Thomas Heller [Thu, 12 Jul 2007 19:19:43 +0000 (19:19 +0000)]
Accept bytes as parameter to foreign functions without prototype.
These are passed as byte strings (unicode strings are passed as wide
character strings).
Thomas Heller [Thu, 12 Jul 2007 11:00:22 +0000 (11:00 +0000)]
Py_USING_UNICODE is no longer defined on Windows (on posix it is still
in pyconfig.h). Remove the workaround for the broken (in Python 2.3)
PyUnicode_(From|As)WideChar functions.
Make this a pure-ASCII file. This used to have a Latin-1 coding and
contain several string literals using non-ASCII characters. Because
of the pain of displaying those characters, it's better to use \xXX
escapes for these instead of typing the accented characters.
Apologies if I missed something; it didn't look like there was anything
that depended on this file being encoded in Latin-1 or containing non-ASCII
characters.
Fix for tet_bigmem.py (the failure was caused by a spurious reference
to 'file', which no longer exists).
Fix inspired by Christian Heimes, SF# 1752195.
Fixes for str/uni/bytes for gettext.py. test_gettext.py passes.
Fix by Christian Heimes, SF# 1751958, who writes:
I tested the fixes with the Zope3 zope.app.locales packages.
The mo files are loaded and parsed w/o any problem.
The translation with gettext.gettext is working as expected.
Fixed a refcount leak in _PyUnicode_AsDefaultEncodedString().
Found by Amaury Forgeot d'Arc, SF# 1752317.
(Also removed a redundant check for !errors.)
Get this test working after converting map to return an iterator
................
r56202 | neal.norwitz | 2007-07-09 04:30:09 +0300 (Mon, 09 Jul 2007) | 37 lines
Merged revisions 56124-56201 via svnmerge from
svn+ssh://pythondev@svn.python.org/python/trunk
copy built files to the PCBuild directory, where tools like
distutils or external build processes can find them.
........
r56176 | kurt.kaiser | 2007-07-05 15:03:39 -0700 (Thu, 05 Jul 2007) | 10 lines
Many calls to tk.call involve an arglist containing a single tuple.
Calls using METH_OLDARGS unpack this tuple; calls using METH_VARARG
don't. Tcl's concatenation of args was affected; IDLE doesn't start.
Modify Tkapp_Call() to unpack single tuple arglists.
Fix typo in comment
........
................
r56251 | neal.norwitz | 2007-07-11 10:01:01 +0300 (Wed, 11 Jul 2007) | 1 line
Get working with map returning an iterator (had to fix whitespace too)
................
r56255 | thomas.wouters | 2007-07-11 13:41:37 +0300 (Wed, 11 Jul 2007) | 6 lines
Clean up merge glitch or copy-paste error (the entire module was duplicated,
except the first half even had some more copy-paste errors, referring to
listcomps and genexps instead of setcomps)
................
r56256 | thomas.wouters | 2007-07-11 15:16:01 +0300 (Wed, 11 Jul 2007) | 14 lines
Dict comprehensions. Still needs doc changes (like many python-3000 features
;-). It generates bytecode similar to:
x = {}
for k, v in (generator here):
x[k] = v
except there is no tuple-packing and -unpacking involved. Trivial
measurement suggests it's significantly faster than dict(generator here) (in
the order of 2 to 3 times as fast) but I have not done extensive
measurements.
................
r56263 | guido.van.rossum | 2007-07-11 15:36:26 +0300 (Wed, 11 Jul 2007) | 3 lines
Patch 1724999 by Ali Gholami Rudi -- avoid complaints about dict size
change during iter in destroy call.
................
Patch by Christian Heimes to change self.assert_(x == y) into
self.assertEqual(x, y). (Christian used self.failUnlessEqual(),
but the double negative makes it hard to grok, so I changed it.)
Make test_descr.py pass. Had to disable a few tests, remove references
to 'file', and fix a bunch of subtleties in the behavior of objects
related to overriding __str__. Also disabled a few tests that I couldn't
see how to fix but that seemed to be checking silly stuff only.
Fix test_hashlib regression caused by my fixing hexdigest() to return
a text string instead of bytes. (This being hex digits it *should*
be a text string -- that it wasn't before was a misunderstanding on
my part of the C code.)
