]> granicus.if.org Git - python/log
python
17 years agoBug #1746880: Correctly install DLLs into system32 folder on Win64.
Martin v. Löwis [Thu, 30 Aug 2007 18:27:06 +0000 (18:27 +0000)]
Bug #1746880: Correctly install DLLs into system32 folder on Win64.

17 years agoMore raise statement normalization.
Collin Winter [Thu, 30 Aug 2007 18:18:27 +0000 (18:18 +0000)]
More raise statement normalization.

17 years agoKill more lingering string exceptions in Lib/plat-mac/.
Collin Winter [Thu, 30 Aug 2007 18:11:48 +0000 (18:11 +0000)]
Kill more lingering string exceptions in Lib/plat-mac/.

17 years ago1/0 -> True/False cleanup.
Collin Winter [Thu, 30 Aug 2007 18:11:17 +0000 (18:11 +0000)]
1/0 -> True/False cleanup.

17 years agoCalling execvp with an empty argument list raises ValueError, but we
Thomas Heller [Thu, 30 Aug 2007 17:57:21 +0000 (17:57 +0000)]
Calling execvp with an empty argument list raises ValueError, but we
want to test for OSError.

17 years agoEnable the trunc() test.
Guido van Rossum [Thu, 30 Aug 2007 17:46:57 +0000 (17:46 +0000)]
Enable the trunc() test.

17 years agoAdd numbers.py. I suspect this is an old version, but Jeffrey is out
Guido van Rossum [Thu, 30 Aug 2007 17:45:54 +0000 (17:45 +0000)]
Add numbers.py.  I suspect this is an old version, but Jeffrey is out
of town, and it will have to do for now.

17 years agoFix a lingering string exception.
Collin Winter [Thu, 30 Aug 2007 17:37:22 +0000 (17:37 +0000)]
Fix a lingering string exception.

17 years agoFix typo
Neal Norwitz [Thu, 30 Aug 2007 17:22:44 +0000 (17:22 +0000)]
Fix typo

17 years agoAdd 2.6 and 3.0 to the silly list.
Guido van Rossum [Thu, 30 Aug 2007 17:21:12 +0000 (17:21 +0000)]
Add 2.6 and 3.0 to the silly list.

17 years agoUpdate patch level to 3.0a1. We're close enough.
Guido van Rossum [Thu, 30 Aug 2007 17:20:01 +0000 (17:20 +0000)]
Update patch level to 3.0a1.  We're close enough.

17 years agoUpdate Misc/NEWS. Add RELNOTES.
Guido van Rossum [Thu, 30 Aug 2007 17:16:55 +0000 (17:16 +0000)]
Update Misc/NEWS.  Add RELNOTES.

17 years agoForbid an empty argument list in execv call.
Thomas Heller [Thu, 30 Aug 2007 17:15:14 +0000 (17:15 +0000)]
Forbid an empty argument list in execv call.

Fixes issue 1039.

17 years agoMove definition of TZNAME_ENCODING further up.
Martin v. Löwis [Thu, 30 Aug 2007 15:40:24 +0000 (15:40 +0000)]
Move definition of TZNAME_ENCODING further up.

17 years agoSet startinpos before calling the error handler.
Walter Dörwald [Thu, 30 Aug 2007 15:35:26 +0000 (15:35 +0000)]
Set startinpos before calling the error handler.

17 years agoRewrap line.
Walter Dörwald [Thu, 30 Aug 2007 15:29:21 +0000 (15:29 +0000)]
Rewrap line.

17 years agoPort to Python 3.0.
Martin v. Löwis [Thu, 30 Aug 2007 15:23:04 +0000 (15:23 +0000)]
Port to Python 3.0.

17 years agoImplement PYTHONNOERRORWINDOW.
Martin v. Löwis [Thu, 30 Aug 2007 14:57:25 +0000 (14:57 +0000)]
Implement PYTHONNOERRORWINDOW.

17 years agoMove the copyright and license stuff to the end.
Guido van Rossum [Thu, 30 Aug 2007 14:51:05 +0000 (14:51 +0000)]
Move the copyright and license stuff to the end.

17 years agoExplicitly convert tzname to Unicode. Fixes #1040.
Martin v. Löwis [Thu, 30 Aug 2007 14:37:48 +0000 (14:37 +0000)]
Explicitly convert tzname to Unicode. Fixes #1040.

17 years agoMore email package related repairs. This fixes smtplib's import and use of
Barry Warsaw [Thu, 30 Aug 2007 14:28:55 +0000 (14:28 +0000)]
More email package related repairs.  This fixes smtplib's import and use of
email.base64mime, but test_smtplib still has failures for me.  They are
timeout errors so think they're more related to my current wacky network setup
than bugs remaining in the code related to the email package.

