svn+ssh://pythondev@svn.python.org/python/branches/p3yk
................
r55413 | fred.drake | 2007-05-17 12:30:10 -0700 (Thu, 17 May 2007) | 1 line
fix argument name in documentation; match the implementation
................
r55430 | jack.diederich | 2007-05-18 06:39:59 -0700 (Fri, 18 May 2007) | 1 line
Implements class decorators, PEP 3129.
................
r55432 | guido.van.rossum | 2007-05-18 08:09:41 -0700 (Fri, 18 May 2007) | 2 lines
obsubmit.
................
r55434 | guido.van.rossum | 2007-05-18 09:39:10 -0700 (Fri, 18 May 2007) | 3 lines
Fix bug in test_inspect. (I presume this is how it should be fixed;
Jack Diedrich, please verify.)
................
r55460 | brett.cannon | 2007-05-20 00:31:57 -0700 (Sun, 20 May 2007) | 4 lines
Remove the imageop module. With imgfile already removed in Python 3.0 and
rgbimg gone in Python 2.6 the unit tests themselves were made worthless. Plus
third-party libraries perform the same function much better.
................
r55469 | neal.norwitz | 2007-05-20 11:28:20 -0700 (Sun, 20 May 2007) | 118 lines
Merged revisions 55324-55467 via svnmerge from
svn+ssh://pythondev@svn.python.org/python/trunk
........
r55348 | georg.brandl | 2007-05-15 13:19:34 -0700 (Tue, 15 May 2007) | 4 lines
HTML-escape the plain traceback in cgitb's HTML output, to prevent
the traceback inadvertently or maliciously closing the comment and
injecting HTML into the error page.
........
r55372 | neal.norwitz | 2007-05-15 21:33:50 -0700 (Tue, 15 May 2007) | 6 lines
Port rev 55353 from Guido:
Add what looks like a necessary call to PyErr_NoMemory() when PyMem_MALLOC()
fails.
Will backport.
........
r55377 | neal.norwitz | 2007-05-15 22:06:33 -0700 (Tue, 15 May 2007) | 1 line
Mention removal of some directories for obsolete platforms
........
r55380 | brett.cannon | 2007-05-15 22:50:03 -0700 (Tue, 15 May 2007) | 2 lines
Change the maintainer of the BeOS port.
........
r55383 | georg.brandl | 2007-05-16 06:44:18 -0700 (Wed, 16 May 2007) | 2 lines
Bug #1719995: don't use deprecated method in sets example.
........
r55386 | neal.norwitz | 2007-05-16 13:05:11 -0700 (Wed, 16 May 2007) | 5 lines
Fix bug in marshal where bad data would cause a segfault due to
lack of an infinite recursion check.
Contributed by Damien Miller at Google.
........
r55389 | brett.cannon | 2007-05-16 15:42:29 -0700 (Wed, 16 May 2007) | 6 lines
Remove the gopherlib module. It has been raising a DeprecationWarning since
Python 2.5.
Also remove gopher support from urllib/urllib2. As both imported gopherlib the
usage of the support would have raised a DeprecationWarning.
........
r55394 | raymond.hettinger | 2007-05-16 18:08:04 -0700 (Wed, 16 May 2007) | 1 line
calendar.py gets no benefit from xrange() instead of range()
........
r55395 | brett.cannon | 2007-05-16 19:02:56 -0700 (Wed, 16 May 2007) | 3 lines
Complete deprecation of BaseException.message. Some subclasses were directly
accessing the message attribute instead of using the descriptor.
........
r55396 | neal.norwitz | 2007-05-16 23:11:36 -0700 (Wed, 16 May 2007) | 4 lines
Reduce the max stack depth to see if this fixes the segfaults on
Windows and some other boxes. If this is successful, this rev should
be backported. I'm not sure how close to the limit we should push this.
........
r55397 | neal.norwitz | 2007-05-16 23:23:50 -0700 (Wed, 16 May 2007) | 4 lines
Set the depth to something very small to try to determine if the
crashes on Windows are really due to the stack size or possibly
some other problem.
........
r55398 | neal.norwitz | 2007-05-17 00:04:46 -0700 (Thu, 17 May 2007) | 4 lines
Last try for tweaking the max stack depth. 5000 was the original value,
4000 didn't work either. 1000 does work on Windows. If 2000 works,
that will hopefully be a reasonable balance.
