]> git.ipfire.org Git - thirdparty/Python/cpython.git/log
thirdparty/Python/cpython.git
23 years agoBump version numbers.
Fred Drake [Mon, 25 Mar 2002 19:38:17 +0000 (19:38 +0000)] 
Bump version numbers.

23 years agoBump Windows build number to 33.
Tim Peters [Mon, 25 Mar 2002 19:30:40 +0000 (19:30 +0000)] 
Bump Windows build number to 33.

23 years agoBump release to 2.2.1c2.
Tim Peters [Mon, 25 Mar 2002 19:28:43 +0000 (19:28 +0000)] 
Bump release to 2.2.1c2.

23 years agobackport jackjansen's checkin of
Michael W. Hudson [Mon, 25 Mar 2002 17:40:43 +0000 (17:40 +0000)] 
backport jackjansen's checkin of
    revision 1.211 of socketmodule.c

Due to interaction between the MSL C library and the GUSI I/O library I can get reads from sockets to work consistently either for unbuffered binary files or for buffered binary files, but not for both:-(

The workaround is to force socket.makefile() to disable buffering for binary files.

Fixes bug 534625. 2.2.1 candidate.

23 years agoAdded description of binhex.Error.
Fred Drake [Mon, 25 Mar 2002 16:37:39 +0000 (16:37 +0000)] 
Added description of binhex.Error.

23 years agoAdded a note about preference file problems for non-privileged OSX
Jack Jansen [Mon, 25 Mar 2002 14:53:11 +0000 (14:53 +0000)] 
Added a note about preference file problems for non-privileged OSX
users. Stopgap for bug #534158.

23 years agonnorwitz's fix for:
Michael W. Hudson [Mon, 25 Mar 2002 13:59:28 +0000 (13:59 +0000)] 
nnorwitz's fix for:

[ 532618 ] install fails if build .so fails

23 years agoBackport akuchling's netrc test.
Michael W. Hudson [Mon, 25 Mar 2002 13:23:53 +0000 (13:23 +0000)] 
Backport akuchling's netrc test.

23 years agoThis commit was manufactured by cvs2svn to create branch
cvs2svn [Mon, 25 Mar 2002 13:23:53 +0000 (13:23 +0000)] 
This commit was manufactured by cvs2svn to create branch
'release22-maint'.

23 years agobackport akuchling's checkin of
Michael W. Hudson [Mon, 25 Mar 2002 13:22:14 +0000 (13:22 +0000)] 
backport akuchling's checkin of
    revision 1.14 of netrc.py

[Bug #532115]  netrc module was broken
   * 'macdef' (macro definition) wasn't parsed correctly
   * account value not reset for a subsequent 'default' line
   * typo: 'whitepace' -> 'whitespace'

Bugfix candidate.

23 years agobackport tim_one's checkin of
Michael W. Hudson [Mon, 25 Mar 2002 13:21:41 +0000 (13:21 +0000)] 
backport tim_one's checkin of
    revision 2.55 of complexobject.c

SF bug 533198:  Complex power underflow raises exception.
Konrad was too kind.  Not only did it raise an exception, the specific
exception it raised made no sense.  These are old bugs in complex_pow()
and friends:

1. Raising 0 to a negative power isn't a range error, it's a domain
   error, so changed c_pow() to set errno to EDOM in that case instead
   of ERANGE.

2. Changed complex_pow() to:

A. Used the Py_ADJUST_ERANGE2 macro to try to clear errno of a spurious
   ERANGE error due to underflow in the libm pow() called by c_pow().

B. Produced different exceptions depending on the errno value:
   i) For errno==EDOM, raise ZeroDivisionError instead of ValueError.
      This is for consistency with the non-complex cases 0.0**-2 and
      0**-2 and 0L**-2.
   ii) For errno==ERANGE, raise OverflowError.

Bugfix candidate.

23 years agobackport akuchling's checkin of
Michael W. Hudson [Mon, 25 Mar 2002 13:15:52 +0000 (13:15 +0000)] 
backport akuchling's checkin of
    revision 1.142 of README

[Bug #532136] Change path in README, as suggested by Ralph Corderoy

Bugfix candidate.

23 years agobackport akuchling's checkin of
Michael W. Hudson [Mon, 25 Mar 2002 13:15:28 +0000 (13:15 +0000)] 
backport akuchling's checkin of
    revision 1.63 of install.py

Revert part of previous patch: several install_* subcommands expect
    .compile to be None, and set it to true if it is.
    Caught by Pearu Peterson.

Bugfix candidate, if the previous change is accepted for
release22-maint.

