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".
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
(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
........
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.
........
svn+ssh://pythondev@svn.python.org/python/trunk
........
r62260 | gregory.p.smith | 2008-04-10 01:11:56 +0200 (Thu, 10 Apr 2008) | 2 lines
better diagnostics
........
r62261 | gregory.p.smith | 2008-04-10 01:16:37 +0200 (Thu, 10 Apr 2008) | 3 lines
Raise SystemError when size < 0 is passed into PyString_FromStringAndSize,
PyBytes_FromStringAndSize or PyUnicode_FromStringAndSize. [issue2587]
........
r62266 | neal.norwitz | 2008-04-10 07:46:39 +0200 (Thu, 10 Apr 2008) | 5 lines
Remove the test file before writing it in case there is no write permission.
This might help fix some of the failures on Windows box(es). It doesn't hurt
either way and ensure the tests are a little more self contained (ie have
less assumptions).
........
r62271 | gregory.p.smith | 2008-04-10 21:50:36 +0200 (Thu, 10 Apr 2008) | 2 lines
get rid of assert (size >= 0) now that an explicit if (size < 0) is in the code.
........
r62277 | andrew.kuchling | 2008-04-10 23:27:10 +0200 (Thu, 10 Apr 2008) | 1 line
Remove forward-looking statement
........
r62278 | andrew.kuchling | 2008-04-10 23:28:51 +0200 (Thu, 10 Apr 2008) | 1 line
Add punctuation
........
r62279 | andrew.kuchling | 2008-04-10 23:29:01 +0200 (Thu, 10 Apr 2008) | 1 line
Use issue directive
........
r62289 | thomas.heller | 2008-04-11 15:05:38 +0200 (Fri, 11 Apr 2008) | 3 lines
Move backwards compatibility macro to the correct place;
PyIndex_Check() was introduced in Python 2.5.
........
r62290 | thomas.heller | 2008-04-11 16:20:26 +0200 (Fri, 11 Apr 2008) | 2 lines
Performance improvements.
........
r62293 | christian.heimes | 2008-04-12 15:03:03 +0200 (Sat, 12 Apr 2008) | 2 lines
Applied patch #2617 from Frank Wierzbicki wit some extras from me
-J and -X are now reserved for Jython and non-standard arguments (e.g. IronPython). I've added some extra comments to make sure the reservation don't get missed in the future.
........
r62294 | georg.brandl | 2008-04-12 20:11:18 +0200 (Sat, 12 Apr 2008) | 2 lines
Use absolute path in sys.path.
........
r62295 | georg.brandl | 2008-04-12 20:36:09 +0200 (Sat, 12 Apr 2008) | 2 lines
#2615: small consistency update by Jeroen Ruigrok van der Werven.
........
r62296 | georg.brandl | 2008-04-12 21:00:20 +0200 (Sat, 12 Apr 2008) | 2 lines
Add Jeroen.
........
r62297 | georg.brandl | 2008-04-12 21:05:37 +0200 (Sat, 12 Apr 2008) | 2 lines
Don't offend snake lovers.
........
r62298 | gregory.p.smith | 2008-04-12 22:37:48 +0200 (Sat, 12 Apr 2008) | 2 lines
fix compiler warnings
........
r62302 | gregory.p.smith | 2008-04-13 00:24:04 +0200 (Sun, 13 Apr 2008) | 3 lines
socket.error inherits from IOError, it no longer needs listing in
the all_errors tuple.
........
r62303 | brett.cannon | 2008-04-13 01:44:07 +0200 (Sun, 13 Apr 2008) | 8 lines
Re-implement the 'warnings' module in C. This allows for usage of the
'warnings' code in places where it was previously not possible (e.g., the
parser). It could also potentially lead to a speed-up in interpreter start-up
if the C version of the code (_warnings) is imported over the use of the
Python version in key places.
Closes issue #1631171.
........
r62304 | gregory.p.smith | 2008-04-13 02:03:25 +0200 (Sun, 13 Apr 2008) | 3 lines
Adds a profile-opt target for easy compilation of a python binary using
gcc's profile guided optimization.
........
r62305 | brett.cannon | 2008-04-13 02:18:44 +0200 (Sun, 13 Apr 2008) | 3 lines
Fix a bug in PySys_HasWarnOption() where it was not properly checking the
length of the list storing the warning options.
........
r62306 | brett.cannon | 2008-04-13 02:25:15 +0200 (Sun, 13 Apr 2008) | 2 lines
Fix an accidental bug of an non-existent init function.