........
r55412 | fred.drake | 2007-05-17 12:29:58 -0700 (Thu, 17 May 2007) | 1 line
fix argument name in documentation; match the implementation
........
r55427 | neal.norwitz | 2007-05-17 22:47:16 -0700 (Thu, 17 May 2007) | 1 line
Verify neither dumps or loads overflow the stack and segfault.
........
r55446 | collin.winter | 2007-05-18 16:11:24 -0700 (Fri, 18 May 2007) | 1 line
Backport PEP 3110's new 'except' syntax to 2.6.
........
r55448 | raymond.hettinger | 2007-05-18 18:11:16 -0700 (Fri, 18 May 2007) | 1 line
Improvements to NamedTuple's implementation, tests, and documentation
........
r55449 | raymond.hettinger | 2007-05-18 18:50:11 -0700 (Fri, 18 May 2007) | 1 line
Fix beginner mistake -- don't mix spaces and tabs.
........
r55450 | neal.norwitz | 2007-05-18 20:48:47 -0700 (Fri, 18 May 2007) | 1 line
Clear data so random memory does not get freed. Will backport.
........
r55452 | neal.norwitz | 2007-05-18 21:34:55 -0700 (Fri, 18 May 2007) | 3 lines
Whoops, need to pay attention to those test failures.
Move the clear to *before* the first use, not after.
........
r55453 | neal.norwitz | 2007-05-18 21:35:52 -0700 (Fri, 18 May 2007) | 1 line
Give some clue as to what happened if the test fails.
........
r55455 | georg.brandl | 2007-05-19 11:09:26 -0700 (Sat, 19 May 2007) | 2 lines
Fix docstring for add_package in site.py.
........
r55458 | brett.cannon | 2007-05-20 00:09:50 -0700 (Sun, 20 May 2007) | 2 lines
Remove the rgbimg module. It has been deprecated since Python 2.5.
........
r55465 | nick.coghlan | 2007-05-20 04:12:49 -0700 (Sun, 20 May 2007) | 1 line
Fix typo in example (should be backported, but my maintenance branch is woefully out of date)
........
................
r55472 | brett.cannon | 2007-05-20 12:06:18 -0700 (Sun, 20 May 2007) | 2 lines
Remove imageop from the Windows build process.
................
r55486 | neal.norwitz | 2007-05-20 23:59:52 -0700 (Sun, 20 May 2007) | 1 line
Remove callable() builtin
................
r55506 | neal.norwitz | 2007-05-22 00:43:29 -0700 (Tue, 22 May 2007) | 78 lines
Merged revisions 55468-55505 via svnmerge from
svn+ssh://pythondev@svn.python.org/python/trunk
........
r55468 | neal.norwitz | 2007-05-20 11:06:27 -0700 (Sun, 20 May 2007) | 1 line
rotor is long gone.
........
r55470 | neal.norwitz | 2007-05-20 11:43:00 -0700 (Sun, 20 May 2007) | 1 line
Update directories/files at the top-level.
........
r55471 | brett.cannon | 2007-05-20 12:05:06 -0700 (Sun, 20 May 2007) | 2 lines
Try to remove rgbimg from Windows builds.
........
r55474 | brett.cannon | 2007-05-20 16:17:38 -0700 (Sun, 20 May 2007) | 4 lines
Remove the macfs module. This led to the deprecation of macostools.touched();
it completely relied on macfs and is a no-op on OS X according to code
comments.
........
r55476 | brett.cannon | 2007-05-20 16:56:18 -0700 (Sun, 20 May 2007) | 3 lines
Move imgfile import to the global namespace to trigger an import error ASAP to
prevent creation of a test file.
........
r55477 | brett.cannon | 2007-05-20 16:57:38 -0700 (Sun, 20 May 2007) | 3 lines
Cause posixfile to raise a DeprecationWarning. Documented as deprecated since
Ptyhon 1.5.
........
r55479 | andrew.kuchling | 2007-05-20 17:03:15 -0700 (Sun, 20 May 2007) | 1 line
Note removed modules
........
r55481 | martin.v.loewis | 2007-05-20 21:35:47 -0700 (Sun, 20 May 2007) | 2 lines
Add Alexandre Vassalotti.
........
r55482 | george.yoshida | 2007-05-20 21:41:21 -0700 (Sun, 20 May 2007) | 4 lines
fix against r55474 [Remove the macfs module]
Remove "libmacfs.tex" from Makefile.deps and mac/mac.tex.
