Commit Graph

5839 Commits

Author SHA1 Message Date
Serhiy Storchaka 3ade66c203 Issue #17998: Fix an internal error in regular expression engine. 2013-08-03 19:26:33 +03:00
Ned Deily 8074364f7a Issue #17557: Fix os.getgroups() to work with the modified behavior of
getgroups(2) on OS X 10.8.  Original patch by Mateusz Lenik.
2013-08-01 21:19:09 -07:00
Raymond Hettinger 662908b5e5 Restore the data block size to 62.
The former block size traded away good fit within cache lines in
order to gain faster division in deque_item().  However, compilers
are getting smarter and can now replace the slow division operation
with a fast integer multiply and right shift.  Accordingly, it makes
sense to go back to a size that lets blocks neatly fill entire
cache-lines.

GCC-4.8 and CLANG 4.0 both compute "x // 62" with something
roughly equivalent to "x * 9520900167075897609 >> 69".
2013-07-28 02:34:42 -07:00
Raymond Hettinger 69468146b4 Issue #18513: Add workaround for OS X 10.8 cexp bug that leads to wrong cmath.rect(0.0,-0.0) results. 2013-07-20 10:56:58 -07:00
Raymond Hettinger b77ed2c54b Backport c952f3d122ae: Tweak the deque struct by moving the least used fields
(maxlen and weakref) to the end.
2013-07-16 02:34:19 -07:00
Raymond Hettinger 90180c1c3f Move the leftlink to the end of the block structure.
The current pattern of memory access will update both the leftlink and
rightlink at the same time, so they should be positioned side-by-side
for better cache locality.

Keeping the leftlink at the front of the structure would make sense
only if the paired updates were eliminated by backporting changesets
49a9c734304d, 3555cc0ca35b, ae9ee46bd471, and 744dd749e25b.  However,
that isn't likely to happen, so we're better off with the leftlink at
the end of the structure.
2013-07-16 01:59:30 -07:00
Serhiy Storchaka fab65428d5 Issue #18101: Tcl.split() now process Unicode strings nested in a tuple as it
do with byte strings.

Added tests for Tcl.split() and tcl.splitline().
2013-07-11 20:32:48 +03:00
Richard Oudkerk 41072db709 Issue #17097: Make multiprocessing ignore EINTR. 2013-07-01 18:45:28 +01:00
Benjamin Peterson 73d6aca5d2 reapply f1dc30a1be72 2013-06-25 11:35:44 -07:00
Benjamin Peterson 227f0faed2 reapply 5accb0ac8bfb 2013-06-25 11:34:48 -07:00
Benjamin Peterson 3968e29959 reapply f1dc30a1be72 2013-06-25 11:26:20 -07:00
Benjamin Peterson 13dd1f947e reapply f1dc30a1be72 2013-06-25 11:25:26 -07:00
Victor Stinner c1a44269da Issue #18135: ssl.SSLSocket.write() now raises an OverflowError if the input
string in longer than 2 gigabytes. The ssl module does not support partial
write.
2013-06-25 00:48:02 +02:00
Benjamin Peterson 478b08ee5a also backout f1dc30a1be72 for not being a bugfix 2013-06-23 11:38:11 -07:00
Victor Stinner 4807df41ad Issue #18135: Fix a possible integer overflow in ssl.SSLSocket.write()
for strings longer than 2 gigabytes.
2013-06-23 15:15:10 +02:00
Benjamin Peterson 10c74d28e4 backout 5accb0ac8bfb; needs more discussion on python-dev 2013-06-22 11:16:36 -07:00
Andrew Kuchling aa6c297316 #18113: avoid segfault if Py_XDECREF triggers code that calls set_panel_userptr again
Problem noted & original patch by Serhiy Storchaka; I tweaked the patch a bit.
2013-06-22 12:33:05 -04:00
Raymond Hettinger 22b4b4cef4 Arrange structure to match the common access patterns. 2013-06-22 00:51:01 -07:00
Christian Heimes 3aa138fe1d Fix for r84195: add HAVE_ALLOCA_H to configure and only include alloca.h if it's available 2013-06-18 13:25:24 +02:00
Victor Stinner c7b0990a05 ctypes: AIX needs an explicit #include <alloca.h> to get alloca() 2013-06-17 22:03:35 +02:00
Andrew Kuchling a49dcc51b8 #18113: Objects associated to a curses.panel object with set_userptr() were leaked.
Reported by Atsuo Ishimoto.
2013-06-15 13:53:10 -04:00
Raymond Hettinger 03512c18cb Fix comment blocks. Adjust blocksize to a power-of-two for better divmod computations. 2013-06-14 01:06:33 -04:00
Ronald Oussoren 0e6283e68a Ensure that the fix for #17269 also works on OSX 10.4
AI_NUMERICSERV isn't defined on OSX 10.4.
2013-06-10 10:35:36 +02:00
Ronald Oussoren 16c52a3376 Issue #17269: Workaround for a platform bug in getaddrinfo on OSX
Without this patch socket.getaddrinfo crashed when called
with some unusual argument combinations.
2013-05-24 13:45:27 +02:00
Serhiy Storchaka cb1a0283c3 Issue #17979: Fixed the re module in build with --disable-unicode. 2013-05-21 22:53:33 +03:00
doko@ubuntu.com 9855249a9f - Fix typos in the multiprocessing module. 2013-05-15 16:54:16 +02:00
Ezio Melotti 5f4ba6b4e0 #16523: improve attrgetter/itemgetter/methodcaller documentation. 2013-05-08 10:53:11 +03:00
Antoine Pitrou 1e18102a72 Issue #17928: Fix test_structmembers on 64-bit big-endian machines.
(_testcapi isn't Py_ssize_t-clean, the "s#" code should use an int for length)
2013-05-08 02:07:13 +02:00
Antoine Pitrou 38f1afe8d1 Issue #17926: Fix dbm.__contains__ on 64-bit big-endian machines. 2013-05-08 01:51:37 +02:00
Antoine Pitrou 31bc8bef63 Issue #17289: The readline module now plays nicer with external modules or applications changing the rl_completer_word_break_characters global variable.
Initial patch by Bradley Froehle.
2013-05-06 21:51:03 +02:00
Charles-Francois Natali 880359c2bd Issue #12181: select module: Fix struct kevent definition on OpenBSD 64-bit
platforms. Patch by Federico Schwindt.
2013-05-06 21:21:57 +02:00
Antoine Pitrou d8931c375a Issue #14173: Avoid crashing when reading a signal handler during interpreter shutdown. 2013-05-04 23:16:59 +02:00
Gregory P. Smith 06544eacdf This local change was lost during the fixing of issue17192 to update
libffi to 3.0.13.  (i'm not sure if it is needed anymore but see
issue 10309 for details which makes no mention of upstream)
2013-04-30 00:08:51 -07:00
Serhiy Storchaka 8493a04e0f Issue #17857: Prevent build failures with pre-3.5.0 versions of sqlite3,
such as was shipped with Centos 5 and Mac OS X 10.4.