........
r62308 | andrew.kuchling | 2008-04-13 03:05:59 +0200 (Sun, 13 Apr 2008) | 1 line
Mention -J, -X
........
r62311 | benjamin.peterson | 2008-04-13 04:20:05 +0200 (Sun, 13 Apr 2008) | 2 lines
Give the "Interactive Interpreter Changes" section in 2.6 whatsnew a unique link name
........
r62313 | brett.cannon | 2008-04-13 04:42:36 +0200 (Sun, 13 Apr 2008) | 3 lines
Fix test_warnings by making the state of things more consistent for each test
when it is run.
........
r62314 | skip.montanaro | 2008-04-13 05:17:30 +0200 (Sun, 13 Apr 2008) | 2 lines
spelling
........
r62315 | georg.brandl | 2008-04-13 09:07:44 +0200 (Sun, 13 Apr 2008) | 2 lines
Fix markup.
........
r62319 | christian.heimes | 2008-04-13 11:30:17 +0200 (Sun, 13 Apr 2008) | 1 line
Fix compiler warning Include/warnings.h:19:28: warning: no newline at end of file
........
r62320 | christian.heimes | 2008-04-13 11:33:24 +0200 (Sun, 13 Apr 2008) | 1 line
Use PyString_InternFromString instead of PyString_FromString for static vars
........
r62321 | christian.heimes | 2008-04-13 11:37:05 +0200 (Sun, 13 Apr 2008) | 1 line
Added new files to the pcbuild files
........
svn+ssh://pythondev@svn.python.org/python/trunk
........
r61834 | raymond.hettinger | 2008-03-24 07:07:49 +0100 (Mon, 24 Mar 2008) | 1 line
Tighten documentation for Random.triangular.
........
r61841 | raymond.hettinger | 2008-03-24 09:17:39 +0100 (Mon, 24 Mar 2008) | 1 line
Issue 2460: Make Ellipsis objects copyable.
........
r61842 | georg.brandl | 2008-03-24 10:34:34 +0100 (Mon, 24 Mar 2008) | 2 lines
#1700821: add a note to audioop docs about signedness of sample formats.
........
r61851 | christian.heimes | 2008-03-24 20:57:42 +0100 (Mon, 24 Mar 2008) | 1 line
Added quick hack for bzr
........
r61852 | christian.heimes | 2008-03-24 20:58:17 +0100 (Mon, 24 Mar 2008) | 1 line
Added quick hack for bzr
........
r61853 | amaury.forgeotdarc | 2008-03-24 22:04:10 +0100 (Mon, 24 Mar 2008) | 4 lines
Issue2469: Correct a typo I introduced at r61793: compilation error with UCS4 builds.
All buildbots compile with UCS2...
........
r61863 | neal.norwitz | 2008-03-25 05:17:38 +0100 (Tue, 25 Mar 2008) | 2 lines
Fix a bunch of UnboundLocalErrors when the tests fail.
........
r61864 | neal.norwitz | 2008-03-25 05:18:18 +0100 (Tue, 25 Mar 2008) | 2 lines
Try to fix a bunch of compiler warnings on Win64.
........
r61869 | neal.norwitz | 2008-03-25 07:35:10 +0100 (Tue, 25 Mar 2008) | 3 lines
Don't try to close a non-open file.
Don't let file removal cause the test to fail.
........
r61870 | neal.norwitz | 2008-03-25 08:00:39 +0100 (Tue, 25 Mar 2008) | 7 lines
Try to get this test to be more stable:
* disable gc during the test run because we are spawning objects and there
was an exception when calling Popen.__del__
* Always set an alarm handler so the process doesn't exit if the test fails
(should probably add assertions on the value of hndl_called in more places)
* Using a negative time causes Linux to treat it as zero, so disable that test.
........
r61874 | gregory.p.smith | 2008-03-25 08:31:28 +0100 (Tue, 25 Mar 2008) | 2 lines
Use a 32-bit unsigned int here, a long is not needed.
........
r61889 | georg.brandl | 2008-03-25 12:59:51 +0100 (Tue, 25 Mar 2008) | 2 lines
Move declarations to block start.
........
Raymond Hettinger
Oren Milman
Victor Stinner
Martin Panter
Serhiy Storchaka
Raymond Hettinger
Brett Cannon
Jesus Cea
Kristján Valur Jónsson
Ezio Melotti
Georg Brandl
Benjamin Peterson
Mark Dickinson
Antoine Pitrou
Hirokazu Yamamoto
Christian Heimes