Tim Peters [Mon, 7 Jun 2004 23:17:48 +0000 (23:17 +0000)]
SF 952807: Unpickling pickled instances of subclasses of
datetime.date, datetime.datetime and datetime.time could yield insane
objects. Thanks to Jiwon Seo for a fix.
Kurt B. Kaiser [Sun, 6 Jun 2004 17:41:41 +0000 (17:41 +0000)]
Backporting [ 778323 ] Tk Dialog Upon Subprocess Socket Error
Added a Tk error dialog to run.py inform the user if the subprocess can't
connect to the user GUI process. Added a timeout to the GUI's listening
socket. Added Tk error dialogs to PyShell.py to announce a failure to bind the
port or connect to the subprocess. Clean up error handling during connection
initiation phase. This is an update of Python Patch 778323.
M NEWS.txt
M PyShell.py
M ScriptBinding.py
M idlever.py
M run.py
[Bug #954364] inspect.getframeinfo() sometimes produces incorrect traceback line #s; fix is to look at tb.tb_lineno, not tb.frame.f_lineno. Patch from Robin Becker and me.
Jack Jansen [Thu, 3 Jun 2004 21:36:58 +0000 (21:36 +0000)]
Backport of 1.45 and 1.46:
- Replace backticks with repr() or "%r"
- Fix for #862941: "run with commandline python" crashed. Fixed.
- Prefer to use pythonw (if it exists) for "run with commandline python".
Jack Jansen [Thu, 3 Jun 2004 14:38:38 +0000 (14:38 +0000)]
Backport of 1.47:
Fix for #932977: MacOSX does not pass the whole pathname in argv[0] for
#!-scripts, only the filename part, and this can lead to incorrect
initialization of sys.path and sys.executable if there is another python
on $PATH before the one used in #!.
The fix was picked up from the darwinports crowd, thanks!
Jack Jansen [Thu, 3 Jun 2004 13:34:41 +0000 (13:34 +0000)]
Backport of 1.23 and 1.24:
Very large scripts folders could crash the IDE, because it runs out
of Menu IDs (of which there are only 255 in Carbon). Fixed by stopping
examining the scripts folder when we allocate menu ID 200.
Jack Jansen [Thu, 3 Jun 2004 13:07:39 +0000 (13:07 +0000)]
Backport of 1.31-1.34:
- Added a downloader using urllib2 in stead of curl, based on code
donated by Kevin Ollivier. This is now the default downloader.
- Added a watcher mechanism, whereby downloaders and unpackers (and,
later builders) can give status feedback to the user. When running
pimp as a command line tool in verbose mode print this output.
- Force option should be applied to a single package, not recursively
to its dependencies. Fixes #733819.
- Don't use "dict" as a variable, it shadows the builtin. Spotted by
Bob Ippolito.
Two issues spotted by Ronald OUssoren:
- there were no accessor functions for the global per-database fields
- packages and their dependencies were installed in order in stead
of in reverse order.
Greg Ward [Thu, 3 Jun 2004 01:53:13 +0000 (01:53 +0000)]
SF #965425: fix wordsep_re so hyphenated words are handled correctly
when preceded by any punctuation, not just a sequence of more hyphens.
(That was a special case so Optik and Docutils could wrap long options
like --foo-bar correctly; this change generalizes the special-case.)
Greg Ward [Thu, 3 Jun 2004 01:51:27 +0000 (01:51 +0000)]
SF #965425: add WrapTestCase.test_punct_hyphens() to ensure that
hyphenated words wrapped in punctuation, like "foo-bar" or [ding-dong],
are split correctly.
Gregory P. Smith [Sat, 15 May 2004 21:37:34 +0000 (21:37 +0000)]
Remove the offending added bit of a test case that doesn't apply to
the python 2.3.x branch as the bsddb.set_get_returns_none() default
behaviour is different than on 2.4 and leter. (yes a correct test
case for this branch could be written; removing this bad test gets
things not-failing fastest)
Barry Warsaw [Thu, 13 May 2004 23:14:31 +0000 (23:14 +0000)]
Added a sample message for the test for SF bug # 846938. I'm naming this
msg_40.txt because msg_3[6-9].txt are already used in email3/py2.4 and I'm
going to forward port this test.
Barry Warsaw [Thu, 13 May 2004 23:12:33 +0000 (23:12 +0000)]
_parsebody(): Do not create subparts unless the container has a main type of
'multipart' and the boundary is defined. This fixes SF bug # 846938, and
several recent email-sig bugs where something like a text/html message also
had a boundary parameter. This would later crash the Generator, which only
consulted the Content-Type to decide how to generate the message (and it would
expect just a string, but find a list there instead).
Greg Ward [Thu, 13 May 2004 01:45:33 +0000 (01:45 +0000)]
SF #847346: add test_fix_sentence_endings() to ensure that sentence
endings get fixed even if the input is too short to wrap. Also throw
in a bunch of other tests for this rather tricky and under-tested
feature.
In WrapTestCase.setUp(), don't bother setting fix_sentence_endings=True
for self.wrapper -- just use a separate wrapper in the two tests that
care about it, namely test_whitespace() and test_fix_sentence_endings().
Thomas Heller [Wed, 12 May 2004 20:04:18 +0000 (20:04 +0000)]
Version '2.3.4c1'
File name 'Python-2.3.4c1.exe'
Change company to 'Python Software Foundation'
Add year 2004 to the copyright.
Include tix into the distribution.
Tim Peters [Tue, 11 May 2004 18:11:11 +0000 (18:11 +0000)]
Update copyright years. In the Windows resource, change the name of the
company to the PSF (Zope Corp and PythonLabs no longer have anything to
do with this!).
Barry Warsaw [Tue, 11 May 2004 02:07:03 +0000 (02:07 +0000)]
In order to fix SF bug # 824977, we replace calloc()/free() calls in
binascii_a2b_qp() and binascii_b2a_qp() with calls to PyMem_Malloc() and
PyMem_Free(). These won't return NULL unless the allocations actually fail,
so it won't trigger a bogus memory error on some platforms <cough>AIX</cough>
when passed a length of zero.