23 years agobackport akuchling's checkin of
Michael W. Hudson [Mon, 25 Mar 2002 13:15:04 +0000 (13:15 +0000)] 
backport akuchling's checkin of
    revision 1.62 of install.py

Add missing Boolean options
Remove unused no_compile flag
Initialize the Boolean attribute .compile to 0 instead of None

Bugfix candidate.

23 years agoThomas said this was OK.
Michael W. Hudson [Mon, 25 Mar 2002 13:14:23 +0000 (13:14 +0000)] 
Thomas said this was OK.

backport akuchling's checkin of
    revision 1.30 of bdist_wininst.py

Add unlisted Boolean options.  Thomas H., can you please check that I
    got this right?

Bugfix candidate, unless Thomas notes a problem.

23 years agobackport akuchling's checkin of
Michael W. Hudson [Mon, 25 Mar 2002 13:13:45 +0000 (13:13 +0000)] 
backport akuchling's checkin of
    revision 1.28 of bdist_rpm.py

[Bug #517451] bdist_rpm didn't list all of its Boolean options.
   (Someone should check the other commands for this same error.)

Bugfix candidate.

23 years agobackport nascheme's checkin of
Michael W. Hudson [Mon, 25 Mar 2002 13:11:47 +0000 (13:11 +0000)] 
backport nascheme's checkin of
    revision 1.60 of pydoc.py

Quote href properly.

23 years agobackport nascheme's checkin of
Michael W. Hudson [Mon, 25 Mar 2002 13:11:31 +0000 (13:11 +0000)] 
backport nascheme's checkin of
    revision 1.59 of pydoc.py

Remove unnecessary \b.  It was causing the RE to miss the tailing
slash on strings like "http://www.python.org/ is good".

23 years agobackport nascheme's checkin of
Michael W. Hudson [Mon, 25 Mar 2002 12:34:15 +0000 (12:34 +0000)] 
backport nascheme's checkin of
    revision 1.27 of cmd.py

Flush stdout before reading next command.  Closes SF bug 526357.

23 years agobackport nascheme's checkin of
Michael W. Hudson [Mon, 25 Mar 2002 12:33:56 +0000 (12:33 +0000)] 
backport nascheme's checkin of
    revision 2.102 of sysmodule.c

Fix wording of sys.exit docstring.  Close SF bug 534113.

23 years agoBackport of 1.27, 1.29 and 1.30:
Jack Jansen [Mon, 25 Mar 2002 12:33:48 +0000 (12:33 +0000)] 
Backport of 1.27, 1.29 and 1.30:
CodeWarrior includes Waste and Internet Config nowadays, obviating the need for separate downloads.

- Weaklink InterfaceLib in _Res module
- forgot to pass libraryflags and stdlibraryflags on to ppc/carbon
  project generation.

First half of fix to 531398.

Weaklink most toolbox modules, improving backward compatibility. Modules will no longer fail to load if a single routine is missing on the curent OS version, inShould finally fix 531398. 2.2.1 candidate.se an exception.

23 years agoRepeat trick with test_calendar.py
Michael W. Hudson [Mon, 25 Mar 2002 12:33:24 +0000 (12:33 +0000)] 
Repeat trick with test_calendar.py

23 years agoSlap HEAD version of calendar.py onto branch.
Michael W. Hudson [Mon, 25 Mar 2002 12:31:48 +0000 (12:31 +0000)] 
Slap HEAD version of calendar.py onto branch.

Hope this was right thing to do.

23 years agobackport akuchling's checkin of
Michael W. Hudson [Mon, 25 Mar 2002 12:25:26 +0000 (12:25 +0000)] 
backport akuchling's checkin of
    revision 2.19 of traceback.h

[Bug #528914] PyTraceBack_Store/Fetch were deleted in 1997, but their
              prototypes remain.  Noted by Yakov Markovitch.

Bugfix candidate.

23 years agobackport akuchling's checkin of
Michael W. Hudson [Mon, 25 Mar 2002 12:24:58 +0000 (12:24 +0000)] 
backport akuchling's checkin of
    revision 1.47 of httplib.py

[Bug #531616] Make HTTPS work again by adding a sendall method to the
FakeSocket class.  Without it, the sendall() call got the method on
the underlying socket object, and that messed up SSL.

Does httplib use other methods of sockets that FakeSocket doesn't support?
Someone should take a look...  (I'll try to give it a once-over.)

2.2.1 bugfix candidate.

23 years agoBackport of _Cmmodule.c 1.8 and cmsupport.py 1.6:
Jack Jansen [Mon, 25 Mar 2002 12:24:50 +0000 (12:24 +0000)] 
Backport of _Cmmodule.c 1.8 and cmsupport.py 1.6:
Weaklink most toolbox modules, improving backward compatibility. Modules will no longer fail to load if a single routine is missing on the curent OS version, in stead calling the missing routine will raise an exception.

