]> granicus.if.org Git - python/log
python
23 years agomoving to new config system
Steven M. Gava [Thu, 3 Jan 2002 11:51:07 +0000 (11:51 +0000)]
moving to new config system

23 years agoIn MachoPython, don't simulate argc/argv unless we have argc=1 and argv[1]
Jack Jansen [Wed, 2 Jan 2002 22:53:38 +0000 (22:53 +0000)]
In MachoPython, don't simulate argc/argv unless we have argc=1 and argv[1]
starts with "-psn_". This means the drag-and-drop continues to work as
expected, but we can now also do
/Applications/Python.app/Contents/MacOS/python script.py
from the command line, which is a lot easier with debugging. Pressing <option>
at this point also has the expected effect of opening the options dialog!

23 years agoFix typo in label
Andrew M. Kuchling [Wed, 2 Jan 2002 16:38:19 +0000 (16:38 +0000)]
Fix typo in label

23 years ago- Added DrawThemeTextBox()
Just van Rossum [Wed, 2 Jan 2002 15:11:44 +0000 (15:11 +0000)]
- Added DrawThemeTextBox()
- fixed GetThemeTextDimensions(): it has an in/out Point arg, not just out.

23 years agorectified copy/paste oversight in comment
Just van Rossum [Wed, 2 Jan 2002 14:59:03 +0000 (14:59 +0000)]
rectified copy/paste oversight in comment

23 years ago- Added support for optional MenuObj arguments
Just van Rossum [Wed, 2 Jan 2002 14:48:36 +0000 (14:48 +0000)]
- Added support for optional MenuObj arguments
- Added a bunch of calls as functions with an optional
  MenuObj first argument. The same calls already
  exist as methods, but then the first arg isn't
  optional... The method versions could go as far as I'm
  concerned. Jack?

23 years agoAdded some support for unicode arguments.
Jack Jansen [Tue, 1 Jan 2002 22:43:13 +0000 (22:43 +0000)]
Added some support for unicode arguments.

23 years agoAdded some support for Uicode arguments. Still leaves something to be desired, especi...
Jack Jansen [Tue, 1 Jan 2002 22:40:08 +0000 (22:40 +0000)]
Added some support for Uicode arguments. Still leaves something to be desired, especially because of poor unicode support in PyArg_ParseTuple.

23 years agoUpdated to current pathnames.
Jack Jansen [Tue, 1 Jan 2002 22:39:07 +0000 (22:39 +0000)]
Updated to current pathnames.

23 years agoImprove grammar
Neal Norwitz [Tue, 1 Jan 2002 21:14:12 +0000 (21:14 +0000)]
Improve grammar

23 years agoPatch #497126: Always compile dl.
Martin v. Löwis [Tue, 1 Jan 2002 20:18:30 +0000 (20:18 +0000)]
Patch #497126: Always compile dl.

23 years agoPatch #494783: Rename cmp_op enumerators.
Martin v. Löwis [Tue, 1 Jan 2002 19:59:11 +0000 (19:59 +0000)]
Patch #494783: Rename cmp_op enumerators.

23 years agoSF Patch #494867 test file methods
Neal Norwitz [Tue, 1 Jan 2002 19:11:13 +0000 (19:11 +0000)]
SF Patch #494867 test file methods

Test that the file methods raise ValueError when called on a closed file.
Test .isatty()
Test name, closed attributes

23 years agoSF Patch #494863, file.xreadlines() should raise ValueError if file is closed
Neal Norwitz [Tue, 1 Jan 2002 19:07:13 +0000 (19:07 +0000)]
SF Patch #494863, file.xreadlines() should raise ValueError if file is closed

This makes xreadlines behave like all other file methods
(other than close() which just returns).

23 years agoPatch #497098: build support for GNU/Hurd.
Martin v. Löwis [Tue, 1 Jan 2002 18:41:33 +0000 (18:41 +0000)]
Patch #497098: build support for GNU/Hurd.

23 years agoUse the Carbon scrap manager interface if the old interface isn't available.
Jack Jansen [Mon, 31 Dec 2001 15:08:04 +0000 (15:08 +0000)]
Use the Carbon scrap manager interface if the old interface isn't available.

23 years agoUpdated to optionally use Carbon Scrap manager.
Jack Jansen [Mon, 31 Dec 2001 15:02:56 +0000 (15:02 +0000)]
Updated to optionally use Carbon Scrap manager.

