Change PyAPI_FUNC(type), PyAPI_DATA(type) and PyMODINIT_FUNC macros
of pyport.h when Py_BUILD_CORE_MODULE is defined.
The Py_BUILD_CORE_MODULE define must be now be used to build a C
extension as a dynamic library accessing Python internals: export the
PyInit_xxx() function in DLL exports on Windows.
Changes:
* Py_BUILD_CORE_BUILTIN and Py_BUILD_CORE_MODULE now imply
Py_BUILD_CORE directy in pyport.h.
* ceval.c compilation now fails with an error if Py_BUILD_CORE is not
defined, just to ensure that Python is build with the correct
defines.
* setup.py now compiles _pickle.c with Py_BUILD_CORE_MODULE define.
* setup.py compiles _json.c with Py_BUILD_CORE_MODULE define, rather
than Py_BUILD_CORE_BUILTIN define
* PCbuild/pythoncore.vcxproj: Add Py_BUILD_CORE_BUILTIN define.
* Disable x87 control word for non-x86 targets
On msvc, x87 control word is only available on x86 target. Need to disable it for other targets to prevent compiling problems.
* Include immintrin.h on x86 and x64 only
Immintrin.h is only available on x86 and x64. Need to disable it for other targets to prevent compiling problems.
* Add Py_STATIC_INLINE() macro to declare a "static inline" function.
If the compiler supports it, try to always inline the function even if no
optimization level was specified.
* Modify pydtrace.h to use Py_STATIC_INLINE() when WITH_DTRACE is
not defined.
* Add an unit test on Py_DECREF() to make sure that
_Py_NegativeRefcount() reports the correct filename.
Fix the pthread+semaphore implementation of
PyThread_acquire_lock_timed() when called with timeout > 0 and
intr_flag=0: recompute the timeout if sem_timedwait() is interrupted
by a signal (EINTR).
See also the PEP 475.
The pthread implementation of PyThread_acquire_lock() now fails with
a fatal error if the timeout is larger than PY_TIMEOUT_MAX, as done
in the Windows implementation.
The check prevents any risk of overflow in PyThread_acquire_lock().
Add also PY_DWORD_MAX constant.
Issue #29234: Inlining _PyStack_AsTuple() into callers increases their stack
consumption, Disable inlining to optimize the stack consumption.
Add _Py_NO_INLINE: use __attribute__((noinline)) of GCC and Clang.
It reduces the stack consumption, bytes per call, before => after:
test_python_call: 1040 => 976 (-64 B)
test_python_getitem: 976 => 912 (-64 B)
test_python_iterator: 1120 => 1056 (-64 B)
=> total: 3136 => 2944 (- 192 B)
When Python is not compiled with PGO, the performance of Python on call_simple
and call_method microbenchmarks depend highly on the code placement. In the
worst case, the performance slowdown can be up to 70%.
The GCC __attribute__((hot)) attribute helps to keep hot code close to reduce
the risk of such major slowdown. This attribute is ignored when Python is
compiled with PGO.
The following functions are considered as hot according to statistics collected
by perf record/perf report:
* _PyEval_EvalFrameDefault()
* call_function()
* _PyFunction_FastCall()
* PyFrame_New()
* frame_dealloc()
* PyErr_Occurred()
computation as the overflow behavior of signed integers is undefined.
NOTE: This change is smaller compared to 3.2 as much of this cleanup had
already been done. I added the comment that my change in 3.2 added so that the
code would match up. Otherwise this just adds or synchronizes appropriate UL
designations on some constants to be pedantic.
In practice we require compiling everything with -fwrapv which forces overflow
to be defined as twos compliment but this keeps the code cleaner for checkers
or in the case where someone has compiled it without -fwrapv or their
compiler's equivalent. We could work to get rid of the -fwrapv requirement
in 3.4 but that requires more planning.
Found by Clang trunk's Undefined Behavior Sanitizer (UBSan).
Cleanup only - no functionality or hash values change.
computation as the overflow behavior of signed integers is undefined.
NOTE: This change is smaller compared to 3.2 as much of this cleanup had
already been done. I added the comment that my change in 3.2 added so that the
code would match up. Otherwise this just adds or synchronizes appropriate UL
designations on some constants to be pedantic.
In practice we require compiling everything with -fwrapv which forces overflow
to be defined as twos compliment but this keeps the code cleaner for checkers
or in the case where someone has compiled it without -fwrapv or their
compiler's equivalent.
Found by Clang trunk's Undefined Behavior Sanitizer (UBSan).
Cleanup only - no functionality or hash values change.
computation as the overflow behavior of signed integers is undefined.