Made test_file pass. This meant adding support for read(-1) and read()
to even the most basic file object (I also added readall() which may
be a better API). Also, not all the tests requiring specific failure
modes could be saved. And there were the usual str/bytes issues.
I made sure test_io.py still passes (io.py is now most thoroughly
tested by combining test_file.py and test_io.py).
Upon exit, flush stdout and stderr (twice: before and after the code that
attempts to delete all modules). This makes test_subprocess work again.
(I can't quite figure out why stdout/stderr don't get deleted properly,
which would flush them anyway, but that's a separate issue.)
Make test_uuid passed.
Change the UUID properties to use @property.
Change the UUID API so that .bytes and .bytes_le return, and the
corresponding arguments to __init__() require, values of type 'bytes'.
Changes to ctypes and Mac toolbox glue that fix test_threading and test_platform.
However, test_ctypes is still broken -- and apparently more than before.
Two necessary fixes (but not enough to make tests pass):
- Use os.urandom() as the only source of random bytes.
- Don't reference socket._fileobject; use io.BufferedReader instead.
Fix a place where floor division would be in order.
........
r56135 | guido.van.rossum | 2007-07-01 06:13:54 +0200 (Sun, 01 Jul 2007) | 28 lines
Make map() and filter() identical to itertools.imap() and .ifilter(),
respectively.
I fixed two bootstrap issues, due to the dynamic import of itertools:
1. Starting python requires that map() and filter() are not used until
site.py has added build/lib.<arch> to sys.path.
2. Building python requires that setup.py and distutils and everything
they use is free of map() and filter() calls.
Beyond this, I only fixed the tests in test_builtin.py.
Others, please help fixing the remaining tests that are now broken!
The fixes are usually simple:
a. map(None, X) -> list(X)
b. map(F, X) -> list(map(F, X))
c. map(lambda x: F(x), X) -> [F(x) for x in X]
d. filter(F, X) -> list(filter(F, X))
e. filter(lambda x: P(x), X) -> [x for x in X if P(x)]
Someone, please also contribute a fixer for 2to3 to do this.
It can leave map()/filter() calls alone that are already
inside a list() or sorted() call or for-loop.
Only in rare cases have I seen code that depends on map() of lists
of different lengths going to the end of the longest, or on filter()
of a string or tuple returning an object of the same type; these
will need more thought to fix.
........
r56136 | guido.van.rossum | 2007-07-01 06:22:01 +0200 (Sun, 01 Jul 2007) | 3 lines
Make it so that test_decimal fails instead of hangs, to help automated
test runners.
........
r56139 | georg.brandl | 2007-07-01 18:20:58 +0200 (Sun, 01 Jul 2007) | 2 lines
Fix a few test cases after the map->imap change.
........
r56142 | neal.norwitz | 2007-07-02 06:38:12 +0200 (Mon, 02 Jul 2007) | 1 line
Get a bunch more tests passing after converting map/filter to return iterators.
........
r56147 | guido.van.rossum | 2007-07-02 15:32:02 +0200 (Mon, 02 Jul 2007) | 4 lines
Fix the remaining failing unit tests (at least on OSX).
Also tweaked urllib2 so it doesn't raise socket.gaierror when
all network interfaces are turned off.
........
Guido van Rossum [Sat, 30 Jun 2007 23:44:36 +0000 (23:44 +0000)]
Fix a failure that was only apparent on big-endian machines:
the argument corresponding to 'c' in PyArg_ParseTuple() must be an int,
not a char! (This is new -- Walter Doerwald changed it in r56044.
Note sure this was a good idea.)
Also removed a debug printf() call that was causing compiler warnings.
Regenerate to add True, False, None.
................
r56069 | neal.norwitz | 2007-06-21 22:31:56 -0700 (Thu, 21 Jun 2007) | 1 line
Get the doctest working again after adding None, True, and False as kewyords.
................
r56070 | neal.norwitz | 2007-06-21 23:25:33 -0700 (Thu, 21 Jun 2007) | 1 line
Add space to error message.
................
r56071 | neal.norwitz | 2007-06-21 23:40:04 -0700 (Thu, 21 Jun 2007) | 6 lines
Get pybench working, primarily
* Use print function
* Stop using string module
* Use sorted instead of assuming dict methods return lists
* Convert range result to a list
................
r56089 | collin.winter | 2007-06-26 10:31:48 -0700 (Tue, 26 Jun 2007) | 1 line
Fix AttributeError in distutils/dir_util.py.