This also r57693 that got clobbered with the sandbox sync, and fixes a couple
of other minor problems that cropped up.  I will kill the sandbox branch next.

The email package now has 11F/11E.

17 years agoTwo changes (not enough to make the test pass though) having to do with
Guido van Rossum [Thu, 30 Aug 2007 14:02:43 +0000 (14:02 +0000)]
Two changes (not enough to make the test pass though) having to do with
the bytes() constructor when passing str.

17 years agoUse s# instead of t# for nis_match.
Martin v. Löwis [Thu, 30 Aug 2007 10:21:51 +0000 (10:21 +0000)]
Use s# instead of t# for nis_match.

17 years agoStop testing for encoded file names, as Python 3 does
Martin v. Löwis [Thu, 30 Aug 2007 10:08:57 +0000 (10:08 +0000)]
Stop testing for encoded file names, as Python 3 does
not support them, anyway.

17 years agoTry to fix test_plistlib so it uses bytes consistently in this call
Neal Norwitz [Thu, 30 Aug 2007 06:37:08 +0000 (06:37 +0000)]
Try to fix test_plistlib so it uses bytes consistently in this call

17 years agoUse inspect.py and clean up.
Kurt B. Kaiser [Thu, 30 Aug 2007 06:35:15 +0000 (06:35 +0000)]
Use inspect.py and clean up.

17 years agoTry to fix test_aepack by comparing bytes with bytes.
Neal Norwitz [Thu, 30 Aug 2007 06:32:09 +0000 (06:32 +0000)]
Try to fix test_aepack by comparing bytes with bytes.

17 years agoWrite bytes instead of unicode
Neal Norwitz [Thu, 30 Aug 2007 06:20:05 +0000 (06:20 +0000)]
Write bytes instead of unicode

17 years agoTry to get test_pep352 and maybe test_platform to work on a Windows buildbot
Neal Norwitz [Thu, 30 Aug 2007 06:16:26 +0000 (06:16 +0000)]
Try to get test_pep352 and maybe test_platform to work on a Windows buildbot

17 years agoSince the filename is encoded, we must add bytes to it. The test
Neal Norwitz [Thu, 30 Aug 2007 05:59:15 +0000 (05:59 +0000)]
Since the filename is encoded, we must add bytes to it.  The test
still fails, but gets a little further with this change.

17 years agoAdd uuids for 3.0.
Martin v. Löwis [Thu, 30 Aug 2007 05:45:17 +0000 (05:45 +0000)]
Add uuids for 3.0.

17 years agoTry to fix the problem on the Windows buildbot where this code:
Neal Norwitz [Thu, 30 Aug 2007 05:44:54 +0000 (05:44 +0000)]
Try to fix the problem on the Windows buildbot where this code:

  b * sys.maxint  # from test_bytes.py line 379

Caused:  SystemError: error return without exception set

17 years agoBump version numbers.
Martin v. Löwis [Thu, 30 Aug 2007 05:40:17 +0000 (05:40 +0000)]
Bump version numbers.

17 years agoStop using the find function on the string module, use the string method.
Neal Norwitz [Thu, 30 Aug 2007 05:35:41 +0000 (05:35 +0000)]
Stop using the find function on the string module, use the string method.
This will hopefully fix the problem on a Windows buildbot with test_sundry.

17 years agoForward Windows build instructions to PCbuild.
Martin v. Löwis [Thu, 30 Aug 2007 05:34:10 +0000 (05:34 +0000)]
Forward Windows build instructions to PCbuild.

17 years agoMake this module importable, this gets test___all__ to pass.
Neal Norwitz [Thu, 30 Aug 2007 05:19:29 +0000 (05:19 +0000)]
Make this module importable, this gets test___all__ to pass.

17 years agoGeneral cleanup, raise normalization in Lib/distutils.
Collin Winter [Thu, 30 Aug 2007 03:52:21 +0000 (03:52 +0000)]
General cleanup, raise normalization in Lib/distutils.

17 years agoRaise statement normalization in Lib/ctypes/.
Collin Winter [Thu, 30 Aug 2007 03:47:13 +0000 (03:47 +0000)]
Raise statement normalization in Lib/ctypes/.

17 years agoOops. I copied a slightly older version of the email package from the sandbox.
Guido van Rossum [Thu, 30 Aug 2007 03:46:43 +0000 (03:46 +0000)]
Oops.  I copied a slightly older version of the email package from the sandbox.
This should restore the email package in the py3k branch to exactly what's in
the sandbox.

This wipes out 1-2 fixes made post-copy, which I'll re-apply shortly.

17 years agoUse the new name for the parser module.
Neal Norwitz [Thu, 30 Aug 2007 03:10:46 +0000 (03:10 +0000)]
Use the new name for the parser module.