Should finally fix 531398. 2.2.1 candidate.

Also blacklisted some constants with definitions that were not Python-compatible.

23 years agobackport bwarsaw's checkin of
Michael W. Hudson [Mon, 25 Mar 2002 12:24:24 +0000 (12:24 +0000)] 
backport bwarsaw's checkin of
    revision 4.8 of python-mode.el

(py-temp-directory): Add /var/tmp to the list of directories this
searches.  This is added after /tmp.  Closes SF bug #505488, except
that /var/tmp comes after /tmp instead of the patch's suggestion of
putting it before /usr/tmp.

23 years agoBackport of _Qdmodule.c 1.8, qdsupport.py 1.38:
Jack Jansen [Mon, 25 Mar 2002 12:20:43 +0000 (12:20 +0000)] 
Backport of _Qdmodule.c 1.8, qdsupport.py 1.38:
Weaklink most toolbox modules, improving backward compatibility. Modules will no longer fail to load if a single routine is missing on the curent OS version, in stead calling the missing routine will raise an exception.

Should finally fix 531398. 2.2.1 candidate.

Also blacklisted some constants with definitions that were not Python-compatible.

23 years agoBackport of _Dragmodule.c 1.8, dragsupport.py 1.10:
Jack Jansen [Mon, 25 Mar 2002 12:17:28 +0000 (12:17 +0000)] 
Backport of _Dragmodule.c 1.8, dragsupport.py 1.10:
Weaklink most toolbox modules, improving backward compatibility. Modules will no longer fail to load if a single routine is missing on the curent OS version, in stead calling the missing routine will raise an exception.

Should finally fix 531398. 2.2.1 candidate.

Also blacklisted some constants with definitions that were not Python-compatible.

23 years agoBackport of _Evtmodule.c 1.6, evtsupport.py 1.15:
Jack Jansen [Mon, 25 Mar 2002 11:14:28 +0000 (11:14 +0000)] 
Backport of _Evtmodule.c 1.6, evtsupport.py 1.15:
Weaklink most toolbox modules, improving backward compatibility. Modules will no longer fail to load if a single routine is missing on the curent OS version, in stead calling the missing routine will raise an exception.

Should finally fix 531398. 2.2.1 candidate.

Also blacklisted some constants with definitions that were not Python-compatible.

23 years agoBackport of _Fmmodule.c 1.6 and fmsupport.py 1.6:
Jack Jansen [Mon, 25 Mar 2002 10:56:59 +0000 (10:56 +0000)] 
Backport of _Fmmodule.c 1.6 and fmsupport.py 1.6:
Weaklink most toolbox modules, improving backward compatibility. Modules will no longer fail to load if a single routine is missing on the curent OS version, in stead calling the missing routine will raise an exception.

Should finally fix 531398. 2.2.1 candidate.

Also blacklisted some constants with definitions that were not Python-compatible.

23 years agoBackport of _Icnmodule.c 1.5, icnsupport.py 1.7:
Jack Jansen [Mon, 25 Mar 2002 10:45:21 +0000 (10:45 +0000)] 
Backport of _Icnmodule.c 1.5, icnsupport.py 1.7:
Weaklink most toolbox modules, improving backward compatibility. Modules will no longer fail to load if a single routine is missing on the curent OS version, in stead calling the missing routine will raise an exception.

Should finally fix 531398. 2.2.1 candidate.

Also blacklisted some constants with definitions that were not Python-compatible.

23 years agoBackport of _Qdoffsmodule.c 1.7, qdoffssupport.py 1.8:
Jack Jansen [Mon, 25 Mar 2002 10:43:35 +0000 (10:43 +0000)] 
Backport of _Qdoffsmodule.c 1.7, qdoffssupport.py 1.8:
Weaklink most toolbox modules, improving backward compatibility. Modules will no longer fail to load if a single routine is missing on the curent OS version, in stead calling the missing routine will raise an exception.

Should finally fix 531398. 2.2.1 candidate.

Also blacklisted some constants with definitions that were not Python-compatible.

23 years agoBackport of _TEmodule.c 1.8, tesupport.py 1.10:
Jack Jansen [Mon, 25 Mar 2002 10:41:18 +0000 (10:41 +0000)] 
Backport of _TEmodule.c 1.8, tesupport.py 1.10:
Weaklink most toolbox modules, improving backward compatibility. Modules will no longer fail to load if a single routine is missing on the curent OS version, in stead calling the missing routine will raise an exception.

Should finally fix 531398. 2.2.1 candidate.

Also blacklisted some constants with definitions that were not Python-compatible.

