Remove the old comment suggesting that it was desireable to have
blocksize+2 as a multiple of the cache line length. That would
have made sense only if the block structure start point was always
aligned to a cache line boundary. However, the memory allocations
are 16 byte aligned, so we don't really have control over whether
the struct spills across cache line boundaries.
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".
* Add comment explaining the endpoint checks
* Only do the checks in a debug build
* Simplify newblock() to only require a length argument
and leave the link updates to the calling code.
* Also add comment for the freelisting logic.
The division and modulo calculation in deque_item() can be compiled
to fast bitwise operations when the BLOCKLEN is a power of two.
Timing before:
~/cpython $ py -m timeit -r7 -s 'from collections import deque' -s 'd=deque(range(10))' 'd[5]'
10000000 loops, best of 7: 0.0627 usec per loop
Timing after:
~/cpython $ py -m timeit -r7 -s 'from collections import deque' -s 'd=deque(range(10))' 'd[5]'
10000000 loops, best of 7: 0.0581 usec per loop
* Clarified comment on the impact of BLOCKLEN on deque_index
(with a power-of-two, the division and modulo
computations are done with a right-shift and bitwise-and).
* Clarified comment on the overflow check to note that
it is general and not just applicable the 64-bit builds.
* In deque._rotate(), the "deque->" indirections are
factored-out of the loop (loop invariant code motion),
leaving the code cleaner looking and slightly faster.
* In deque._rotate(), replaced the memcpy() with an
equivalent loop. That saved the memcpy setup time
and allowed the pointers to move in their natural
leftward and rightward directions.
See comparative timings at: http://pastebin.com/p0RJnT5N
(Reviewed by Georg Brandl.)
Also made similar changes to deque.reverse() though this wasn't
strictly necessary (the edge case cannot occur with two pointers
moving to meet in the middle). Making the change in reverse()
was more a matter of future-proofing.
svn+ssh://pythondev@svn.python.org/python/trunk
........
r68112 | benjamin.peterson | 2009-01-01 00:48:39 +0100 (Thu, 01 Jan 2009) | 1 line
#4795 inspect.isgeneratorfunction() should return False instead of None
........
r68115 | benjamin.peterson | 2009-01-01 05:04:41 +0100 (Thu, 01 Jan 2009) | 1 line
simplfy code
........
r68120 | georg.brandl | 2009-01-01 13:15:31 +0100 (Thu, 01 Jan 2009) | 4 lines
#4228: Pack negative values the same way as 2.4
in struct's L format.
........
r68133 | antoine.pitrou | 2009-01-01 16:38:03 +0100 (Thu, 01 Jan 2009) | 1 line
fill in actual issue number in tests
........
r68141 | benjamin.peterson | 2009-01-01 17:43:12 +0100 (Thu, 01 Jan 2009) | 1 line
fix highlighting
........
r68142 | benjamin.peterson | 2009-01-01 18:29:49 +0100 (Thu, 01 Jan 2009) | 2 lines
welcome to 2009, Python!
........
r68145 | amaury.forgeotdarc | 2009-01-02 01:03:54 +0100 (Fri, 02 Jan 2009) | 5 lines
#4801 _collections module fails to build on cygwin.
_PyObject_GC_TRACK is the macro version of PyObject_GC_Track,
and according to documentation it should not be used for extension modules.
........
r68146 | ronald.oussoren | 2009-01-02 11:44:46 +0100 (Fri, 02 Jan 2009) | 2 lines
Fix for issue4472: "configure --enable-shared doesn't work on OSX"
........
r68148 | ronald.oussoren | 2009-01-02 11:48:31 +0100 (Fri, 02 Jan 2009) | 2 lines
Forgot to add a NEWS item in my previous checkin
........
r68149 | ronald.oussoren | 2009-01-02 11:50:48 +0100 (Fri, 02 Jan 2009) | 2 lines
Fix for issue4780
........
svn+ssh://pythondev@svn.python.org/python/trunk
........
r68128 | antoine.pitrou | 2009-01-01 15:11:22 +0100 (jeu., 01 janv. 2009) | 3 lines
Issue #3680: Reference cycles created through a dict, set or deque iterator did not get collected.