17 years agoRevert revisions 57472, 57474, and 57477 which disabled some tests
Neal Norwitz [Thu, 30 Aug 2007 03:06:59 +0000 (03:06 +0000)]
Revert revisions 57472, 57474, and 57477 which disabled some tests
when the email package was removed.  Email was restored, so restore the tests.

17 years agoFix a more bytes/str confusion.
Barry Warsaw [Thu, 30 Aug 2007 02:10:49 +0000 (02:10 +0000)]
Fix a more bytes/str confusion.

Use str.encode('raw-unicode-escape') consistently instead of bytes(string).

Remove the convert_eols argument from base64mime.decode().  This matches
previous API changes done to the quoprimime module.

17 years agoRaise statement normalization in Lib/.
Collin Winter [Thu, 30 Aug 2007 01:19:48 +0000 (01:19 +0000)]
Raise statement normalization in Lib/.

17 years agoCopying the email package back, despite its failings.
Guido van Rossum [Thu, 30 Aug 2007 01:15:14 +0000 (01:15 +0000)]
Copying the email package back, despite its failings.

17 years agoFix an outdated URL in a SyntaxError message.
Guido van Rossum [Thu, 30 Aug 2007 00:10:46 +0000 (00:10 +0000)]
Fix an outdated URL in a SyntaxError message.

17 years agoWhen clobbering, also clean the docs.
Guido van Rossum [Wed, 29 Aug 2007 23:48:29 +0000 (23:48 +0000)]
When clobbering, also clean the docs.

17 years agoRevert r57685 (weird merge result).
Collin Winter [Wed, 29 Aug 2007 23:41:34 +0000 (23:41 +0000)]
Revert r57685 (weird merge result).

17 years agoMake it work with Py3k.
Guido van Rossum [Wed, 29 Aug 2007 23:37:34 +0000 (23:37 +0000)]
Make it work with Py3k.

17 years agoRaise statement normalization in Lib/test/.
Collin Winter [Wed, 29 Aug 2007 23:37:32 +0000 (23:37 +0000)]
Raise statement normalization in Lib/test/.

17 years ago(no commit message)
Collin Winter [Wed, 29 Aug 2007 23:37:10 +0000 (23:37 +0000)]

17 years ago(no commit message)
Collin Winter [Wed, 29 Aug 2007 23:36:47 +0000 (23:36 +0000)]

17 years agoDon't delete test output when "make clean" is run -- only when "make
Guido van Rossum [Wed, 29 Aug 2007 23:35:30 +0000 (23:35 +0000)]
Don't delete test output when "make clean" is run -- only when "make
clobber" is run.  (Thanks Thomas W. for pointing this out!)

17 years agoModernize clean and funny targets.
Guido van Rossum [Wed, 29 Aug 2007 23:24:02 +0000 (23:24 +0000)]
Modernize clean and funny targets.

17 years agoMostly rewritten, much shorter README for Py3k.
Guido van Rossum [Wed, 29 Aug 2007 23:03:30 +0000 (23:03 +0000)]
Mostly rewritten, much shorter README for Py3k.

17 years ago[Oops, I forgot half of the patch.]
Guido van Rossum [Wed, 29 Aug 2007 20:39:13 +0000 (20:39 +0000)]
[Oops, I forgot half of the patch.]
Patch # 1050 by Amaury Forgeot d'Arc.
On Windows, debug builds insert stack probes, and recursive functions
tend to exhaust the stack faster.
This patch reduces the marshal maximum depth from 2000 to 1500 for debug
builds only. Optimized builds are not affected.
This allows test_marshal to pass with debug builds.

17 years agoAndother bytes/str comparison caught by Jeremy's change.
Guido van Rossum [Wed, 29 Aug 2007 20:24:57 +0000 (20:24 +0000)]
Andother bytes/str comparison caught by Jeremy's change.

17 years agoChange string literal to bytes.
Jeremy Hylton [Wed, 29 Aug 2007 19:09:54 +0000 (19:09 +0000)]
Change string literal to bytes.

17 years agoMake terminator constants bytes.
Jeremy Hylton [Wed, 29 Aug 2007 19:08:30 +0000 (19:08 +0000)]
Make terminator constants bytes.

17 years agoChange string literals to bytes, since they are compared to bytes.
Jeremy Hylton [Wed, 29 Aug 2007 19:07:40 +0000 (19:07 +0000)]
Change string literals to bytes, since they are compared to bytes.

17 years agoFix issue # 1037 (sort of).
Guido van Rossum [Wed, 29 Aug 2007 18:54:41 +0000 (18:54 +0000)]
Fix issue # 1037 (sort of).

