000)]
Various edits

21 years agoremove "support" for BerkeleyDB 3.1, it hasn't worked for a long time
Gregory P. Smith [Thu, 13 Nov 2003 08:30:03 +0000 (08:30 +0000)]
remove "support" for BerkeleyDB 3.1, it hasn't worked for a long time

21 years agoPatch #839877: Remove unused lambda expression.
Martin v. Löwis [Thu, 13 Nov 2003 07:45:39 +0000 (07:45 +0000)]
Patch #839877: Remove unused lambda expression.

21 years agoPatch #804543: strdup saved locales. Backported to 2.3.
Martin v. Löwis [Thu, 13 Nov 2003 07:43:21 +0000 (07:43 +0000)]
Patch #804543: strdup saved locales. Backported to 2.3.

21 years agoSF bug 840829: weakref callbacks and gc corrupt memory.
Tim Peters [Wed, 12 Nov 2003 20:43:28 +0000 (20:43 +0000)]
SF bug 840829:  weakref callbacks and gc corrupt memory.

subtype_dealloc():  This left the dying object exposed to gc, so that
if cyclic gc triggered during the weakref callback, gc tried to delete
the dying object a second time.  That's a disaster.  subtype_dealloc()
had a (I hope!) unique problem here, as every normal dealloc routine
untracks the object (from gc) before fiddling with weakrefs etc.  But
subtype_dealloc has obscure technical reasons for re-registering the
dying object with gc (already explained in a large comment block at
the bottom of the function).

The fix amounts to simply refraining from reregistering the dying object
with gc until after the weakref callback (if any) has been called.

This is a critical bug (hard to predict, and causes seemingly random
memory corruption when it occurs).  I'll backport it to 2.3 later.

21 years agopyexpat stopped building on Windows -- whining about the lack of a
Tim Peters [Wed, 12 Nov 2003 19:30:36 +0000 (19:30 +0000)]
pyexpat stopped building on Windows -- whining about the lack of a
HAVE_MEMMOVE define.  Just defined it on the cmdline (it doesn't
include Python.h, and Fred doesn't want to change the code).

21 years agoFix typo
Raymond Hettinger [Wed, 12 Nov 2003 16:42:10 +0000 (16:42 +0000)]
Fix typo

21 years agoExplain the advantages of reversed.
Raymond Hettinger [Wed, 12 Nov 2003 16:39:30 +0000 (16:39 +0000)]
Explain the advantages of reversed.

21 years agoPresent each feature in terms of what makes it useful or desirable.
Raymond Hettinger [Wed, 12 Nov 2003 16:27:50 +0000 (16:27 +0000)]
Present each feature in terms of what makes it useful or desirable.