........
The 2.x dict.iteritems() returns an iterator,
whereas the 3.0 dict.items() returns a "view",
which is iterable, but not an iterator with its __next__ method.
Patch by Hirokazu Yamamoto.
svn+ssh://pythondev@svn.python.org/python/trunk
........
r66377 | amaury.forgeotdarc | 2008-09-11 00:04:45 +0200 (jeu., 11 sept. 2008) | 8 lines
#3743: PY_FORMAT_SIZE_T is designed for the OS "printf" functions, not for
PyString_FromFormat which has an independent implementation, and uses "%zd".
This makes a difference on win64, where printf needs "%Id" to display
64bit values. For example, queue.__repr__ was incorrect.
Reviewed by Martin von Loewis.
........
svn+ssh://pythondev@svn.python.org/python/trunk
........
r65910 | benjamin.peterson | 2008-08-20 09:07:59 -0500 (Wed, 20 Aug 2008) | 1 line
fix up the multiprocessing docs a little
........
r65977 | christian.heimes | 2008-08-22 14:47:25 -0500 (Fri, 22 Aug 2008) | 3 lines
Silenced compiler warning
Objects/stringlib/find.h:97: warning: 'stringlib_contains_obj' defined but not used
Reviewed by Benjamin Peterson
........
r65980 | christian.heimes | 2008-08-22 15:10:27 -0500 (Fri, 22 Aug 2008) | 3 lines
Fixed two format strings in the _collections module. For example
Modules/_collectionsmodule.c:674: warning: format '%i' expects type 'int', but argument 2 has type 'Py_ssize_t'
Reviewed by Benjamin Peterson
........
r65984 | christian.heimes | 2008-08-22 16:23:47 -0500 (Fri, 22 Aug 2008) | 1 line
d is the correct format string
........
r65986 | mark.hammond | 2008-08-22 19:59:14 -0500 (Fri, 22 Aug 2008) | 2 lines
Fix bug 3625: test issues on 64bit windows. r=pitrou
........
r66000 | benjamin.peterson | 2008-08-23 15:27:43 -0500 (Sat, 23 Aug 2008) | 5 lines
#3643 add a few more checks to _testcapi to prevent segfaults
Author: Victor Stinner
Reviewer: Benjamin Peterson
........
r66011 | neal.norwitz | 2008-08-24 12:27:43 -0500 (Sun, 24 Aug 2008) | 1 line
Ignore a couple more tests that report leaks inconsistently.
........
r66012 | neal.norwitz | 2008-08-24 12:29:53 -0500 (Sun, 24 Aug 2008) | 1 line
Use the actual blacklist of leaky tests
........
r66014 | georg.brandl | 2008-08-24 13:11:07 -0500 (Sun, 24 Aug 2008) | 2 lines
#3654: fix duplicate test method name. Review by Benjamin P.
........
r66017 | benjamin.peterson | 2008-08-24 16:55:03 -0500 (Sun, 24 Aug 2008) | 1 line
remove note about unimplemented feature
........
r66020 | brett.cannon | 2008-08-24 18:15:19 -0500 (Sun, 24 Aug 2008) | 1 line
Clarify that some attributes/methods are listed somewhat separately because they are not part of the threading API.
........
svn+ssh://pythondev@svn.python.org/python/trunk
........
r60679 | raymond.hettinger | 2008-02-09 02:18:42 +0100 (Sat, 09 Feb 2008) | 1 line
Make ABC containers inherit as documented.
........
r60684 | raymond.hettinger | 2008-02-09 04:34:52 +0100 (Sat, 09 Feb 2008) | 1 line
Merge with r60683.
........
r60687 | raymond.hettinger | 2008-02-09 05:37:49 +0100 (Sat, 09 Feb 2008) | 1 line
Add -3 warnings that set.copy(), dict.copy(), and defaultdict.copy() will go away in Py3.x
........
r60689 | raymond.hettinger | 2008-02-09 11:04:19 +0100 (Sat, 09 Feb 2008) | 1 line
Metaclass declaration is inherited
........
r60691 | raymond.hettinger | 2008-02-09 11:06:20 +0100 (Sat, 09 Feb 2008) | 1 line
Temporarily disable this test. It's been broken for a week.