17 years agoMake it an error to compare a bytes object and a Unicode object.
Jeremy Hylton [Wed, 29 Aug 2007 18:47:16 +0000 (18:47 +0000)]
Make it an error to compare a bytes object and a Unicode object.

17 years agoPatch # 1050 by Amaury Forgeot d'Arc.
Guido van Rossum [Wed, 29 Aug 2007 18:44:54 +0000 (18:44 +0000)]
Patch # 1050 by Amaury Forgeot d'Arc.
On Windows, debug builds insert stack probes, and recursive functions
tend to exhaust the stack faster.
This patch reduces the marshal maximum depth from 2000 to 1500 for debug
builds only. Optimized builds are not affected.
This allows test_marshal to pass with debug builds.

17 years ago1. Debugger was failing to start due to DictProxy limitations.
Kurt B. Kaiser [Wed, 29 Aug 2007 18:44:24 +0000 (18:44 +0000)]
1. Debugger was failing to start due to DictProxy limitations.
2. Fix some debug prints in RemoteDebugger.py - use py3k syntax.

17 years agoPatch # 1048 by Amaury Forgeot d'Arc.
Guido van Rossum [Wed, 29 Aug 2007 18:42:15 +0000 (18:42 +0000)]
Patch # 1048 by Amaury Forgeot d'Arc.
test_float crashes on Windows, because the %zd format is used in a call
to PyOS_snprintf().
The attached patch properly uses PY_FORMAT_SIZE_T.

17 years agoThree patches from issue #1047, by Amaury Forgeot d'Arc:
Guido van Rossum [Wed, 29 Aug 2007 18:38:11 +0000 (18:38 +0000)]
Three patches from issue #1047, by Amaury Forgeot d'Arc:

1/ getargs.diff adds the 'Z' and 'Z#' format specifiers for
PyArg_ParseTuple. They mimic z and z# for unicode strings, by accepting
a Unicode or None (in which case the Py_UNICODE* pointer is set to
NULL). With doc and tests.

2/ subprocess.diff converts file PC/_subprocess.c to unicode. We use the
Unicode version of the win32 api (and Z conversion from previous patch)

3/ stdout.diff: sys.stdout must not convert the line endings, Windows
already does it.
Without this patch, when redirecting the output of python, the file
contains \r\r\n for each line. (test_subprocess did catch this)

However, I (GvR) removed the change to _fileio.c (included in the
patches) that prevents closing file descripors < 3 from being closed;
I think that needs to be solved in a different way.

17 years agoIn rseponse to bug# 1029, force the newline default for StringIO to "\n",
Guido van Rossum [Wed, 29 Aug 2007 18:31:16 +0000 (18:31 +0000)]
In rseponse to bug# 1029, force the newline default for StringIO to "\n",
so that even on Windows, after s.write("x\n"), s.getvalue() == "x\n".

17 years agoUnset PYTHONPATH, for my convenience.
Guido van Rossum [Wed, 29 Aug 2007 18:15:48 +0000 (18:15 +0000)]
Unset PYTHONPATH, for my convenience.

17 years agoInsist that the argument to TextIOWrapper.write() is a basestring
Guido van Rossum [Wed, 29 Aug 2007 18:10:08 +0000 (18:10 +0000)]
Insist that the argument to TextIOWrapper.write() is a basestring
instance.  This was effectively already the case, but the error
reporting was lousy.

17 years agoUse the correct type for variables corresponding to 'u' formats.
Guido van Rossum [Wed, 29 Aug 2007 17:28:42 +0000 (17:28 +0000)]
Use the correct type for variables corresponding to 'u' formats.

17 years agoConvert various string literals to bytes.
Jeremy Hylton [Wed, 29 Aug 2007 17:26:34 +0000 (17:26 +0000)]
Convert various string literals to bytes.

17 years agoPEP 7 clean ups, no change in behavior.
Jeremy Hylton [Wed, 29 Aug 2007 16:21:59 +0000 (16:21 +0000)]
PEP 7 clean ups, no change in behavior.

Reflow long lines.
Remove whitespace inside parents.
Move open braces off lines by themselves.

17 years agoFix test failures caused by missing/incorrect conversion to bytes.
Jeremy Hylton [Wed, 29 Aug 2007 14:56:40 +0000 (14:56 +0000)]
Fix test failures caused by missing/incorrect conversion to bytes.

17 years agoFix up brokenness with hashing, now hashlib is strict in requiring bytes too.
Guido van Rossum [Wed, 29 Aug 2007 14:26:52 +0000 (14:26 +0000)]
Fix up brokenness with hashing, now hashlib is strict in requiring bytes too.

17 years agoStraggler (forgot to save this earlier).
Guido van Rossum [Wed, 29 Aug 2007 14:26:02 +0000 (14:26 +0000)]
Straggler (forgot to save this earlier).