23 years agoUse the Carbon scrap manager interface if the old interface isn't available.
Jack Jansen [Mon, 31 Dec 2001 14:53:05 +0000 (14:53 +0000)]
Use the Carbon scrap manager interface if the old interface isn't available.

23 years agoAdded support for the Carbon scrap manager (finally).
Jack Jansen [Mon, 31 Dec 2001 14:52:03 +0000 (14:52 +0000)]
Added support for the Carbon scrap manager (finally).

23 years agoOops, forgot to mark CreateScrollBarControl and CreateSliderControl
Just van Rossum [Mon, 31 Dec 2001 09:50:32 +0000 (09:50 +0000)]
Oops, forgot to mark CreateScrollBarControl and CreateSliderControl
as Carbon-only

23 years agomade radio button labels readable under OSX
Just van Rossum [Mon, 31 Dec 2001 08:58:44 +0000 (08:58 +0000)]
made radio button labels readable under OSX

23 years agohalf-hearted stab at supported the 2.2 object model better.
Just van Rossum [Mon, 31 Dec 2001 08:57:57 +0000 (08:57 +0000)]
half-hearted stab at supported the 2.2 object model better.

23 years agoadded support for live feedback in scrollbars
Just van Rossum [Mon, 31 Dec 2001 08:56:52 +0000 (08:56 +0000)]
added support for live feedback in scrollbars

23 years ago- added support for ControlActionProcs, exposing the following calls:
Just van Rossum [Sun, 30 Dec 2001 21:25:26 +0000 (21:25 +0000)]
- added support for ControlActionProcs, exposing the following calls:
    - ctl.SetControlAction()
    - CreateScrollBarControl()
    - CreateSliderControl()
- print traceback when callbacks fail

23 years agoPrint encoded versions of the file names in test application. Fixes #496084
Martin v. Löwis [Sun, 30 Dec 2001 14:43:56 +0000 (14:43 +0000)]
Print encoded versions of the file names in test application. Fixes #496084

23 years agotest_support: add a docstring to vereq().
Tim Peters [Sat, 29 Dec 2001 17:34:57 +0000 (17:34 +0000)]
test_support:  add a docstring to vereq().
test_complex:  repair new test's usage of vereq().

23 years agoSF Patch #497487 add test to compare conjugate of a complex number
Neal Norwitz [Sat, 29 Dec 2001 14:31:46 +0000 (14:31 +0000)]
SF Patch #497487 add test to compare conjugate of a complex number

23 years agoSF Patch #494873 add tests for complex numbers including calls to int()/long()
Neal Norwitz [Sat, 29 Dec 2001 01:02:21 +0000 (01:02 +0000)]
SF Patch #494873 add tests for complex numbers including calls to int()/long()

23 years agoSF Patch #494874 add tests for int()/long() invalid parameters
Neal Norwitz [Sat, 29 Dec 2001 00:35:20 +0000 (00:35 +0000)]
SF Patch #494874 add tests for int()/long() invalid parameters

23 years agoSF Patch #494872 test repr() of a built-in module
Neal Norwitz [Sat, 29 Dec 2001 00:25:42 +0000 (00:25 +0000)]
SF Patch #494872 test repr() of a built-in module

23 years agoSF Patch #494876, test invalid parameters to pow()
Neal Norwitz [Sat, 29 Dec 2001 00:16:09 +0000 (00:16 +0000)]
SF Patch #494876, test invalid parameters to pow()

23 years agoThread.__bootstrap(): ignore exceptions in the self.__delete() call in
Guido van Rossum [Fri, 28 Dec 2001 22:07:09 +0000 (22:07 +0000)]
Thread.__bootstrap(): ignore exceptions in the self.__delete() call in
the finally clause.  An exception here could happen when a daemon
thread exits after the threading module has already been trashed by
the import finalization, and there's not much of a point in trying to
insist doing the cleanup in that stage.

This should fix SF bug ##497111: active_limbo_lock error at program
exit.

2.1.2 and 2.2.1 Bugfix candidate!

23 years agoFix for SF bug ##497426: can't deepcopy recursive new objects
Guido van Rossum [Fri, 28 Dec 2001 21:39:03 +0000 (21:39 +0000)]
Fix for SF bug ##497426: can't deepcopy recursive new objects

deepcopy(), _reconstruct(): pass the memo to the other function, so
that recursive data structures built out of new-style objects may be
deeply copied correctly.