........
r60695 | nick.coghlan | 2008-02-09 16:28:09 +0100 (Sat, 09 Feb 2008) | 1 line
Issue 2021: Allow NamedTemporaryFile and SpooledTemporaryFile to be used as context managers. (The NamedTemporaryFile fix should be considered for backporting to 2.5)
........
svn+ssh://pythondev@svn.python.org/python/trunk
........
r60618 | walter.doerwald | 2008-02-06 15:31:55 +0100 (Wed, 06 Feb 2008) | 6 lines
Remove month parameter from Calendar.yeardatescalendar(),
Calendar.yeardays2calendar() and Calendar.yeardayscalendar() as the methods
don't have such a parameter. Fixes issue #2017.
Rewrap content to 80 chars.
........
r60622 | facundo.batista | 2008-02-06 20:28:49 +0100 (Wed, 06 Feb 2008) | 4 lines
Fixes issue 1959. Converted tests to unittest.
Thanks Giampaolo Rodola.
........
r60626 | thomas.heller | 2008-02-06 21:29:17 +0100 (Wed, 06 Feb 2008) | 3 lines
Fixed refcounts and error handling.
Should not be merged to py3k branch.
........
r60630 | mark.dickinson | 2008-02-06 23:10:50 +0100 (Wed, 06 Feb 2008) | 4 lines
Issue 1979: Make Decimal comparisons (other than !=, ==) involving NaN
raise InvalidOperation (and return False if InvalidOperation is trapped).
........
r60632 | mark.dickinson | 2008-02-06 23:25:16 +0100 (Wed, 06 Feb 2008) | 2 lines
Remove incorrect usage of :const: in documentation.
........
r60634 | georg.brandl | 2008-02-07 00:45:51 +0100 (Thu, 07 Feb 2008) | 2 lines
Revert accidental changes to test_queue in r60605.
........
r60636 | raymond.hettinger | 2008-02-07 01:54:20 +0100 (Thu, 07 Feb 2008) | 1 line
Issue 2025: Add tuple.count() and tuple.index() to follow the ABC in collections.Sequence.
........
r60637 | mark.dickinson | 2008-02-07 02:14:23 +0100 (Thu, 07 Feb 2008) | 2 lines
Fix broken link in decimal documentation.
........
r60638 | mark.dickinson | 2008-02-07 02:42:06 +0100 (Thu, 07 Feb 2008) | 3 lines
IEEE 754 should be IEEE 854; give precise reference for
comparisons involving NaNs.
........
r60639 | raymond.hettinger | 2008-02-07 03:12:52 +0100 (Thu, 07 Feb 2008) | 1 line
Return ints instead of longs for tuple.count() and tuple.index().
........
r60640 | raymond.hettinger | 2008-02-07 04:10:33 +0100 (Thu, 07 Feb 2008) | 1 line
Merge 60627.
........
r60641 | raymond.hettinger | 2008-02-07 04:25:46 +0100 (Thu, 07 Feb 2008) | 1 line
Merge r60628, r60631, and r60633. Register UserList and UserString will the appropriate ABCs.
........
r60642 | brett.cannon | 2008-02-07 08:47:31 +0100 (Thu, 07 Feb 2008) | 3 lines
Cast a struct to a void pointer so as to do a type-safe pointer comparison
(mistmatch found by clang).
........
r60643 | brett.cannon | 2008-02-07 09:04:07 +0100 (Thu, 07 Feb 2008) | 2 lines
Remove unnecessary curly braces around an int literal.
........
r60644 | andrew.kuchling | 2008-02-07 12:43:47 +0100 (Thu, 07 Feb 2008) | 1 line
Update URL
........
r60645 | facundo.batista | 2008-02-07 17:16:29 +0100 (Thu, 07 Feb 2008) | 4 lines
Fixes issue 2026. Tests converted to unittest. Thanks
Giampaolo Rodola.