23 years agoBackport of _Qtmocule.c 1.7, qtsupport.py 1.19:
Jack Jansen [Mon, 25 Mar 2002 10:38:57 +0000 (10:38 +0000)] 
Backport of _Qtmocule.c 1.7, qtsupport.py 1.19:
Weaklink most toolbox modules, improving backward compatibility. Modules will no longer fail to load if a single routine is missing on the curent OS version, in stead calling the missing routine will raise an exception.

Should finally fix 531398. 2.2.1 candidate.

Also blacklisted some constants with definitions that were not Python-compatible.

23 years agoBackport of _Appmodule.c 1.10, appsupport.py 1.14:
Jack Jansen [Mon, 25 Mar 2002 10:30:36 +0000 (10:30 +0000)] 
Backport of _Appmodule.c 1.10, appsupport.py 1.14:
Weaklink most toolbox modules, improving backward compatibility. Modules will no longer fail to load if a single routine is missing on the curent OS version, in stead calling the missing routine will raise an exception.

Should finally fix 531398. 2.2.1 candidate.

Also blacklisted some constants with definitions that were not Python-compatible.

23 years agoBackport of 1.20:
Jack Jansen [Mon, 25 Mar 2002 10:25:56 +0000 (10:25 +0000)] 
Backport of 1.20:
Weaklink most toolbox modules, improving backward compatibility. Modules will no longer fail to load if a single routine is missing on the curent OS version, in stead calling the missing routine will raise an exception.

Should finally fix 531398. 2.2.1 candidate.

Also blacklisted some constants with definitions that were not Python-compatible.

23 years agoBackport of 1.7:
Jack Jansen [Mon, 25 Mar 2002 10:23:56 +0000 (10:23 +0000)] 
Backport of 1.7:
Weaklink most toolbox modules, improving backward compatibility. Modules will no longer fail to load if a single routine is missing on the curent OS version, inAlso blacklisted some constants with definitions that were not Python-compatible.

23 years agoBackport of _AEmodule.c 1.10 and aesupport.py 1.26:
Jack Jansen [Mon, 25 Mar 2002 10:21:42 +0000 (10:21 +0000)] 
Backport of _AEmodule.c 1.10 and aesupport.py 1.26:
Weaklink most toolbox modules, improving backward compatibility. Modules will no longer fail to load if a single routine is missing on the curent OS version, inAlso blacklisted some constants with definitions that were not Python-compatible.

23 years agoFix broken HTML in the head; this was reported by a user as causing Opera 6.01
Fred Drake [Fri, 22 Mar 2002 17:23:03 +0000 (17:23 +0000)] 
Fix broken HTML in the head; this was reported by a user as causing Opera 6.01
to crash.  The user has reported the problem to Opera, but we still should
generate something that passes for HTML.

23 years agoI was just testing you. Neil Norwitz gets the prize. (Fixed a typo
Barry Warsaw [Fri, 22 Mar 2002 16:41:18 +0000 (16:41 +0000)] 
I was just testing you.  Neil Norwitz gets the prize.  (Fixed a typo
in the description of the email package patch.)

23 years agoAdded a NEWS item for the fix of SF bug #531966 in the email package.
Barry Warsaw [Fri, 22 Mar 2002 16:25:54 +0000 (16:25 +0000)] 
Added a NEWS item for the fix of SF bug #531966 in the email package.

Also, added a header for 2.2.1c2 (if there isn't a c2, change this to
2.2.1 final).

23 years ago_handle_multipart(): Fixes for SF bug #531966. Specifically two
Barry Warsaw [Fri, 22 Mar 2002 16:21:56 +0000 (16:21 +0000)] 
_handle_multipart(): Fixes for SF bug #531966.  Specifically two
situations are handled now: a multipart/* containing no payload
(i.e. never set), and a multipart/* containing a scalar payload
(i.e. Message.add_payload() having been called exactly once, not
passing in a sequence object).

_make_boundary(): Fixed bogus cut-n-paste error (self as first arg).

I will merge these changes into the standalone email package and
Python 2.3 separately.

23 years agotest_no_parts_in_a_multipart(): A test for the layout of a
Barry Warsaw [Fri, 22 Mar 2002 16:19:30 +0000 (16:19 +0000)] 
test_no_parts_in_a_multipart(): A test for the layout of a
multipart/mixed message with no attachments.

test_one_part_in_a_multipart(): A test for the layout of a
multipart/mixed message with a single attachment.

test_seq_parts_in_a_multipart(): A test for the layout of a
multipart/mixed message with a single attachment that happens to be a
sequence of length one.

These tests ensure no regressions on the fix for SF bug #531966.

I will merge these into the standalone email package and Python 2.3
trunk separately.