2.2.1 bugfix!

23 years agoFix for SF bug ##497426: can't deepcopy recursive new objects
Guido van Rossum [Fri, 28 Dec 2001 21:33:22 +0000 (21:33 +0000)]
Fix for SF bug ##497426: can't deepcopy recursive new objects

deepcopy(), _reconstruct(): pass the memo to the other function, so
that recursive data structures built out of new-style objects may be
deeply copied correctly.

2.2.1 bugfix!

23 years agoAdd fcntl.h constants from glibc 2.2.4. Fixes #496171.
Martin v. Löwis [Fri, 28 Dec 2001 21:08:12 +0000 (21:08 +0000)]
Add fcntl.h constants from glibc 2.2.4. Fixes #496171.

23 years agoSF patch #497420 (Eduardo Pérez): ftplib: ftp anonymous password
Guido van Rossum [Fri, 28 Dec 2001 20:57:14 +0000 (20:57 +0000)]
SF patch #497420 (Eduardo Pérez): ftplib: ftp anonymous password

Instead of sending the real user and host, use "anonymous@" (i.e. no
host name at all!) as the default anonymous FTP password.  This avoids
privacy violations.

23 years agoUpdate the docstring too. :-)
Guido van Rossum [Fri, 28 Dec 2001 20:54:55 +0000 (20:54 +0000)]
Update the docstring too. :-)

23 years agoSF patch #497420 (Eduardo Pérez): ftplib: ftp anonymous password
Guido van Rossum [Fri, 28 Dec 2001 20:54:28 +0000 (20:54 +0000)]
SF patch #497420 (Eduardo Pérez): ftplib: ftp anonymous password

Instead of sending the real user and host, use "anonymous@" (i.e. no
host name at all!) as the default anonymous FTP password.  This avoids
privacy violations.

23 years agoAdded someone.
Jack Jansen [Thu, 27 Dec 2001 23:37:49 +0000 (23:37 +0000)]
Added someone.

23 years agoPatches by Jens B. Jorgensen with small mods by me:
Jack Jansen [Thu, 27 Dec 2001 23:35:43 +0000 (23:35 +0000)]
Patches by Jens B. Jorgensen with small mods by me:
- Converted the templates to use ANSI C prototypes (finally!)
- Use re in stead of deprecated regex

23 years agoMerging changes from release22-branch.
Jack Jansen [Thu, 27 Dec 2001 23:01:18 +0000 (23:01 +0000)]
Merging changes from release22-branch.

23 years agoAnother name; should be added to Python 2.2.1.
Fred Drake [Thu, 27 Dec 2001 18:40:18 +0000 (18:40 +0000)]
Another name; should be added to Python 2.2.1.

23 years agoAdded some missing index entries, noted by L. Peter Deutsch.
Fred Drake [Thu, 27 Dec 2001 18:38:10 +0000 (18:38 +0000)]
Added some missing index entries, noted by L. Peter Deutsch.
This should be included in Python 2.2.1.

23 years agoUpdated -- all Demo subdirectories are once again listed here, in
Guido van Rossum [Thu, 27 Dec 2001 16:57:49 +0000 (16:57 +0000)]
Updated -- all Demo subdirectories are once again listed here, in
alphabetical order (!), and the obsolete 'extend' directory is no
longer mentioned.

This and the erasure of the extend directory are 2.2.1 bugfix
candidates (but only of you want to be thorough -- it's not like
anybody cares :-).

23 years agoRemoving this directory; it's no longer needed now that Misc/Makefile.pre.in
Guido van Rossum [Thu, 27 Dec 2001 16:48:27 +0000 (16:48 +0000)]
Removing this directory; it's no longer needed now that Misc/Makefile.pre.in
no longer exists.  Docs for distutils are plenty elsewhere.

23 years ago_reduce(): Avoid infinite recursion in the pickler when self.__class__
Guido van Rossum [Thu, 27 Dec 2001 16:27:28 +0000 (16:27 +0000)]
_reduce(): Avoid infinite recursion in the pickler when self.__class__
doesn't have the _HEAPTYPE flag set, e.g. for time.struct_time and
posix.stat_result.

This fixes the immediate symptoms of SF bug #496873 (cPickle /
time.struct_time loop), replacing the infinite loop with an exception.