........
r60646 | christian.heimes | 2008-02-07 18:15:30 +0100 (Thu, 07 Feb 2008) | 1 line
Added some statistics code to dict and list object code. I wanted to test how a larger freelist affects the reusage of freed objects. Contrary to my gut feelings 80 objects is more than fine for small apps. I haven't profiled a large app yet.
........
r60648 | facundo.batista | 2008-02-07 20:06:52 +0100 (Thu, 07 Feb 2008) | 6 lines
Fixes Issue 1401. When redirected, a possible POST get converted
to GET, so it loses its payload. So, it also must lose the
headers related to the payload (if it has no content any more,
it shouldn't indicate content length and type).
........
r60649 | walter.doerwald | 2008-02-07 20:30:22 +0100 (Thu, 07 Feb 2008) | 3 lines
Clarify that the output of TextCalendar.formatmonth() and
TextCalendar.formatyear() for custom instances won't be influenced by calls
to the module global setfirstweekday() function. Fixes#2018.
........
r60651 | walter.doerwald | 2008-02-07 20:48:34 +0100 (Thu, 07 Feb 2008) | 3 lines
Fix documentation for Calendar.iterweekdays(): firstweekday is a property.
Fixes second part of #2018.
........
r60653 | walter.doerwald | 2008-02-07 20:57:32 +0100 (Thu, 07 Feb 2008) | 2 lines
Fix typo in docstring for Calendar.itermonthdays().
........
r60655 | raymond.hettinger | 2008-02-07 21:04:37 +0100 (Thu, 07 Feb 2008) | 1 line
The float conversion recipe is simpler in Py2.6
........
r60657 | raymond.hettinger | 2008-02-07 21:10:49 +0100 (Thu, 07 Feb 2008) | 1 line
Fix typo
........
r60660 | brett.cannon | 2008-02-07 23:27:10 +0100 (Thu, 07 Feb 2008) | 3 lines
Make sure a switch statement does not have repetitive case statements.
Error found through LLVM post-2.1 svn.
........
r60661 | christian.heimes | 2008-02-08 01:11:31 +0100 (Fri, 08 Feb 2008) | 1 line
Deallocate content of the dict free list on interpreter shutdown
........
r60662 | christian.heimes | 2008-02-08 01:14:34 +0100 (Fri, 08 Feb 2008) | 1 line
Use prefix decrement
........
r60663 | amaury.forgeotdarc | 2008-02-08 01:56:02 +0100 (Fri, 08 Feb 2008) | 5 lines
issue 2045: Infinite recursion when printing a subclass of defaultdict,
if default_factory is set to a bound method.
Will backport.
........
r60667 | jeffrey.yasskin | 2008-02-08 07:45:40 +0100 (Fri, 08 Feb 2008) | 2 lines
Oops! 2.6's Rational.__ne__ didn't work.
........
r60671 | hyeshik.chang | 2008-02-08 18:10:20 +0100 (Fri, 08 Feb 2008) | 2 lines
Update big5hkscs codec to conform to the HKSCS:2004 revision.
........
r60673 | raymond.hettinger | 2008-02-08 23:30:04 +0100 (Fri, 08 Feb 2008) | 4 lines
Remove unnecessary modulo division.
The preceding test guarantees that 0 <= i < len.
........
r60674 | raymond.hettinger | 2008-02-09 00:02:27 +0100 (Sat, 09 Feb 2008) | 1 line
Speed-up __iter__() mixin method.
........
r60675 | raymond.hettinger | 2008-02-09 00:34:21 +0100 (Sat, 09 Feb 2008) | 1 line
Fill-in missing Set comparisons
........
r60677 | raymond.hettinger | 2008-02-09 00:57:06 +0100 (Sat, 09 Feb 2008) | 1 line
Add advice on choosing between DictMixin and MutableMapping
........
Martin Panter
Raymond Hettinger
Eric Snow
Benjamin Peterson
Victor Stinner
Andrew Svetlov
Jesus Cea
Kristján Valur Jónsson
Antoine Pitrou
Martin v. Löwis
Brian Curtin
Victor Stinner
Georg Brandl
Mark Dickinson
Amaury Forgeot d'Arc
Nick Coghlan
Christian Heimes