23 years agoBackport of _Resmodule.c 1.10 and ressupport.py 1.21:
Jack Jansen [Fri, 22 Mar 2002 15:48:44 +0000 (15:48 +0000)] 
Backport of _Resmodule.c 1.10 and ressupport.py 1.21:
Generate with weaklink stubs, so missing routines (on MacOS 8.6 and earlier)
don't cause import failure.

Fixes 531398, 2.2.1 candidate.

23 years agoI forgot to regenerate this. Oops.
Michael W. Hudson [Fri, 22 Mar 2002 11:22:24 +0000 (11:22 +0000)] 
I forgot to regenerate this.  Oops.

23 years agoMerge of 1.13: checked in to convert from BinHex to AppleSingle.
Jack Jansen [Thu, 21 Mar 2002 20:19:30 +0000 (20:19 +0000)] 
Merge of 1.13: checked in to convert from BinHex to AppleSingle.

23 years agoMerge rev. 1.5: Updated to new Universal Headers.
Jack Jansen [Thu, 21 Mar 2002 14:56:48 +0000 (14:56 +0000)] 
Merge rev. 1.5: Updated to new Universal Headers.

Also, this checkin should make the file MacBinary again.

23 years agoSF# 522426, add doc for common parameter for filecmp.cmpfiles()
Neal Norwitz [Wed, 20 Mar 2002 18:54:03 +0000 (18:54 +0000)] 
SF# 522426, add doc for common parameter for filecmp.cmpfiles()

23 years agoAdjust some poor wording in the text that explains what events are used
Fred Drake [Tue, 19 Mar 2002 14:37:14 +0000 (14:37 +0000)] 
Adjust some poor wording in the text that explains what events are used
for (reported by Keith Briggs).
Wrap some very long lines.

23 years agosync() has different return values depending on whether we build with GUSI or not.
Jack Jansen [Tue, 19 Mar 2002 11:01:20 +0000 (11:01 +0000)] 
sync() has different return values depending on whether we build with GUSI or not.

23 years agoFiles used for 2.2.1c1 distribution.
Jack Jansen [Tue, 19 Mar 2002 10:59:38 +0000 (10:59 +0000)] 
Files used for 2.2.1c1 distribution.

23 years agochar/unsigned char mixup.
Jack Jansen [Tue, 19 Mar 2002 10:59:26 +0000 (10:59 +0000)] 
char/unsigned char mixup.

23 years agoClarify that copy_reg.pickle() is not intended for use with "classic" classes.
Fred Drake [Tue, 19 Mar 2002 03:33:14 +0000 (03:33 +0000)] 
Clarify that copy_reg.pickle() is not intended for use with "classic" classes.
This was stated before, but a minor grammatical error made it difficult to be
sure of the meaning.
This closes SF bug #530143.

23 years agoRecord 2.2.1c1 release date.
Tim Peters [Tue, 19 Mar 2002 02:13:47 +0000 (02:13 +0000)] 
Record 2.2.1c1 release date.

23 years agoTo make 'urllib.py -t' run again, change FTP URL to a file that actually
Andrew M. Kuchling [Mon, 18 Mar 2002 22:19:24 +0000 (22:19 +0000)] 
To make 'urllib.py -t' run again, change FTP URL to a file that actually
   exists.

23 years agoThis commit was manufactured by cvs2svn to create tag 'r221c1'. v2.2.1c1
cvs2svn [Mon, 18 Mar 2002 16:47:35 +0000 (16:47 +0000)] 
This commit was manufactured by cvs2svn to create tag 'r221c1'.

23 years agoFix up unescaped tilde; reported by several people.
Fred Drake [Mon, 18 Mar 2002 16:47:35 +0000 (16:47 +0000)] 
Fix up unescaped tilde; reported by several people.

23 years agoRemove extra verb; reported by Detlef Lannert.
Fred Drake [Mon, 18 Mar 2002 16:44:46 +0000 (16:44 +0000)] 
Remove extra verb; reported by Detlef Lannert.

23 years agoBackport of _CFmodule.c 1.11 and _Winmodule 1.8:
Jack Jansen [Mon, 18 Mar 2002 15:54:29 +0000 (15:54 +0000)] 
Backport of _CFmodule.c 1.11 and _Winmodule 1.8:
Lurking bug found by patch for 531291: FSSpecs should be passed to
Py_BuildValue by address, not by value.

23 years agoBackport of _Resmodule.c 1.10 and ressupport.py 1.20:
Jack Jansen [Mon, 18 Mar 2002 15:38:24 +0000 (15:38 +0000)] 
Backport of _Resmodule.c 1.10 and ressupport.py 1.20:
Some of the newer routines return an OSErr, like most toolbox calls,
in stead of depending on ResError(). Second half of fix for #531291.

23 years agoMake StringIO work in non-unicode builds.
Michael W. Hudson [Mon, 18 Mar 2002 13:31:31 +0000 (13:31 +0000)] 
Make StringIO work in non-unicode builds.