17 years agoRefuse to compute digests from PyUnicode (str) instances.
Guido van Rossum [Wed, 29 Aug 2007 14:21:45 +0000 (14:21 +0000)]
Refuse to compute digests from PyUnicode (str) instances.
This breaks a few things that I'll patch up in a minute.

17 years agoOops, remove an abort() I put in for debugging.
Guido van Rossum [Wed, 29 Aug 2007 14:05:23 +0000 (14:05 +0000)]
Oops, remove an abort() I put in for debugging.

17 years agoFix the one failing test (can't decode twice).
Guido van Rossum [Wed, 29 Aug 2007 14:04:40 +0000 (14:04 +0000)]
Fix the one failing test (can't decode twice).

17 years agoAdd PyUnicode_AsStringAndSize(), which is like PyUnicode_AsString() but
Guido van Rossum [Wed, 29 Aug 2007 13:53:23 +0000 (13:53 +0000)]
Add PyUnicode_AsStringAndSize(), which is like PyUnicode_AsString() but
has an extra (optional) output parameter through which it returns the size.
Use this in a few places where I used PyUnicode_AsString() + strlen(),
and in one new place (which fixes test_pep263).

17 years agoFix test_pyclbr -- _https_connection is optional.
Guido van Rossum [Wed, 29 Aug 2007 13:35:11 +0000 (13:35 +0000)]
Fix test_pyclbr -- _https_connection is optional.

17 years agoFix failure in error handler -- exc[-1] -> exc.args[-1].
Guido van Rossum [Wed, 29 Aug 2007 13:18:47 +0000 (13:18 +0000)]
Fix failure in error handler -- exc[-1] -> exc.args[-1].

17 years agoCorrected missed #if in r57652.
Eric Smith [Wed, 29 Aug 2007 12:43:12 +0000 (12:43 +0000)]
Corrected missed #if in r57652.

17 years agoAdded conditional compilation for '()', which was an allowed sign code in a
Eric Smith [Wed, 29 Aug 2007 12:38:45 +0000 (12:38 +0000)]
Added conditional compilation for '()', which was an allowed sign code in a
previous version of PEP 3101.  It's currently not compiled in, but I want to
leave it because it might be useful in the future and it makes
calc_number_widths() clearer.  It justifies NumberFieldWidths.rsign
and .n__rsign.

17 years agoGet test_bsddb3 passing
Neal Norwitz [Wed, 29 Aug 2007 05:52:49 +0000 (05:52 +0000)]
Get test_bsddb3 passing

17 years agostrings vs bytes, bytes wins again
Gregory P. Smith [Wed, 29 Aug 2007 05:06:13 +0000 (05:06 +0000)]
strings vs bytes, bytes wins again

17 years agoCommit strict str/bytes distinction.
Guido van Rossum [Wed, 29 Aug 2007 04:05:57 +0000 (04:05 +0000)]
Commit strict str/bytes distinction.
From now on, trying to write str to a binary stream
is an error (I'm still working on the reverse).
There are still (at least) two failing tests:
- test_asynchat
- test_urllib2_localnet
but I'm sure these will be fixed by someone.

17 years agoFound a different, more direct way to disable ssl support until it's fixed.
Guido van Rossum [Wed, 29 Aug 2007 03:59:57 +0000 (03:59 +0000)]
Found a different, more direct way to disable ssl support until it's fixed.

17 years agoChange the way the encoding parameter is handled.
Guido van Rossum [Wed, 29 Aug 2007 03:53:53 +0000 (03:53 +0000)]
Change the way the encoding parameter is handled.
This fixes test_doctest with strict bytes/str.

17 years agoMake test_cmd_line work with strict str/bytes.
Guido van Rossum [Wed, 29 Aug 2007 03:47:36 +0000 (03:47 +0000)]
Make test_cmd_line work with strict str/bytes.

17 years agoFix the sqlite test. Blobs should be created using buffer(b"blob"),
Guido van Rossum [Wed, 29 Aug 2007 03:44:33 +0000 (03:44 +0000)]
Fix the sqlite test.  Blobs should be created using buffer(b"blob"),
not buffer("blob").

17 years ago"Fix" a few places that were using PyObject_AsCharBuffer() to convert a string
Guido van Rossum [Wed, 29 Aug 2007 03:34:29 +0000 (03:34 +0000)]
"Fix" a few places that were using PyObject_AsCharBuffer() to convert a string
(PyUnicode these days) to a char* + length.  The fix consists of calling
PyUnicode_AsString() and strlen().  This is not ideal, but AsCharBuffer()
is definitely not the API to use.

