Rather than wrapping the C _isdir function in a Python function,
just import the C _isdir function directly. Additionally, add in the
docstring which was left out.
Éric Araujo [Thu, 9 Jun 2011 05:47:25 +0000 (07:47 +0200)]
Fix example in packaging test_config.
The example C extension used the “three.fast_taunt” name, but no “three” parent
was defined in the setup.cfg. This did not cause a failure nor even print a
warning, we may want to change that.
Brian Curtin [Wed, 8 Jun 2011 23:43:57 +0000 (18:43 -0500)]
Fix #11583. Changed os.path.isdir to use GetFileAttributes instead of os.stat.
By changing to the Windows GetFileAttributes API in nt._isdir we can figure
out if the path is a directory without opening the file via os.stat. This has
the minor benefit of speeding up os.path.isdir by at least 2x for regular
files and 10-15x improvements were seen on symbolic links (which opened the
file multiple times during os.stat). Since os.path.isdir is used in
several places on interpreter startup, we get a minor speedup in startup time.
Brian Curtin [Wed, 8 Jun 2011 23:17:18 +0000 (18:17 -0500)]
Fix #11583. Changed os.path.isdir to use GetFileAttributes instead of os.stat.
By changing to the Windows GetFileAttributes API in nt._isdir we can figure
out if the path is a directory without opening the file via os.stat. This has
the minor benefit of speeding up os.path.isdir by at least 2x for regular
files and 10-15x improvements were seen on symbolic links (which opened the
file multiple times during os.stat). Since os.path.isdir is used in
several places on interpreter startup, we get a minor speedup in startup time.
Antoine Pitrou [Wed, 8 Jun 2011 15:21:55 +0000 (17:21 +0200)]
Issue #9205: concurrent.futures.ProcessPoolExecutor now detects killed
children and raises BrokenProcessPool in such a situation. Previously it
would reliably freeze/deadlock.
Éric Araujo [Wed, 8 Jun 2011 03:29:39 +0000 (05:29 +0200)]
Fix a few misuses of :option: I missed in r86521.
Extract of the commit message:
Fix usage of :option: in the docs (#9312).
:option: is used to create a link to an option of python, not to mark
up any instance of any arbitrary command-line option. These were
changed to ````.
Brett Cannon [Tue, 7 Jun 2011 03:09:10 +0000 (20:09 -0700)]
When building sqlite3, the directory where sqlite.h was found was
always appended to the include directories regardless of whether it
was already in the list of directories. This could cause issue if
sqlite was installed in the same location as another install of
Python. Now a check is done to make sure the directory is not included
twice.
Antoine Pitrou [Mon, 6 Jun 2011 17:35:31 +0000 (19:35 +0200)]
Issue #12040: Expose a new attribute `sentinel` on instances of
:class:`multiprocessing.Process`. Also, fix Process.join() to not use
polling anymore, when given a timeout.
Victor Stinner [Mon, 6 Jun 2011 10:14:23 +0000 (12:14 +0200)]
Issue #12250: regrtest --timeout displays a warning instead of failing with an
error if faulthandler.dump_tracebacks_later() is missing (e.g. if Python is
compiled without threads).
Martin v. Löwis [Mon, 6 Jun 2011 08:27:56 +0000 (10:27 +0200)]
Drop claim about nasty problem.
It's unclear what this was referring to; most likely, it was about sockets
that the application had already closed, in which case it's not a sockets
problem, but an application problem.
Éric Araujo [Sun, 5 Jun 2011 23:54:54 +0000 (01:54 +0200)]
Change reST targets in setup.cfg spec.
I chose “setupcfg” as prefix instead of “packaging-setupcfg” because the scope
of the spec is not limited to packaging: it’s intended as a language-agnostic
document for packaging tools developers as well as Python authors.
Éric Araujo [Sun, 5 Jun 2011 23:52:37 +0000 (01:52 +0200)]
Add a short table of contents to the setup.cfg spec.
I tried shortening the sidebar ToC with the tocdepth option instead, but it has
a bug which caused all headings with a level deeper than the tocdepth value to
all have the same section number, which was a usability regression rather than
in improvement.