Lots of tests fail in non-unicode builds, but I think most of these are
"bugs" in the tests.  I hope so, anyway.

23 years agobackport jackjansen's checkin of
Michael W. Hudson [Mon, 18 Mar 2002 13:10:41 +0000 (13:10 +0000)] 
backport jackjansen's checkin of
    revision 1.2 of ERRNO.py
    revision 1.2 of FILE.py
    revision 1.2 of IN.py

Regenerated for Irix 6.5.

23 years agobackport jackjansen's checkin of
Michael W. Hudson [Mon, 18 Mar 2002 13:10:13 +0000 (13:10 +0000)] 
backport jackjansen's checkin of
    revision 1.3 of regen

Patch by Michael Pruett: make regen work on Irix 6.0 and 6.5.

23 years agoamk's fix attached to
Michael W. Hudson [Mon, 18 Mar 2002 13:03:40 +0000 (13:03 +0000)] 
amk's fix attached to

[ 516299 ] urlparse can get fragments wrong

23 years agoRemove extraneous #define as per effbot's instructions in:
Michael W. Hudson [Mon, 18 Mar 2002 12:59:38 +0000 (12:59 +0000)] 
Remove extraneous #define as per effbot's instructions in:

[ 530285 ] redefining SRE_CODE in Modules/sre.h

Another one for the trunk, later.

23 years agodoerwalter's failing examples from
Michael W. Hudson [Mon, 18 Mar 2002 12:54:51 +0000 (12:54 +0000)] 
doerwalter's failing examples from

[ 529104 ] broken error handling in unicode-escape

23 years agoMartin's fix for
Michael W. Hudson [Mon, 18 Mar 2002 12:47:52 +0000 (12:47 +0000)] 
Martin's fix for

[ 529104 ] broken error handling in unicode-escape

I presume this will need to be fixed on the trunk, too.

Later.

23 years agoFix
Michael W. Hudson [Mon, 18 Mar 2002 12:43:33 +0000 (12:43 +0000)] 
Fix

[ 531306 ] ucs4 build horked.

Classic C mistake, I think.

Also squashed a couple of warnings in the ucs4 build.

23 years agoFinished 2.2.1 release notes.
Jack Jansen [Mon, 18 Mar 2002 10:44:49 +0000 (10:44 +0000)] 
Finished 2.2.1 release notes.

23 years agoFix for SF bug 528132 (Armin Rigo): classmethod().__get__() segfault
Guido van Rossum [Mon, 18 Mar 2002 03:05:36 +0000 (03:05 +0000)] 
Fix for SF bug 528132 (Armin Rigo): classmethod().__get__() segfault

The proper fix is not quite what was submitted; it's really better to
take the class of the object passed rather than calling PyMethod_New
with NULL pointer args, because that can then cause other core dumps
later.

I also added a testcase for the fix to classmethods() in test_descr.py.

I'll apply this to 2.3 too.

23 years agoA first cut at NEWS for 2.2.1c1.
Michael W. Hudson [Sun, 17 Mar 2002 20:47:11 +0000 (20:47 +0000)] 
A first cut at NEWS for 2.2.1c1.

Please comment!  You can find logs at

http://starship.python.net/crew/mwh/cvslog3.txt

on the assumption that you don't want to fight cvs yourself.

In particular, I need a better explanation of the bugs that have
been fixed in the email package.

23 years agoBackport a checkin of jvr's:
Michael W. Hudson [Sun, 17 Mar 2002 19:47:39 +0000 (19:47 +0000)] 
Backport a checkin of jvr's:

on MacOSX/Darwin, use ranlib when building static libs.

I hope this belongs on the branch...

23 years agoBackport a checkin of lemburg's:
Michael W. Hudson [Sun, 17 Mar 2002 19:42:18 +0000 (19:42 +0000)] 
Backport a checkin of lemburg's:

Remove mentioning of -U option in "python -h" output.

23 years agoAnother Guido backport:
Michael W. Hudson [Sun, 17 Mar 2002 19:39:24 +0000 (19:39 +0000)] 
Another Guido backport:

Quick build: clarify that you have to do "make install" as root; OS
info: add info about Red Hat's python and python2.

23 years agoBackport a checkin of Guido's:
Michael W. Hudson [Sun, 17 Mar 2002 19:36:48 +0000 (19:36 +0000)] 
Backport a checkin of Guido's:

Remove stub for unicode.txt.  Resort README in dictionary order.