23 years agoDue to a cut-and-paste error, the type object exported under the name
Guido van Rossum [Thu, 27 Dec 2001 16:23:28 +0000 (16:23 +0000)]
Due to a cut-and-paste error, the type object exported under the name
statvfs_result was in fact the stat_result type object. :-(

2.2.1 bugfix!

23 years agoModified version of patch #496882: echo SimpleStdin readline()
Just van Rossum [Thu, 27 Dec 2001 10:29:07 +0000 (10:29 +0000)]
Modified version of patch #496882: echo SimpleStdin readline()
input to stdout.

23 years agoElaborate the descriptions for onecmd(), precmd(), and postcmd() so they are
Fred Drake [Thu, 27 Dec 2001 05:10:18 +0000 (05:10 +0000)]
Elaborate the descriptions for onecmd(), precmd(), and postcmd() so they are
useful.

23 years agoFix wrongly-named formal parameters in three places: begin_y was used twice
Fred Drake [Wed, 26 Dec 2001 22:08:44 +0000 (22:08 +0000)]
Fix wrongly-named formal parameters in three places: begin_y was used twice
instead of begin_y and begin_x for derwin(), subpad(), and subwin().
Reported for derwin() by Eric Huss.

Added class annotations for the window methods so they would be properly
described in the index.

23 years agoAdded another name.
Fred Drake [Wed, 26 Dec 2001 22:08:35 +0000 (22:08 +0000)]
Added another name.

23 years agoAdded index entries similar to some recommended by Skip, and used the word
Fred Drake [Wed, 26 Dec 2001 20:06:40 +0000 (20:06 +0000)]
Added index entries similar to some recommended by Skip, and used the word
"interpolation" in the text, to make the string formatting material easier to
find.
This closes SF bug #487165.
Bugfix: this should be applied for Python 2.2.1.

23 years agoMake this do the right thing with entries which start with the percent sign,
Fred Drake [Wed, 26 Dec 2001 19:55:14 +0000 (19:55 +0000)]
Make this do the right thing with entries which start with the percent sign,
in response to Skip's comments in SF bug #487165.

Make use of string methods instead of string module functions in most places.
Add (and make the default) a way to collapse symbol entries into a single
"Symbols" section in the generated index.  This is similar to what makeindex
does, but does not include entries beginning with an underscore.

23 years agoMore index entries.
Fred Drake [Wed, 26 Dec 2001 19:48:43 +0000 (19:48 +0000)]
More index entries.

23 years agoClose an improperly-closed verbatim environment.
Fred Drake [Wed, 26 Dec 2001 16:53:48 +0000 (16:53 +0000)]
Close an improperly-closed verbatim environment.
This closes SF patch #496215.

Add a little more detail to the example that had not been closed.

Bugfix: this should be made part of 2.2.1.

23 years agoSF bug #495548: troublesome #define in pyport.h
Tim Peters [Tue, 25 Dec 2001 19:07:38 +0000 (19:07 +0000)]
SF bug #495548:  troublesome #define in pyport.h
Removed the ancient "#define ANY void".

Bugfix candidate?  Hard call.  The bug report claims the existence of
this #define creates conflicts with other packages, which is easy to
believe.  OTOH, some extension authors may still be relying on its
presence.  I'm afraid you can't win on this one.

23 years agoSF bug #496549 -Qnew and in-place division "/=".
Tim Peters [Tue, 25 Dec 2001 18:49:11 +0000 (18:49 +0000)]
SF bug #496549 -Qnew and in-place division "/=".

eval_frame():  Under -Qnew, INPLACE_DIVIDE wasn't getting handed off to
INPLACE_TRUE_DIVIDE (like BINARY_DIVIDE was getting handed off to
BINARY_TRUE_DIVIDE).

Bugfix candidate.

23 years agoDon't set passiveserver to 0 in connect(). See SF bug #495693.
Guido van Rossum [Sun, 23 Dec 2001 13:54:19 +0000 (13:54 +0000)]
Don't set passiveserver to 0 in connect().  See SF bug #495693.

This should definitely be backported to 2.2.1.  I'll leave it to Jack
to decide whether he wants to fix this in MacPython 2.2.

23 years agoFix the erroneous availability annotation for s.makefile() from the last
Fred Drake [Sat, 22 Dec 2001 19:07:58 +0000 (19:07 +0000)]
Fix the erroneous availability annotation for s.makefile() from the last
checkin (my fault!).
Wrap some long lines and fix some markup inconsistencies.