........
r55487 | raymond.hettinger | 2007-05-21 01:13:35 -0700 (Mon, 21 May 2007) | 1 line
Replace assertion with straight error-checking.
........
r55489 | raymond.hettinger | 2007-05-21 09:40:10 -0700 (Mon, 21 May 2007) | 1 line
Allow all alphanumeric and underscores in type and field names.
........
r55490 | facundo.batista | 2007-05-21 10:32:32 -0700 (Mon, 21 May 2007) | 5 lines
Added timeout support to HTTPSConnection, through the
socket.create_connection function. Also added a small
test for this, and updated NEWS file.
........
r55495 | georg.brandl | 2007-05-21 13:34:16 -0700 (Mon, 21 May 2007) | 2 lines
Patch #1686487: you can now pass any mapping after '**' in function calls.
........
r55502 | neal.norwitz | 2007-05-21 23:03:36 -0700 (Mon, 21 May 2007) | 1 line
Document new params to HTTPSConnection
........
r55504 | neal.norwitz | 2007-05-22 00:16:10 -0700 (Tue, 22 May 2007) | 1 line
Stop using METH_OLDARGS
........
r55505 | neal.norwitz | 2007-05-22 00:16:44 -0700 (Tue, 22 May 2007) | 1 line
Stop using METH_OLDARGS implicitly
........
................
PEP 285. Everything described in the PEP is here, and there is even
some documentation. I had to fix 12 unit tests; all but one of these
were printing Boolean outcomes that changed from 0/1 to False/True.
(The exception is test_unicode.py, which did a type(x) == type(y)
style comparison. I could've fixed that with a single line using
issubtype(x, type(y)), but instead chose to be explicit about those
places where a bool is expected.
Still to do: perhaps more documentation; change standard library
modules to return False/True from predicates.
The comment following used to say:
/* We use ~hash instead of hash, as degenerate hash functions, such
as for ints <sigh>, can have lots of leading zeros. It's not
really a performance risk, but better safe than sorry.
12-Dec-00 tim: so ~hash produces lots of leading ones instead --
what's the gain? */
That is, there was never a good reason for doing it. And to the contrary,
as explained on Python-Dev last December, it tended to make the *sum*
(i + incr) & mask (which is the first table index examined in case of
collison) the same "too often" across distinct hashes.
Changing to the simpler "i = hash & mask" reduced the number of string-dict
collisions (== # number of times we go around the lookup for-loop) from about
6 million to 5 million during a full run of the test suite (these are
approximate because the test suite does some random stuff from run to run).
The number of collisions in non-string dicts also decreased, but not as
dramatically.
Note that this may, for a given dict, change the order (wrt previous
releases) of entries exposed by .keys(), .values() and .items(). A number
of std tests suffered bogus failures as a result. For dicts keyed by
small ints, or (less so) by characters, the order is much more likely to be
in increasing order of key now; e.g.,
>>> d = {}
>>> for i in range(10):
... d[i] = i
...
>>> d
{0: 0, 1: 1, 2: 2, 3: 3, 4: 4, 5: 5, 6: 6, 7: 7, 8: 8, 9: 9}
>>>
Unfortunately. people may latch on to that in small examples and draw a
bogus conclusion.
test_support.py
Moved test_extcall's sortdict() into test_support, made it stronger,
and imported sortdict into other std tests that needed it.
test_unicode.py
Excluced cp875 from the "roundtrip over range(128)" test, because
cp875 doesn't have a well-defined inverse for unicode("?", "cp875").
See Python-Dev for excruciating details.
Cookie.py
Chaged various output functions to sort dicts before building
strings from them.
test_extcall
Fiddled the expected-result file. This remains sensitive to native
dict ordering, because, e.g., if there are multiple errors in a
keyword-arg dict (and test_extcall sets up many cases like that), the
specific error Python complains about first depends on native dict
ordering.
message, and tries to make the messages more consistent and helpful when
the wrong number of arguments or duplicate keyword arguments are supplied.
Comes with more tests for test_extcall.py and and an update to an error
message in test/output/test_pyexpat.
If a non-tuple sequence is passed as the *arg, convert it to a tuple
before checking its length.
If named keyword arguments are used in combination with **kwargs, make
a copy of kwargs before inserting the new keys.
Guido van Rossum
Samuele Pedroni
Tim Peters
Jeremy Hylton
Ka-Ping Yee
Fred Drake