23 years agoBackport nnorwitz's checkin of revision 1.119:
Michael W. Hudson [Sun, 17 Mar 2002 19:31:28 +0000 (19:31 +0000)] 
Backport nnorwitz's checkin of revision 1.119:

Fix typo

23 years agoBackport montanero's checkin of revision 1.7:
Michael W. Hudson [Sun, 17 Mar 2002 19:05:18 +0000 (19:05 +0000)] 
Backport montanero's checkin of revision 1.7:

update text to refer to ServerProxy class in preference to Server, which is
only retained for backward compatibility with older versions of the library.

23 years agoBackport my fix from a whiles back:
Michael W. Hudson [Sun, 17 Mar 2002 19:02:10 +0000 (19:02 +0000)] 
Backport my fix from a whiles back:

Fix for

[ #504284 ] Last build problems on AIX

I'm ignoring the suggestion that this should be an autoconf test in the
interests of having a fix today.  Feel free to quibble.

23 years agoBackport a little test from Skip.
Michael W. Hudson [Sun, 17 Mar 2002 18:59:32 +0000 (18:59 +0000)] 
Backport a little test from Skip.

23 years agoSF patch 530070: pydoc regression, from Martin and Guido.
Tim Peters [Sun, 17 Mar 2002 18:57:07 +0000 (18:57 +0000)] 
SF patch 530070: pydoc regression, from Martin and Guido.
Change the way __doc__ is handled, to avoid blowing up on non-string
__doc__ values.

23 years agoBackport Tim's checkin of revision 1.5:
Michael W. Hudson [Sun, 17 Mar 2002 18:05:03 +0000 (18:05 +0000)] 
Backport Tim's checkin of revision 1.5:

This test left a new set of 3 junk files behind each time it was run.

23 years agoStop using not-exposed-in-22x os.O_ constants.
Michael W. Hudson [Sun, 17 Mar 2002 18:02:51 +0000 (18:02 +0000)] 
Stop using not-exposed-in-22x os.O_ constants.

23 years agoSo there is some merit in slogging through ~4800 lines of cvs log.
Michael W. Hudson [Sun, 17 Mar 2002 17:54:32 +0000 (17:54 +0000)] 
So there is some merit in slogging through ~4800 lines of cvs log.

Bring tempfile.py up to date from the trunk.  There have been three
checkins (all by Tim):

SF bug #509805 tempfile.gettempdir not threadsafe
This is an ancient race when multiple threads call gettempdir() (or
anything relying on it) for the first time.

Fixed x-platform via the Big Hammer of rearranging the code to serialize
the first calls.  Subsequent calls are as fast as before.

Note that the Python test suite can't provoke this bug:  it requires
setting up multiple threads making the very first calls into tempfile,
but the test suite uses tempfile several times before getting to
test_threadedtempfile.

Bugfix candidate.

[and]

New TemporaryFile implementation for Windows:  this doesn't need a
TemproraryFileWrapper wrapper anymore, and should be immune from the
problem that a temp file inherited by a spawned process caused an
attempt to close the temp file in the spawning process to blow
up (the unlink in TemporaryFileWrapper.close() blew up with a
"Permission denied" error because, despite that the temp file got
closed in the spawning process, the spawned process still had it open
by virtue of C-level file descriptor inheritance).  In context,
that bug took days to figure out <wink/sigh>.

[and]

Thanks to Detlef Lannert for pointing out a typo in the code that
uses _DummyMutex on platforms without threads.

The first and third of these are pretty clearly bugfixes; I think the
second is too.

23 years agoTake Tim's work on file.truncate out of 2.2.1 again.
Michael W. Hudson [Sun, 17 Mar 2002 15:55:50 +0000 (15:55 +0000)] 
Take Tim's work on file.truncate out of 2.2.1 again.

23 years agoMerge jackjansen's checking of revision 1.5.
Michael W. Hudson [Sat, 16 Mar 2002 18:33:31 +0000 (18:33 +0000)] 
Merge jackjansen's checking of revision 1.5.

23 years agoBackport Tim's work on getting file.truncate working better on Win32.
Michael W. Hudson [Sat, 16 Mar 2002 18:19:33 +0000 (18:19 +0000)] 
Backport Tim's work on getting file.truncate working better on Win32.

"cvs diff | patch" managed to stick the NEWS item in the 2.2 final
section!  I wonder which silly man wrote patch <wink>.

23 years agobackport bwarsaw's checkin of
Michael W. Hudson [Sat, 16 Mar 2002 18:03:42 +0000 (18:03 +0000)] 
backport bwarsaw's checkin of
    revision 4.7 of python-mode.el

(py-honor-comment-indentation, py-compute-indentation): Fix the
implementation to match the documentation for
py-honor-comment-indentation w.r.t. not nil or t value.  In that case
it should still ignore ## for indentation purposes.  Closes SF bug
#523825, w/ patch provided by Christian Stork (mod'd by Barry).

Python 2.2.1 candidate.

