After a long discussion about the problem with Windows GUI apps Guido decided that sys.stdin, stdout and stderr should be None when the C runtime library returns invalid file descriptors for the standard streams.
So far the only known cases are Windows GUI apps and scripts started with pythonw on Windows. The OS restrictions are tight enough to catch the problem on other OSes.
When an unfinished generator-iterator is garbage collected, PyEval_EvalFrameEx
is called with a GeneratorExit exception set. This leads to funny results
if the sys.settrace function itself makes use of generators.
A visible effect is that the settrace function is reset to None.
Another is that the eventual "finally" block of the generator is not called.
It is necessary to save/restore the exception around the call to the trace
function.
This happens a lot with py3k: isinstance() of an ABCMeta instance runs
def __instancecheck__(cls, instance):
"""Override for isinstance(instance, cls)."""
return any(cls.__subclasscheck__(c)
for c in {instance.__class__, type(instance)})
which lets an opened generator expression each time it returns True.
Seems a backport candidate, even if the case is less frequent in 2.5.
Added some more methods and attributes to stdprinter to ease up debugging #1415 on Windows. It also makes the object more file like. Now it can be used as a working replacement for an io instance.
svn+ssh://pythondev@svn.python.org/python/trunk
........
r58931 | vinay.sajip | 2007-11-11 15:27:30 +0100 (Sun, 11 Nov 2007) | 1 line
Fixed a bug reported (in private email, by Robert Crida) in logging configuration whereby child loggers of a logger named in a configuration file, which are not themselves named in the configuration, are disabled when the configuration is applied.
........
r58932 | georg.brandl | 2007-11-11 16:16:16 +0100 (Sun, 11 Nov 2007) | 2 lines
Remove duplication of "this".
........
r58935 | christian.heimes | 2007-11-12 02:15:40 +0100 (Mon, 12 Nov 2007) | 2 lines
Added new decorator syntax to property.__doc__
Guido prefers _x over __x.
........
r58936 | christian.heimes | 2007-11-12 02:20:56 +0100 (Mon, 12 Nov 2007) | 2 lines
Fix for #1427: Error in standard module calendar
the prweek() method is still broken and I can't figure out how it suppose to work.
........
r58938 | andrew.kuchling | 2007-11-12 02:25:21 +0100 (Mon, 12 Nov 2007) | 1 line
Re-word sentence
........
raw string literals. I added a whole bunch of tests but am still not sure
I am testing all paths through the code. I really think the code could be
simplified quite a bit.
On Windows fileno(stdout) and fileno(stderr) can return an invalid file descriptor number (-2 on my machine). It happens only for pythonw.exe but not for python.exe.
Catch the problem ASAP in PyFile_NewStdPrinter(). I've also removed the call to PyErr_BadInternalCall(). It was causing a seg fault because the exceptions aren't available yet.
I figured out that the files may suite other developers. The files take care of most problems and makes building Python on Windows easier. Simply double click on build_env.bat to open a shell and type build.
The patch doesn't do the whole name mangling mambo jambo on purpose. MS sure does some weird things and I don't feel like reimplementing ntpath.normpath in C. If the user does deliberately something stupid he is on his own.
TODO: Backport to 2.6 (maybe 2.5?) and document that users should only do sys.path.append(os.path.normpath(somepath)) if they want to be on the safe side.
No detailed change log; just check out the change log for the py3k-pep3137
branch. The most obvious changes:
- str8 renamed to bytes (PyString at the C level);
- bytes renamed to buffer (PyBytes at the C level);
- PyString and PyUnicode are no longer compatible.
I.e. we now have an immutable bytes type and a mutable bytes type.
The behavior of PyString was modified quite a bit, to make it more
bytes-like. Some changes are still on the to-do list.
svn+ssh://pythondev@svn.python.org/python/trunk
........
r58868 | gregory.p.smith | 2007-11-05 16:19:03 -0800 (Mon, 05 Nov 2007) | 3 lines
Fixes Issue 1385: The hmac module now computes the correct hmac when using
hashes with a block size other than 64 bytes (such as sha384 and sha512).
........
svn+ssh://pythondev@svn.python.org/python/trunk
........
r58822 | brett.cannon | 2007-11-02 23:47:02 -0700 (Fri, 02 Nov 2007) | 2 lines
Add a missing quotation mark.
........
r58840 | skip.montanaro | 2007-11-04 07:56:52 -0800 (Sun, 04 Nov 2007) | 2 lines
Note change to get_dialect semantics in 2.5. Will backport to 2.5.
........
r58844 | georg.brandl | 2007-11-04 09:43:49 -0800 (Sun, 04 Nov 2007) | 2 lines
Fix syntax for versionchanged markup.
........
r58850 | gregory.p.smith | 2007-11-04 18:32:26 -0800 (Sun, 04 Nov 2007) | 9 lines
Fixes bug 477182 on pybsddb.sf.net. DB objects now load the flags and
pay attention to them when opening an existing database. This means
that d[] behaves properly even on databases previously created with DB_DUP
or DB_DUPSORT flags to allow duplicate keys.
http://sourceforge.net/tracker/index.php?func=detail&aid=477182&group_id=13900&atid=113900
Do not backport, this bugfix could be considered an API change.