................
r56124 | guido.van.rossum | 2007-06-29 18:04:31 -0700 (Fri, 29 Jun 2007) | 30 lines
Merged revisions 56014-56123 via svnmerge from
svn+ssh://pythondev@svn.python.org/python/trunk
Added exclude keyword argument to the TarFile.add() method.
........
r56023 | lars.gustaebel | 2007-06-18 13:05:55 -0700 (Mon, 18 Jun 2007) | 3 lines
Added missing \versionchanged tag for the new exclude
parameter.
........
r56038 | georg.brandl | 2007-06-19 05:36:00 -0700 (Tue, 19 Jun 2007) | 2 lines
Bug #1737864: allow empty message in logging format routines.
........
r56040 | georg.brandl | 2007-06-19 05:38:20 -0700 (Tue, 19 Jun 2007) | 2 lines
All these tests have been flaky wrt reporting leaks. Disable them.
................
r56003 | neal.norwitz | 2007-06-15 19:41:09 -0700 (Fri, 15 Jun 2007) | 1 line
Fix it so test_os no longer reports ref leaks by clearing all the caches
the ABCMeta stores on the class. Apply this to all the ABC collections
as well as the class of os.environ which inherits from an ABC collection.
................
r56013 | neal.norwitz | 2007-06-17 19:56:31 -0700 (Sun, 17 Jun 2007) | 4 lines
This should make it a little easier when adding new collections which use ABCs.
The new subclass won't need to be listed in regrtest.
It will be even better when Guido adds weakrefs. :-)
................
r56014 | martin.v.loewis | 2007-06-17 20:15:51 -0700 (Sun, 17 Jun 2007) | 1 line
Drop inline, as it's not support by VS 2003.
................
r56015 | martin.v.loewis | 2007-06-17 20:17:19 -0700 (Sun, 17 Jun 2007) | 1 line
Expect long objects for DWORD values.
................
r56016 | martin.v.loewis | 2007-06-17 20:18:01 -0700 (Sun, 17 Jun 2007) | 1 line
Drop modules that have been deleted.
................
r56017 | martin.v.loewis | 2007-06-17 20:18:55 -0700 (Sun, 17 Jun 2007) | 1 line
Bump DLL version number to 30.
................
r56018 | neal.norwitz | 2007-06-17 20:55:43 -0700 (Sun, 17 Jun 2007) | 62 lines
Merged revisions 55951-56013 via svnmerge from
svn+ssh://pythondev@svn.python.org/python/trunk
Add T_PYSSIZET in structmember.h: This can be used for
Py_ssize_t members.
Simplify the implementation of UnicodeError objects:
start and end attributes are now stored directly as
Py_ssize_t members, which simplifies various get and
set functions.
........
r55975 | martin.v.loewis | 2007-06-14 13:46:25 -0700 (Thu, 14 Jun 2007) | 3 lines
Patch #1734014: Use _I64_MAX instead of LLONG_MAX.
Will backport to 2.5.
........
r55984 | neal.norwitz | 2007-06-14 20:11:41 -0700 (Thu, 14 Jun 2007) | 4 lines
urllib2_localnet says it leaks probably due to threads. So ignore it.
popen2 is also complaining probably for similar reasons.
make install always reports failure, so don't mail in this case.
........
r56001 | andrew.kuchling | 2007-06-15 15:43:03 -0700 (Fri, 15 Jun 2007) | 1 line
Add a word
........
r56005 | martin.v.loewis | 2007-06-16 03:08:43 -0700 (Sat, 16 Jun 2007) | 2 lines
Guido van Rossum [Fri, 15 Jun 2007 00:00:12 +0000 (00:00 +0000)]
Patch by Ron Adam to make repr(str8(...)) return something looking like
s'...' instead of '...', allowing it to be distinguished from unicode
strings. This doesn't roundtrip -- for now, that's intentional, as the
goal is to rip these out rather than to make it a feature, and making
them stand out (breaking 8 more tests) helps us track them down so we
can rip them out.
Guido van Rossum [Thu, 14 Jun 2007 03:27:55 +0000 (03:27 +0000)]
Modernize the super() call in ABCMeta.__new__() -- I had messed with
it when I thought something was broken, and forgotten to restore it
before checking in (maybe I did a svn revert which had the wrong effect?).