23 years agobackport my checkin of
Michael W. Hudson [Sat, 16 Mar 2002 18:02:20 +0000 (18:02 +0000)] 
backport my checkin of
    revision 1.54 of os.py

Fix

[ 530236 ] os.py assumes existence of statvfs_resul

This was pretty dense of me.  Sorry.

2.2.1 candidate.

23 years agoThis checkin backport two checkins by Skip.
Michael W. Hudson [Sat, 16 Mar 2002 18:01:05 +0000 (18:01 +0000)] 
This checkin backport two checkins by Skip.

backport montanaro's checkin of
    revision 1.24 of calendar.py

make _localized_name instances work more like the tuples they replaced.  In
particular, negative indexes work and they are limited by the actual length
of the names they represent (weekday and month names).  This closes bug
#503202.

[and then]

Corrected _localized_name.__getitem__ based on code in patch 503202 (which I
thought was just a bug report, so didn't notice - doh!).  This handles
slicing, which v 1.23 didn't.

23 years agobackport loewis' checkin of
Michael W. Hudson [Sat, 16 Mar 2002 17:58:21 +0000 (17:58 +0000)] 
backport loewis' checkin of
    revision 2.23 of pypcre.c

Include Python.h first. Fixes #530159.

23 years agobackport gvanrossum's checkin of
Michael W. Hudson [Sat, 16 Mar 2002 17:57:26 +0000 (17:57 +0000)] 
backport gvanrossum's checkin of
    revision 1.121 of test_descr.py

"Fix" for SF bug #520644: __slots__ are not pickled.

As promised in my response to the bug report, I'm not really fixing
it; in fact, one could argule over what the proper fix should do.
Instead, I'm adding a little magic that raises TypeError if you try to
pickle an instance of a class that has __slots__ but doesn't define or
override __getstate__.  This is done by adding a bozo __getstate__
that always raises TypeError.

Bugfix candidate (also the checkin to typeobject.c, of course).

23 years agobackport gvanrossum's checkin of
Michael W. Hudson [Sat, 16 Mar 2002 17:56:51 +0000 (17:56 +0000)] 
backport gvanrossum's checkin of
    revision 2.129 of typeobject.c

"Fix" for SF bug #520644: __slots__ are not pickled.

As promised in my response to the bug report, I'm not really fixing
it; in fact, one could argule over what the proper fix should do.
Instead, I'm adding a little magic that raises TypeError if you try to
pickle an instance of a class that has __slots__ but doesn't define or
override __getstate__.  This is done by adding a bozo __getstate__
that always raises TypeError.

23 years agobackport loewis' checkin of
Michael W. Hudson [Sat, 16 Mar 2002 17:54:20 +0000 (17:54 +0000)] 
backport loewis' checkin of
    revision 2.26 of _localemodule.c

Verify arguments for nl_langinfo. Fixes #528879.

23 years agoNaughty Fred forgot to mark this as a bugfix candidate.
Michael W. Hudson [Sat, 16 Mar 2002 17:53:51 +0000 (17:53 +0000)] 
Naughty Fred forgot to mark this as a bugfix candidate.

backport fdrake's checkin of
    revision 1.5 of log.py

Set/update self.cwd properly.

23 years agoFix stupid typo in example.
Fred Drake [Sat, 16 Mar 2002 13:53:23 +0000 (13:53 +0000)] 
Fix stupid typo in example.

23 years agoMarkup error: braces not properly marked in dictionary display
Fred Drake [Sat, 16 Mar 2002 06:35:01 +0000 (06:35 +0000)] 
Markup error: braces not properly marked in dictionary display
grammer productions.
Fixes SF bug #520959.

23 years agoClarify the descriptions of the positive and negative lookbehind assertions.
Fred Drake [Sat, 16 Mar 2002 05:10:22 +0000 (05:10 +0000)] 
Clarify the descriptions of the positive and negative lookbehind assertions.
Added examples of positive lookbehind assertions.
This closes SF bug #529708.

23 years agopdfTeX 0.13 and 0.14 do not agree on the name of the macro, and I will not
Fred Drake [Sat, 16 Mar 2002 04:52:55 +0000 (04:52 +0000)] 
pdfTeX 0.13 and 0.14 do not agree on the name of the macro, and I will not
change the installed version on either of the machines I use to format the
docs.  Instead, use a compatibility hack to support both versions.  This is
also better for external users of the Python styles.

23 years agoRevise the markup related to the grammar productions to increase the
Fred Drake [Fri, 15 Mar 2002 23:18:05 +0000 (23:18 +0000)] 
Revise the markup related to the grammar productions to increase the
level of predictability.  This is not really "good" markup, but is arguably
better than we had before.
This closes SF bug #523117.