17 years agoModified parsing of format strings, so that we always return
Eric Smith [Wed, 29 Aug 2007 03:22:59 +0000 (03:22 +0000)]
Modified parsing of format strings, so that we always return
a tuple (literal, field_name, format_spec, conversion).

literal will always be a string, but might be of zero length.
field_name will be None if there is no markup text
format_spec will be a (possibly zero length) string if
  field_name is non-None
conversion will be a one character string, or None

This makes the Formatter class, and especially it's parse()
method, easier to understand.

Suggestion was by Jim Jewett, inspired by the "tail" of an
elementtree node.

Also, fixed a reference leak in fieldnameiter_next.

17 years agoMake gettext work with strict str/bytes.
Guido van Rossum [Wed, 29 Aug 2007 03:08:55 +0000 (03:08 +0000)]
Make gettext work with strict str/bytes.

17 years agoFix this test. How could it ever have worked?!
Guido van Rossum [Wed, 29 Aug 2007 02:57:31 +0000 (02:57 +0000)]
Fix this test.  How could it ever have worked?!

17 years agoForgot one. This makes test_urllib2.py pass.
Guido van Rossum [Wed, 29 Aug 2007 02:30:49 +0000 (02:30 +0000)]
Forgot one.  This makes test_urllib2.py pass.

17 years agoMake test_httplib pass.
Guido van Rossum [Wed, 29 Aug 2007 02:00:20 +0000 (02:00 +0000)]
Make test_httplib pass.

17 years agoMake test_urllib be strict about str/bytes.
Guido van Rossum [Wed, 29 Aug 2007 01:53:26 +0000 (01:53 +0000)]
Make test_urllib be strict about str/bytes.
(One change to httplib.py, but not enough for test_httplib.)

17 years agoDisable this test too.
Guido van Rossum [Wed, 29 Aug 2007 01:46:57 +0000 (01:46 +0000)]
Disable this test too.

17 years agoDisable test_ssl until ssl.py has been fixed.
Guido van Rossum [Wed, 29 Aug 2007 01:26:39 +0000 (01:26 +0000)]
Disable test_ssl until ssl.py has been fixed.

17 years agoFix buglet in sliceobjects, they were not returning Py_NotImplemented when
Thomas Wouters [Tue, 28 Aug 2007 23:07:26 +0000 (23:07 +0000)]
Fix buglet in sliceobjects, they were not returning Py_NotImplemented when
compared against something other than sliceobjects.

17 years agoMerge the trunk changes in. Breaks socket.ssl for now.
Thomas Wouters [Tue, 28 Aug 2007 21:37:11 +0000 (21:37 +0000)]
Merge the trunk changes in. Breaks socket.ssl for now.

Merged revisions 57392-57619 via svnmerge from
svn+ssh://pythondev@svn.python.org/python/trunk

........
  r57395 | georg.brandl | 2007-08-24 19:23:23 +0200 (Fri, 24 Aug 2007) | 2 lines

  Bug #1011: fix rfc822.Message.getheader docs.
........
  r57397 | georg.brandl | 2007-08-24 19:38:49 +0200 (Fri, 24 Aug 2007) | 2 lines

  Patch #1006: port test_winreg to unittest.
........
  r57398 | georg.brandl | 2007-08-24 19:46:54 +0200 (Fri, 24 Aug 2007) | 2 lines

  Fix #1012: wrong URL to :mod:`site` in install/index.rst.
........
  r57399 | georg.brandl | 2007-08-24 20:07:52 +0200 (Fri, 24 Aug 2007) | 2 lines

  Patch #1008: port test_signal to unittest.
........
  r57400 | georg.brandl | 2007-08-24 20:22:54 +0200 (Fri, 24 Aug 2007) | 2 lines

  Port test_frozen to unittest.
........
  r57401 | georg.brandl | 2007-08-24 20:27:43 +0200 (Fri, 24 Aug 2007) | 2 lines

  Document new utility functions in test_support.
........
  r57402 | georg.brandl | 2007-08-24 20:30:06 +0200 (Fri, 24 Aug 2007) | 2 lines

  Remove test_rgbimg output file, there is no test_rgbimg.py.
........
  r57403 | georg.brandl | 2007-08-24 20:35:27 +0200 (Fri, 24 Aug 2007) | 2 lines

  Remove output file for test_ossaudiodev, also properly close the dsp object.
........
  r57404 | georg.brandl | 2007-08-24 20:46:27 +0200 (Fri, 24 Aug 2007) | 2 lines

  Convert test_linuxaudiodev to unittest. Fix a wrong finally clause in test_ossaudiodev.
........
  r57406 | collin.winter | 2007-08-24 21:13:58 +0200 (Fri, 24 Aug 2007) | 1 line