23 years agoRegenerated for Linux 2.2.4.
Martin v. Löwis [Sat, 22 Dec 2001 15:23:50 +0000 (15:23 +0000)]
Regenerated for Linux 2.2.4.

23 years agoAdd TCP socket options from glibc 2.2.4. Fixes #495680.
Martin v. Löwis [Sat, 22 Dec 2001 15:05:32 +0000 (15:05 +0000)]
Add TCP socket options from glibc 2.2.4. Fixes #495680.
2.2.1 bugfix candidate.

23 years agoWindows build: close out 2.2, prep for 2.3.
Tim Peters [Fri, 21 Dec 2001 22:06:12 +0000 (22:06 +0000)]
Windows build:  close out 2.2, prep for 2.3.

23 years agoAdded 2.3a1 section.
Tim Peters [Fri, 21 Dec 2001 21:36:50 +0000 (21:36 +0000)]
Added 2.3a1 section.

23 years agoAnd we start all over again!
Barry Warsaw [Fri, 21 Dec 2001 20:05:33 +0000 (20:05 +0000)]
And we start all over again!

23 years agoMerge of the release22 branch changes back into the trunk.
Barry Warsaw [Fri, 21 Dec 2001 20:04:22 +0000 (20:04 +0000)]
Merge of the release22 branch changes back into the trunk.

23 years agoAdd notes that fromfd() and s.makefile() are Unix-specific.
Fred Drake [Fri, 21 Dec 2001 17:45:03 +0000 (17:45 +0000)]
Add notes that fromfd() and s.makefile() are Unix-specific.
This fixes SF bug #495896.

Fix up various markup consistency & style guide conformance nits.

23 years agoDoc changes on the trunk will not be in Python 2.2, so let's call it 2.2+.
Fred Drake [Fri, 21 Dec 2001 16:46:28 +0000 (16:46 +0000)]
Doc changes on the trunk will not be in Python 2.2, so let's call it 2.2+.

23 years agoPyOS_vsnprintf(): Change PyMem_Malloc() call to PyMem_MALLOC() macro,
Barry Warsaw [Fri, 21 Dec 2001 16:32:15 +0000 (16:32 +0000)]
PyOS_vsnprintf(): Change PyMem_Malloc() call to PyMem_MALLOC() macro,
(ditto for PyMem_Free() -> PyMem_FREE()) to fix and close SF bug
#495875 on systems that HAVE_SNPRINTF=0.

Check in on both release-22 branch and trunk.

23 years agoSuggested by Pete Shinners: treat .m and .mm files as source code.
Andrew M. Kuchling [Fri, 21 Dec 2001 15:34:17 +0000 (15:34 +0000)]
Suggested by Pete Shinners: treat .m and .mm files as source code.

Question for Jack Jansen: is this reasonable?

Candidate for 2.2 release branch (if Jack thinks it's OK).

23 years agoforward-patch from release21-maint branch:
Anthony Baxter [Fri, 21 Dec 2001 05:13:37 +0000 (05:13 +0000)]
forward-patch from release21-maint branch:
  Make dumbdbm merely "dumb", rather than "terminally broken". Without this
  patch, it's almost impossible to use dumbdbm _without_ causing horrible
  datalossage. With this patch, dumbdbm passes my own horrible torture test,
  as well as the roundup test suite.

  dumbdbm really could do with a smidgin of a rewrite or two, but that's not
  suitable for the release21-maint branch.

23 years ago1.00 at last!
Andrew M. Kuchling [Fri, 21 Dec 2001 04:39:11 +0000 (04:39 +0000)]
1.00 at last!
Describe super() very briefly
A few minor reformattings and wording changes
Set the release date (presumably tomorrow...)

23 years agoAdd a reference to the signal module to the os.kill() description.
Fred Drake [Fri, 21 Dec 2001 03:58:47 +0000 (03:58 +0000)]
Add a reference to the signal module to the os.kill() description.
This closes SF bug #495609.

23 years agoFix typo in httplib example.
Fred Drake [Fri, 21 Dec 2001 03:52:04 +0000 (03:52 +0000)]
Fix typo in httplib example.
This fixes SF bug #495221.