In practice we require compiling everything with -fwrapv which forces overflow
to be defined as twos compliment but this keeps the code cleaner for checkers
or in the case where someone has compiled it without -fwrapv or their
compiler's equivalent.
Found by Clang trunk's Undefined Behavior Sanitizer (UBSan).
Cleanup only - no functionality or hash values change.
to a single #define instead of having several copies in several files.
This excludes the Modules/ tree (datetime and expat both have a copy
for their own purposes with no need for it to be the same).
(instances of int, float, complex, decimal.Decimal and
fractions.Fraction) that makes it easy to maintain the invariant that
hash(x) == hash(y) whenever x and y have equal value.
svn+ssh://pythondev@svn.python.org/python/trunk
........
r76561 | mark.dickinson | 2009-11-28 12:30:36 +0000 (Sat, 28 Nov 2009) | 5 lines
Include ieeefp.h (when available) in pyport.h instead of individually in
Objects/floatobject.c and Objects/complexobject.c. This should silence
compiler warnings about implicit declaration of the 'finite' function
on Solaris.
........
problems for binary distributions of Python in situations
where the build machine has SSE2 but the target machine
does not.
Therefore, don't enable SSE2 instructions automatically on x86.
- incorporate and adapt David Gay's dtoa and strtod
into the Python core
- on platforms where we can use Gay's code (almost
all!), repr(float) is based on the shortest
sequence of decimal digits that rounds correctly.
- add sys.float_repr_style attribute to indicate
whether we're using Gay's code or not
- add autoconf magic to detect and enable SSE2
instructions on x86/gcc
- slight change to repr and str: repr switches
to exponential notation at 1e16 instead of
1e17, str switches at 1e11 instead of 1e12
svn+ssh://pythondev@svn.python.org/python/trunk
........
r70489 | mark.dickinson | 2009-03-20 23:16:14 +0000 (Fri, 20 Mar 2009) | 4 lines
Rewrite Py_ARITHMETIC_RIGHT_SHIFT so that it's valid for all signed
integer types T, not just those for which "unsigned T" is legal.
........
- new configure option --enable-big-digits
- new structseq sys.int_info giving information about the internal format
By default, 30-bit digits are enabled on 64-bit machines but
disabled on 32-bit machines.
svn+ssh://pythondev@svn.python.org/python/trunk
........
r68450 | jeffrey.yasskin | 2009-01-09 10:47:07 -0600 (Fri, 09 Jan 2009) | 3 lines
Fix issue 4884, preventing a crash in the socket code when python is compiled
with llvm-gcc and run with a glibc <2.10.
........
r68480 | vinay.sajip | 2009-01-10 07:38:04 -0600 (Sat, 10 Jan 2009) | 1 line
Minor documentation changes cross-referencing NullHandler to the documentation on configuring logging in a library.
........
r68481 | vinay.sajip | 2009-01-10 07:42:04 -0600 (Sat, 10 Jan 2009) | 1 line
Corrected an incorrect self-reference.
........
r68493 | benjamin.peterson | 2009-01-10 11:18:55 -0600 (Sat, 10 Jan 2009) | 1 line
rewrite verbose conditionals
........
r68495 | benjamin.peterson | 2009-01-10 11:36:44 -0600 (Sat, 10 Jan 2009) | 1 line
tp_iter only exists with Py_TPFLAGS_HAVE_ITER #4901
........
r68501 | vinay.sajip | 2009-01-10 13:22:57 -0600 (Sat, 10 Jan 2009) | 1 line
Corrected minor typo and added .currentmodule directives to fix missing cross-references.
........
r68512 | benjamin.peterson | 2009-01-10 16:42:10 -0600 (Sat, 10 Jan 2009) | 1 line
make tests fail if they can't be imported
........
r68514 | benjamin.peterson | 2009-01-10 17:41:59 -0600 (Sat, 10 Jan 2009) | 1 line
move seealso to a more appropiate place
........
r68515 | benjamin.peterson | 2009-01-10 17:49:08 -0600 (Sat, 10 Jan 2009) | 1 line
macos 9 isn't supported
........
r68534 | gregory.p.smith | 2009-01-11 11:53:33 -0600 (Sun, 11 Jan 2009) | 2 lines
correct email address
........
r68535 | gregory.p.smith | 2009-01-11 11:57:54 -0600 (Sun, 11 Jan 2009) | 9 lines
Update the documentation for binascii and zlib crc32/adler32 functions
to better describe the signed vs unsigned return value behavior on
different platforms and versions of python. Mention the workaround to
make them all return the same thing by using & 0xffffffff.