........
r58851 | gregory.p.smith | 2007-11-04 18:56:31 -0800 (Sun, 04 Nov 2007) | 3 lines
Add the bsddb.db.DBEnv.lock_id_free method.
Improve test_lock's tempdir creation and cleanup.
........
r58852 | gregory.p.smith | 2007-11-05 01:06:28 -0800 (Mon, 05 Nov 2007) | 3 lines
* db->get_types is only available in BerkeleyDB >= 4.2
* get compiling with older versions of python again for a stand alone release.
........
r58853 | gregory.p.smith | 2007-11-05 01:07:40 -0800 (Mon, 05 Nov 2007) | 2 lines
* db->get_flags is only available in BerkeleyDB >= 4.2
........
r58854 | mark.summerfield | 2007-11-05 01:22:48 -0800 (Mon, 05 Nov 2007) | 3 lines
Added cross-references between the various archive file formats.
........
r58857 | mark.summerfield | 2007-11-05 06:38:50 -0800 (Mon, 05 Nov 2007) | 5 lines
Clarified the fact that you can have comments for individual archive
members even though comments to the archive itself aren't currently
supported.
........
Added PyUnicode_DecodeFSDefaultAndSize
Fixed a problem with the sys.path code that caused a segfault on Windows when the path contains non ASCII chars. The code for sys.executable, exec_prefix and prefix should be fixed, too.
svn+ssh://pythondev@svn.python.org/python/trunk
........
r58745 | georg.brandl | 2007-11-01 10:19:33 -0700 (Thu, 01 Nov 2007) | 2 lines
#1364: os.lstat is available on Windows too, as an alias to os.stat.
........
r58750 | christian.heimes | 2007-11-01 12:48:10 -0700 (Thu, 01 Nov 2007) | 1 line
Backport of import tests for bug http://bugs.python.org/issue1293 and bug http://bugs.python.org/issue1342
........
r58751 | christian.heimes | 2007-11-01 13:11:06 -0700 (Thu, 01 Nov 2007) | 1 line
Removed non ASCII text from test as requested by Guido. Sorry :/
........
r58753 | georg.brandl | 2007-11-01 13:37:02 -0700 (Thu, 01 Nov 2007) | 2 lines
Fix markup glitch.
........
r58757 | gregory.p.smith | 2007-11-01 14:08:14 -0700 (Thu, 01 Nov 2007) | 4 lines
Fix bug introduced in revision 58385. Database keys could no longer
have NULL bytes in them. Replace the errant strdup with a
malloc+memcpy. Adds a unit test for the correct behavior.
........
r58758 | gregory.p.smith | 2007-11-01 14:15:36 -0700 (Thu, 01 Nov 2007) | 3 lines
Undo revision 58533 58534 fixes. Those were a workaround for
a problem introduced by 58385.
........
r58759 | gregory.p.smith | 2007-11-01 14:17:47 -0700 (Thu, 01 Nov 2007) | 2 lines
false "fix" undone as correct problem was found and fixed.
........
r58765 | mark.summerfield | 2007-11-02 01:24:59 -0700 (Fri, 02 Nov 2007) | 3 lines
Added more file-handling related cross-references.
........
r58766 | nick.coghlan | 2007-11-02 03:09:12 -0700 (Fri, 02 Nov 2007) | 1 line
Fix for bug 1705170 - contextmanager swallowing StopIteration (2.5 backport candidate)
........
r58784 | thomas.heller | 2007-11-02 12:10:24 -0700 (Fri, 02 Nov 2007) | 4 lines
Issue #1292: On alpha, arm, ppc, and s390 linux systems the
--with-system-ffi configure option defaults to "yes" because the
bundled libffi sources are too old.
........
r58785 | thomas.heller | 2007-11-02 12:11:23 -0700 (Fri, 02 Nov 2007) | 1 line
Enable the full ctypes c_longdouble tests again.
........
r58796 | georg.brandl | 2007-11-02 13:06:17 -0700 (Fri, 02 Nov 2007) | 4 lines
Make "hashable" a glossary entry and clarify docs on __cmp__, __eq__ and __hash__.
I hope the concept of hashability is better understandable now.
Thanks to Tim Hatch for pointing out the flaws here.
........
Merge r58757 and r58758 from trunk.
Undoes incorrect dbtables fix and errant strdup introduced as
described below:
r58757 | gregory.p.smith | 2007-11-01 14:08:14 -0700 (Thu, 01 Nov 2007) | 4 lines
Fix bug introduced in revision 58385. Database keys could no longer
have NULL bytes in them. Replace the errant strdup with a
malloc+memcpy. Adds a unit test for the correct behavior.
r58758 | gregory.p.smith | 2007-11-01 14:15:36 -0700 (Thu, 01 Nov 2007) | 3 lines
Undo revision 58533 58534 fixes. Those were a workaround for
a problem introduced by 58385.
Christian Heimes
Amaury Forgeot d'Arc
Guido van Rossum
Thomas Heller
Brett Cannon
Georg Brandl
Gregory P. Smith