  Convert test_pkg to use unittest.
........
  r57408 | georg.brandl | 2007-08-24 21:22:34 +0200 (Fri, 24 Aug 2007) | 2 lines

  Catch the correct errors.
........
  r57409 | georg.brandl | 2007-08-24 21:33:53 +0200 (Fri, 24 Aug 2007) | 2 lines

  Port test_class to unittest. Patch #1671298.
........
  r57415 | collin.winter | 2007-08-24 23:09:42 +0200 (Fri, 24 Aug 2007) | 1 line

  Make test_structmembers pass when run with regrtests's -R flag.
........
  r57455 | nick.coghlan | 2007-08-25 06:32:07 +0200 (Sat, 25 Aug 2007) | 1 line

  Revert misguided attempt at fixing incompatibility between -m and -i switches (better fix coming soon)
........
  r57456 | nick.coghlan | 2007-08-25 06:35:54 +0200 (Sat, 25 Aug 2007) | 1 line

  Revert compile.c changes that shouldn't have been included in previous checkin
........
  r57461 | nick.coghlan | 2007-08-25 12:50:41 +0200 (Sat, 25 Aug 2007) | 1 line

  Fix bug 1764407 - the -i switch now does the right thing when using the -m switch
........
  r57464 | guido.van.rossum | 2007-08-25 17:08:43 +0200 (Sat, 25 Aug 2007) | 4 lines

  Server-side SSL and certificate validation, by Bill Janssen.
  While cleaning up Bill's C style, I may have cleaned up some code
  he didn't touch as well (in _ssl.c).
........
  r57465 | neal.norwitz | 2007-08-25 18:41:36 +0200 (Sat, 25 Aug 2007) | 3 lines

  Try to get this to build with Visual Studio by moving all the variable
  declarations to the beginning of a scope.
........
  r57466 | neal.norwitz | 2007-08-25 18:54:38 +0200 (Sat, 25 Aug 2007) | 1 line

  Fix test so it is skipped properly if there is no SSL support.
........
  r57467 | neal.norwitz | 2007-08-25 18:58:09 +0200 (Sat, 25 Aug 2007) | 2 lines

  Fix a few more variables to try to get this to compile with Visual Studio.
........
  r57473 | neal.norwitz | 2007-08-25 19:25:17 +0200 (Sat, 25 Aug 2007) | 1 line

  Try to get this test to pass for systems that do not have SO_REUSEPORT
........
  r57482 | gregory.p.smith | 2007-08-26 02:26:00 +0200 (Sun, 26 Aug 2007) | 7 lines

  keep setup.py from listing unneeded hash modules (_md5, _sha*) as
  missing when they were not built because _hashlib with openssl provided
  their functionality instead.

  don't build bsddb185 if bsddb was built.
........
  r57483 | neal.norwitz | 2007-08-26 03:08:16 +0200 (Sun, 26 Aug 2007) | 1 line

  Fix typo in docstring (missing c in reacquire)
........
  r57484 | neal.norwitz | 2007-08-26 03:42:03 +0200 (Sun, 26 Aug 2007) | 2 lines

  Spell check (also americanify behaviour, it's almost 3 times as common)
........
  r57503 | neal.norwitz | 2007-08-26 08:29:57 +0200 (Sun, 26 Aug 2007) | 4 lines

  Reap children before the test starts so hopefully SocketServer
  won't find any old children left around which causes an exception
  in collect_children() and the test to fail.
........
  r57510 | neal.norwitz | 2007-08-26 20:50:39 +0200 (Sun, 26 Aug 2007) | 1 line

  Fail gracefully if the cert files cannot be created
........
  r57513 | guido.van.rossum | 2007-08-26 21:35:09 +0200 (Sun, 26 Aug 2007) | 4 lines

  Bill Janssen wrote:
  Here's a patch which makes test_ssl a better player in the buildbots
  environment.  I deep-ended on "try-except-else" clauses.
........
  r57518 | neal.norwitz | 2007-08-26 23:40:16 +0200 (Sun, 26 Aug 2007) | 1 line

  Get the test passing by commenting out some writes (should they be removed?)
........
  r57522 | neal.norwitz | 2007-08-27 00:16:23 +0200 (Mon, 27 Aug 2007) | 3 lines

  Catch IOError for when the device file doesn't exist or the user doesn't have
  permission to write to the device.
........
  r57524 | neal.norwitz | 2007-08-27 00:20:03 +0200 (Mon, 27 Aug 2007) | 5 lines

  Another patch from Bill Janssen that:
  1)  Fixes the bug that two class names are initial-lower-case.
  2)  Replaces the poll waiting for the server to become ready with
      a threading.Event signal.
........
  r57536 | neal.norwitz | 2007-08-27 02:58:33 +0200 (Mon, 27 Aug 2007) | 1 line