Fixes issue4903.
Also needs to be merged into release26-maint, release30-maint, & py3k.
........
r68536 | benjamin.peterson | 2009-01-11 13:48:15 -0600 (Sun, 11 Jan 2009) | 1 line
add email addresses
........
r68552 | vinay.sajip | 2009-01-12 14:36:18 -0600 (Mon, 12 Jan 2009) | 1 line
Minor changes/corrections in markup.
........
r68563 | benjamin.peterson | 2009-01-12 19:49:10 -0600 (Mon, 12 Jan 2009) | 1 line
small logic correction
........
r68570 | raymond.hettinger | 2009-01-13 03:08:32 -0600 (Tue, 13 Jan 2009) | 5 lines
Issue 4922: Incorrect comments for MutableSet.add() and MutableSet.discard().
Needs to be backported to 2.6 and forward ported to 3.0 and 3.1.
........
r68571 | armin.ronacher | 2009-01-13 05:52:23 -0600 (Tue, 13 Jan 2009) | 3 lines
ast.literal_eval can properly evaluate complex numbers now. This fixes issue4907.
........
r68572 | andrew.kuchling | 2009-01-13 07:40:54 -0600 (Tue, 13 Jan 2009) | 1 line
Note that first coord. is left alone
........
r68575 | thomas.heller | 2009-01-13 11:32:28 -0600 (Tue, 13 Jan 2009) | 1 line
Fix refcount leak in error cases. Bug found by coverity.
........
r68582 | georg.brandl | 2009-01-13 16:14:01 -0600 (Tue, 13 Jan 2009) | 2 lines
Use assertRaises.
........
r68596 | amaury.forgeotdarc | 2009-01-13 17:39:22 -0600 (Tue, 13 Jan 2009) | 3 lines
#1162154: inspect.getmembers() now skips attributes that raise AttributeError,
e.g. a __slots__ attribute which has not been set.
........
r68623 | vinay.sajip | 2009-01-15 16:48:13 -0600 (Thu, 15 Jan 2009) | 1 line
Made minor changes/corrections in markup. Added a couple of section headings.
........
r68624 | vinay.sajip | 2009-01-15 17:04:47 -0600 (Thu, 15 Jan 2009) | 1 line
Minor changes/corrections in markup.
........
r68628 | benjamin.peterson | 2009-01-15 20:55:24 -0600 (Thu, 15 Jan 2009) | 1 line
compare with == not is #4946
........
svn+ssh://pythondev@svn.python.org/python/trunk
........
r64114 | gregory.p.smith | 2008-06-11 09:41:16 +0200 (mer., 11 juin 2008) | 6 lines
Merge in release25-maint r60793:
Added checks for integer overflows, contributed by Google. Some are
only available if asserts are left in the code, in cases where they
can't be triggered from Python code.
........
svn+ssh://pythondev@svn.python.org/python/trunk
........
r63562 | martin.v.loewis | 2008-05-23 17:06:50 +0200 (Fri, 23 May 2008) | 2 lines
Patch #1722225: Support QNX 6.
........
r63570 | trent.nelson | 2008-05-23 22:33:14 +0200 (Fri, 23 May 2008) | 1 line
Introduce a user macro named $(externalsDir), which should point to the root directory of where all the external sources should live. Developers can change this value if their external sources live elsewhere. The default of '..\..' matches the current status quo.
........
r63728 | gregory.p.smith | 2008-05-26 23:16:34 +0200 (Mon, 26 May 2008) | 4 lines
Fix issue2589: there was a potential integer overflow leading to
memory corruption on esoteric platforms and incorrect behavior on
normal platforms.
........
r63734 | gregory.p.smith | 2008-05-27 00:07:28 +0200 (Tue, 27 May 2008) | 3 lines
Fix issue2588: Do not execute str[size-1] = '\0' when a 0 size is
passed in. (The assert won't prevent this in non-debug builds).
........
r63784 | raymond.hettinger | 2008-05-29 10:38:23 +0200 (Thu, 29 May 2008) | 1 line
Fix two typos.
........
r63788 | facundo.batista | 2008-05-29 18:39:26 +0200 (Thu, 29 May 2008) | 6 lines
Fixed the semantic of timeout for socket.create_connection and
all the upper level libraries that use it, including urllib2.
Added and fixed some tests, and changed docs correspondingly.
Thanks to John J Lee for the patch and the pusing, :)
........
r63802 | mark.dickinson | 2008-05-30 04:46:53 +0200 (Fri, 30 May 2008) | 2 lines
Fix typo in testSum
........
r63817 | raymond.hettinger | 2008-05-30 20:20:50 +0200 (Fri, 30 May 2008) | 8 lines
* Mark intermedidate computes values (hi, lo, yr) as volatile.