This bug was already fixed in issue14572 for 2.7 only and then it was
backported back from 3.3 in issue17073.
2013-04-28 14:09:47 +03:00
Antoine Pitrou be92971b38 Issue #17710: Fix cPickle raising a SystemError on bogus input. 2013-04-15 21:35:25 +02:00
Serhiy Storchaka 616f2fe28c Issue #17016: Get rid of possible pointer wraparounds and integer overflows
in the re module.  Patch by Nickolai Zeldovich.
2013-04-13 21:15:10 +03:00
doko@ubuntu.com 5ef41abc7c - Issue #17682: Add the _io module to Modules/Setup.dist (commented out). 2013-04-09 17:34:39 +02:00
doko@ubuntu.com 6cb4343d0d - Issue #13150, #17512: sysconfig no longer parses the Makefile and config.h
files when imported, instead doing it at build time.  This makes importing
  sysconfig faster and reduces Python startup time by 20%.
2013-04-08 21:20:09 +02:00
Kristjan Valur Jonsson 7e360b3963 Issue #17662: fix socketmodule compilation with Setup.dist.
Patch contributed by Bohuslav "Slavek" Kabrda (bkabrda)
2013-04-08 16:03:30 +00:00
Serhiy Storchaka 6c467a41bc Revert a premature patch for issue #14010 (changeset d17d10c84d27). 2013-04-06 22:51:29 +03:00
Serhiy Storchaka bb84565cdb Issue #14010: Fix a crash when iterating or deleting deeply nested filters
in itertools module (i.e. itertools.izip(), itertools.chain(), etc).
2013-04-06 22:04:10 +03:00
Benjamin Peterson d6138c426e always allow -1 as a uid 2013-03-23 16:19:04 -05:00
Benjamin Peterson 4d7fc3c5ac undo PyInt -> PyLong change; that was wrong 2013-03-23 15:35:24 -05:00
Benjamin Peterson 31289230e2 return int instead long when possible (#17531) 2013-03-23 15:22:20 -05:00
Kristján Valur Jónsson 9795ca44fb Issue #10212: Support new buffer interface for struct.unpack and
cStringIO
2013-03-19 17:17:47 -07:00
Kristján Valur Jónsson 11f989320e Issue #10296 : Don't handle BreakPoint exceptions using
Structured Exception Handling on windows.
2013-03-19 15:24:10 -07:00
doko@ubuntu.com 4950a3b6cc - Issue #17477: Update the bsddb module to pybsddb 5.3.0, supporting
db-5.x, and dropping support for db-4.1 and db-4.2.
2013-03-19 14:46:29 -07:00
Kristján Valur Jónsson 868f0aac37 issue #9090 : Take the same approach for socketmodule as daytimemodule
when it needs support from timemodule (which is a .so on linux):
link in timemodule.c for the required functions.
2013-03-19 13:53:56 -07:00
R David Murray a39c47aab0 #8862: Fix curses cleanup with getchar is interrupted by a signal.
I have no idea how one would write a test for this.

Patch by July Tikhonov.
2013-03-19 16:26:53 -04:00
Kristján Valur Jónsson 620e36419a issue #9090 : Limit the fix to windows since getting a portable simple
time function on non-windows isn't quite simple.
2013-03-19 13:01:05 -07:00