  Stop using string.join (from the module) to ease upgrade to py3k
........
  r57537 | neal.norwitz | 2007-08-27 03:03:18 +0200 (Mon, 27 Aug 2007) | 1 line

  Make a utility function for handling (printing) an error
........
  r57538 | neal.norwitz | 2007-08-27 03:15:33 +0200 (Mon, 27 Aug 2007) | 4 lines

  If we can't create a certificate, print a warning, but don't fail the test.
  Modified patch from what Bill Janssen sent on python-3000.
........
  r57539 | facundo.batista | 2007-08-27 03:15:34 +0200 (Mon, 27 Aug 2007) | 7 lines

  Ignore test failures caused by 'resource temporarily unavailable'
  exceptions raised in the test server thread, since SimpleXMLRPCServer
  does not gracefully handle them.  Changed number of requests handled
  by tests server thread to one (was 2) because no tests require more
  than one request. [GSoC - Alan McIntyre]
........
  r57561 | guido.van.rossum | 2007-08-27 19:19:42 +0200 (Mon, 27 Aug 2007) | 8 lines

  > Regardless, building a fixed test certificate and checking it in sounds like
  > the better option.  Then the openssl command in the test code can be turned
  > into a comment describing how the test data was pregenerated.

  Here's a patch that does that.

  Bill
........
  r57568 | guido.van.rossum | 2007-08-27 20:42:23 +0200 (Mon, 27 Aug 2007) | 26 lines

  > Some of the code sets the error string in this directly before
  > returning NULL, and other pieces of the code call PySSL_SetError,
  > which creates the error string.  I think some of the places which set
  > the string directly probably shouldn't; instead, they should call
  > PySSL_SetError to cons up the error name directly from the err code.
  > However, PySSL_SetError only works after the construction of an ssl
  > object, which means it can't be used there...  I'll take a longer look
  > at it and see if there's a reasonable fix.

  Here's a patch which addresses this.  It also fixes the indentation in
  PySSL_SetError, bringing it into line with PEP 7, fixes a compile warning
  about one of the OpenSSL macros, and makes the namespace a bit more
  consistent.  I've tested it on FC 7 and OS X 10.4.

  % ./python ./Lib/test/regrtest.py -R :1: -u all test_ssl
  test_ssl
  beginning 6 repetitions
  123456
  ......
  1 test OK.
  [29244 refs]
  %

  [GvR: slightly edited to enforce 79-char line length, even if it required
   violating the style guide.]
........
  r57570 | guido.van.rossum | 2007-08-27 21:11:11 +0200 (Mon, 27 Aug 2007) | 2 lines

  Patch 10124 by Bill Janssen, docs for the new ssl code.
........
  r57574 | guido.van.rossum | 2007-08-27 22:51:00 +0200 (Mon, 27 Aug 2007) | 3 lines

  Patch # 1739906 by Christian Heimes -- add reduce to functools (importing
  it from __builtin__).
........
  r57575 | guido.van.rossum | 2007-08-27 22:52:10 +0200 (Mon, 27 Aug 2007) | 2 lines

  News about functools.reduce.
........
  r57611 | georg.brandl | 2007-08-28 10:29:08 +0200 (Tue, 28 Aug 2007) | 2 lines

  Document rev. 57574.
........
  r57612 | sean.reifschneider | 2007-08-28 11:07:54 +0200 (Tue, 28 Aug 2007) | 2 lines

  Adding basic imputil documentation.
........
  r57614 | georg.brandl | 2007-08-28 12:48:18 +0200 (Tue, 28 Aug 2007) | 2 lines

  Fix some glitches.
........
  r57616 | lars.gustaebel | 2007-08-28 14:31:09 +0200 (Tue, 28 Aug 2007) | 5 lines

  TarFile.__init__() no longer fails if no name argument is passed and
  the fileobj argument has no usable name attribute (e.g. StringIO).

  (will backport to 2.5)
........
  r57619 | thomas.wouters | 2007-08-28 17:28:19 +0200 (Tue, 28 Aug 2007) | 22 lines

  Improve extended slicing support in builtin types and classes. Specifically:

   - Specialcase extended slices that amount to a shallow copy the same way as
     is done for simple slices, in the tuple, string and unicode case.

   - Specialcase step-1 extended slices to optimize the common case for all
     involved types.

   - For lists, allow extended slice assignment of differing lengths as long
     as the step is 1. (Previously, 'l[:2:1] = []' failed even though
     'l[:2] = []' and 'l[:2:None] = []' do not.)

   - Implement extended slicing for buffer, array, structseq, mmap and
     UserString.UserString.

   - Implement slice-object support (but not non-step-1 slice assignment) for
     UserString.MutableString.

   - Add tests for all new functionality.
........