* Expand comments.
* Swap variable names in the sum_exact code so that x and y
are consistently chosen as the larger and smaller magnitude
values respectively.
........
r63827 | raymond.hettinger | 2008-05-31 05:24:31 +0200 (Sat, 31 May 2008) | 1 line
Implement heapq in terms of less-than (to match list.sort()).
........
r63839 | gerhard.haering | 2008-05-31 23:33:27 +0200 (Sat, 31 May 2008) | 2 lines
Fixed rowcount for SELECT statements. They're -1 now (again), for better DB-API 2.0 compliance.
........
r63887 | gregory.p.smith | 2008-06-02 06:05:52 +0200 (Mon, 02 Jun 2008) | 4 lines
Fix issue 2782: be less strict about the format string type in strftime.
Accept unicode and anything else ParseTuple "s#" can deal with. This
matches the time.strftime behavior.
........
r63975 | neal.norwitz | 2008-06-06 06:47:01 +0200 (Fri, 06 Jun 2008) | 3 lines
Aldo Cortesi confirmed this is still needed for OpenBSD 4.2 and 4.3.
(I didn't regen configure, since I don't have a working autoconf.)
........
r63998 | raymond.hettinger | 2008-06-06 23:47:51 +0200 (Fri, 06 Jun 2008) | 1 line
Issue 3501: Make heapq support both __le__ and __lt__.
........
svn+ssh://pythondev@svn.python.org/python/trunk
........
r62380 | christian.heimes | 2008-04-19 01:13:07 +0200 (Sat, 19 Apr 2008) | 3 lines
I finally got the time to update and merge Mark's and my trunk-math branch. The patch is collaborated work of Mark Dickinson and me. It was mostly done a few months ago. The patch fixes a lot of loose ends and edge cases related to operations with NaN, INF, very small values and complex math.
The patch also adds acosh, asinh, atanh, log1p and copysign to all platforms. Finally it fixes differences between platforms like different results or exceptions for edge cases. Have fun :)
........
r62382 | christian.heimes | 2008-04-19 01:40:40 +0200 (Sat, 19 Apr 2008) | 2 lines
Added new files to Windows project files
More Windows related fixes are coming soon
........
r62383 | christian.heimes | 2008-04-19 01:49:11 +0200 (Sat, 19 Apr 2008) | 1 line
Stupid me. Py_RETURN_NAN should actually return something ...
........
svn+ssh://pythondev@svn.python.org/python/trunk
........
r60482 | raymond.hettinger | 2008-01-31 23:07:16 +0100 (Thu, 31 Jan 2008) | 1 line
Minor wordsmithing on docstring
........
r60483 | mark.dickinson | 2008-01-31 23:17:37 +0100 (Thu, 31 Jan 2008) | 5 lines
Issue #1678380. Fix a bug that identifies 0j and -0j when they appear
in the same code unit. The fix is essentially the same as the fix for a
previous bug identifying 0. and -0.
........
r60484 | christian.heimes | 2008-02-01 00:08:23 +0100 (Fri, 01 Feb 2008) | 1 line
Fixed bug #1983: Return from fork() is pid_t, not int
........
r60486 | jeffrey.yasskin | 2008-02-01 07:22:46 +0100 (Fri, 01 Feb 2008) | 4 lines
Move __builtins__.trunc() to math.trunc() per
http://mail.python.org/pipermail/python-dev/2008-January/076626.html and issue
1965.
........
r60487 | jeffrey.yasskin | 2008-02-01 08:05:46 +0100 (Fri, 01 Feb 2008) | 3 lines
Roll back r60248. It's useful to encourage users not to change Rational
instances.
........
r60488 | neal.norwitz | 2008-02-01 08:22:59 +0100 (Fri, 01 Feb 2008) | 1 line
Fix refleak
........
Victor Stinner
Paul Monson
Minmin Gong
Serhiy Storchaka
Victor Stinner
xdegaye
Benjamin Peterson
Eric Snow
Antoine Pitrou
n.d. parker
Joseph Shen
Christian Heimes
Ned Deily
Stefan Krah
Berker Peksag
Steve Dower
Larry Hastings
Christian Heimes
Terry Jan Reedy
Gregory P. Smith
Mark Dickinson
Georg Brandl
Jesus Cea
Mark Dickinson
Victor Stinner
Martin v. Löwis
Daniel Stutzbach
Alexander Belopolsky
Antoine Pitrou
Ronald Oussoren
Amaury Forgeot d'Arc