23 years agoFix up some examples in the tutorial so we don't contradict our own
Fred Drake [Thu, 20 Dec 2001 23:54:56 +0000 (23:54 +0000)]
Fix up some examples in the tutorial so we don't contradict our own
advice on docstrings.
This fixes SF bug #495601.

23 years agoApplying r22c1 branch mods back to the trunk.
Jack Jansen [Thu, 20 Dec 2001 20:41:45 +0000 (20:41 +0000)]
Applying r22c1 branch mods back to the trunk.

23 years agoFix the availability statement for the spawn*() functions to reflect the
Fred Drake [Thu, 20 Dec 2001 17:24:11 +0000 (17:24 +0000)]
Fix the availability statement for the spawn*() functions to reflect the
actual availability on Windows.
This fixes SF bug #495191.

23 years agoRe-commit Ping's patch to the cgi and cgitb documentation, using the
Fred Drake [Thu, 20 Dec 2001 17:13:09 +0000 (17:13 +0000)]
Re-commit Ping's patch to the cgi and cgitb documentation, using the
right version this time.  Thanks, Ping!
(This was from SF patch #494582, "\index -> \indexii" version.)

23 years agotest_parseaddr_empty(): New test for assuring that
Barry Warsaw [Thu, 20 Dec 2001 16:37:27 +0000 (16:37 +0000)]
test_parseaddr_empty(): New test for assuring that
Utils.parseaddr('<>') -- i.e. on an empty address, returns the empty
string.  Built on rfc822, this used to return None.

23 years agoUpdate the documentation links
Andrew M. Kuchling [Thu, 20 Dec 2001 16:33:45 +0000 (16:33 +0000)]
Update the documentation links
Remove reference to this being a draft

23 years agoUse the final patch/bug numbers
Andrew M. Kuchling [Thu, 20 Dec 2001 16:04:24 +0000 (16:04 +0000)]
Use the final patch/bug numbers

23 years agoZZZ.
Guido van Rossum [Thu, 20 Dec 2001 15:56:23 +0000 (15:56 +0000)]
ZZZ.

23 years agoSF patch #495358 (Artur Zaprzala): rfc822.AddressList and "<>" address
Guido van Rossum [Thu, 20 Dec 2001 15:54:48 +0000 (15:54 +0000)]
SF patch #495358 (Artur Zaprzala): rfc822.AddressList and "<>" address

    rfc822.AddressList incorrectly handles empty address.
    "<>" is converted to None and should be "".
    AddressList.__str__() fails on None.
    I got an email with such an address and my program
    failed processing it.

    Example:
    >>> import rfc822

    >>> rfc822.AddressList("<>").addresslist
    [('', None)]
    >>> str(rfc822.AddressList("<>"))
    Traceback (most recent call last):
    File "<stdin>", line 1, in ?
    File "/usr/lib/python2.1/rfc822.py", line 753, in __str__
    return ", ".join(map(dump_address_pair,
    self.addresslist))
    TypeError: sequence item 0: expected string, None found

[His solution: in the internal routine AddrlistClass.getrouteaddr(),
initialize adlist to "".]

23 years agoAnother contributor.
Guido van Rossum [Thu, 20 Dec 2001 13:19:36 +0000 (13:19 +0000)]
Another contributor.

23 years agoWhitespace normalization.
Tim Peters [Thu, 20 Dec 2001 06:18:15 +0000 (06:18 +0000)]
Whitespace normalization.

23 years agoApparently it's Cygwin with a capital C.
Michael W. Hudson [Wed, 19 Dec 2001 22:09:09 +0000 (22:09 +0000)]
Apparently it's Cygwin with a capital C.

23 years agopartial merge with current pythonware codebase:
Fredrik Lundh [Wed, 19 Dec 2001 21:40:04 +0000 (21:40 +0000)]
partial merge with current pythonware codebase:
- use repr instead of implied str for doubles
- updated version number to 1.0.0 (for 2.2 final)

23 years agoFix a typo
Neal Norwitz [Wed, 19 Dec 2001 20:44:13 +0000 (20:44 +0000)]
Fix a typo

23 years agoMore cygwin news. This section is getting a bit long. Oh well.
Michael W. Hudson [Wed, 19 Dec 2001 19:49:58 +0000 (19:49 +0000)]
More cygwin news.  This section is getting a bit long.  Oh well.

23 years agoSF bug #495021: Crash calling os.stat with a trailing backslash
Tim Peters [Wed, 19 Dec 2001 19:05:01 +0000 (19:05 +0000)]
SF bug #495021:  Crash calling os.stat with a trailing backslash
Patch from Mark Hammond, plus code rearrangement and comments from me.
posix_do_stat():  Windows-specific code could try to free() stack
memory in some cases when a path ending with a forward or backward slash
was passed to os.stat().

23 years agoAdd test for pickling new-style class with custom metaclass.
Guido van Rossum [Wed, 19 Dec 2001 16:58:54 +0000 (16:58 +0000)]
Add test for pickling new-style class with custom metaclass.

23 years agoFix for SF bug #494904: Cannot pickle a class with a metaclass,
Guido van Rossum [Wed, 19 Dec 2001 16:57:36 +0000 (16:57 +0000)]
Fix for SF bug #494904: Cannot pickle a class with a metaclass,
reported by Dan Parisien.

23 years agosave(): Fix for SF bug #494904: Cannot pickle a class with a
Guido van Rossum [Wed, 19 Dec 2001 16:56:54 +0000 (16:56 +0000)]
save(): Fix for SF bug #494904: Cannot pickle a class with a
metaclass, reported by Dan Parisien.

Objects that are instances of custom metaclasses, i.e. whose ob_type
is a subclass of PyType_Type, should be pickled the same as new-style
classes (objects whose ob_type is PyType_Type).  This can't be done
through the existing dispatch switches, and the __reduce__ trick
doesn't work for these, since it finds the unbound __reduce__ for
instances of the class (inherited from PyBaseObject_Type).  So check
explicitly using PyType_IsSubtype().

23 years agoPickler.save(): Fix for SF bug #494904: Cannot pickle a class with a
Guido van Rossum [Wed, 19 Dec 2001 16:55:02 +0000 (16:55 +0000)]
Pickler.save(): Fix for SF bug #494904: Cannot pickle a class with a
metaclass, reported by Dan Parisien.

Objects that are instances of custom metaclasses, i.e. whose class is
a subclass of 'type', should be pickled the same as new-style classes
(objects whose class is 'type').  This can't be done through a
dispatch table entry, and the __reduce__ trick doesn't work for these,
since it finds the unbound __reduce__ for instances of the class
(inherited from 'object').  So check explicitly using issubclass().

23 years agoAdd some additional tests that check more proxy behaviors.
Fred Drake [Wed, 19 Dec 2001 16:54:23 +0000 (16:54 +0000)]
Add some additional tests that check more proxy behaviors.

23 years agoproxy_compare(): Make sure that we unwrap both objects being compared if
Fred Drake [Wed, 19 Dec 2001 16:44:30 +0000 (16:44 +0000)]
proxy_compare():  Make sure that we unwrap both objects being compared if
both are proxy objects.

23 years agoFix the test control support for the pickle & cPickle tests so the tests run
Fred Drake [Wed, 19 Dec 2001 16:42:15 +0000 (16:42 +0000)]
Fix the test control support for the pickle & cPickle tests so the tests run
under regrtest.

23 years agoThe test using class initarg failed, because it was lacking a
Guido van Rossum [Wed, 19 Dec 2001 16:38:29 +0000 (16:38 +0000)]
The test using class initarg failed, because it was lacking a
__safe_for_unpickling__ attribute.

23 years agoMake the module docstring agree with reality: the module prvides the
Fred Drake [Wed, 19 Dec 2001 14:27:41 +0000 (14:27 +0000)]
Make the module docstring agree with reality: the module prvides the
"handler()" function, not the "handle()" function.

23 years agoWhen running regen for the plat directories we should use the BUILDEXT
Jack Jansen [Wed, 19 Dec 2001 09:24:40 +0000 (09:24 +0000)]
When running regen for the plat directories we should use the BUILDEXT
extension, not the EXT one, as regen uses the python binary in the build
directory. Fixes #493959.

23 years agoSF bug #494738: binascii_b2a_base64 overwrites memory.
Tim Peters [Wed, 19 Dec 2001 04:41:35 +0000 (04:41 +0000)]
SF bug #494738:  binascii_b2a_base64 overwrites memory.
binascii_b2a_base64():  We didn't allocate enough buffer space for very
short inputs (e.g., a 1-byte input can produce a 5-byte output, but we
only allocated 2 bytes).  I expect that malloc overheads absorbed the
overrun in practice, but computing a correct upper bound is a very simple
change.