Mailman 3 December 2013 - pypy-issue

[issue892] Pypy is slower on longs than it should
by Leonardo Santagada 17 Jan '14

17 Jan '14

New submission from Leonardo Santagada <santagada(a)gmail.com>: In Python Brasil this year we found that this code runs relatively the same on pypy and cpython on linux (and is much worse on osx): fibs = {0: 0, 1: 1} def fib(n): if n in fibs: return fibs[n] if n % 2 == 0: fibs[n] = ((2 * fib((n / 2) - 1)) + fib(n / 2)) * fib(n / 2) else: fibs[n] = (fib((n - 1) / 2) ** 2) + (fib((n + 1) / 2) ** 2) return fibs[n] fib(10000000) original by Henrique Pereira (implementation of a dijkstra algo) the link is here: https://gist.github.com/1256985 Alex Gaynor said that the reason is probably because pypy longs are known to be slower than cpythons. ---------- messages: 3239 nosy: pypy-issue, santagada priority: bug status: unread title: Pypy is slower on longs than it should ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue892> ________________________________________

8 11

[issue1429] CSV module is slow
by mrjbq7 06 Jan '14

06 Jan '14

New submission from mrjbq7 <mrjbq7(a)gmail.com>: Using the 6.7MB airport.csv file from ourairports.com (http://www.ourairports.com/data/airports.csv) saved to "/tmp/airports.csv", and the attached python script (csvtest.py). On Python 2.7.2: reading took 0.137 seconds writing took 0.126 seconds complete took 0.264 seconds On PyPy 1.9.0: reading took 0.627 seconds writing took 0.491 seconds complete took 1.118 seconds ---------- messages: 5485 nosy: mrjbq7, pypy-issue priority: performance bug status: unread title: CSV module is slow ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1429> ________________________________________

3 3

[issue1665] Module re behavior different from CPython in Match.group
by danblack 27 Dec '13

27 Dec '13

New submission from danblack <daniel.subs(a)internode.on.net>: $ pypy -V Python 2.7.3 (352c78d2e80f4a812ae1d8cdbe8c01a7f2e6fbc0, Aug 19 2013, 10:43:46) [PyPy 2.1.0 with GCC 4.8.1 20130603 (Red Hat 4.8.1-1)] (pypyenv) $ ipython Python 2.7.3 (352c78d2e80f4a812ae1d8cdbe8c01a7f2e6fbc0, Aug 19 2013, 10:43:46) In [1]: import re In [2]: m = re.search('(?P<alpha>abc)','abcdef') In [3]: m.group('CAT') --------------------------------------------------------------------------- KeyError Traceback (most recent call last) <ipython-input-3-506c488db4df> in <module>() ----> 1 m.group('CAT') KeyError: 'CAT' CPython (2.7.5) In [1]: import re In [2]: m = re.search('(?P<alpha>abc)','abcdef') In [3]: m.group('CAT') --------------------------------------------------------------------------- IndexError Traceback (most recent call last) <ipython-input-3-506c488db4df> in <module>() ----> 1 m.group('CAT') IndexError: no such group ---------- messages: 6446 nosy: danblack, pypy-issue priority: bug release: 2.1 status: unread title: Module re behavior different from CPython in Match.group ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1665> ________________________________________

2 1

[issue1664] numpypy bit shift is broken
by mattip 24 Dec '13

24 Dec '13

New submission from mattip <matti.picus(a)gmail.com>: Python 2.7.3 (4702ccfceeb9, Dec 23 2013, 21:59:39) [PyPy 2.3.0-alpha0 with MSC v.1500 32 bit] on win32 Type "help", "copyright", "credits" or "license" for more information. >>>> import _numpypy.multiarray as numpy >>>> a=numpy.array(123,dtype='uint64') >>>> a array([123]) >>>> a >> 1 SEGFAULT ---------- messages: 6444 nosy: mattip, pypy-issue priority: bug release: ??? status: unread title: numpypy bit shift is broken ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1664> ________________________________________

2 1

[issue1640] Posix 2001 ftime() is deprecated
by Tobias Oberstein 24 Dec '13

24 Dec '13

New submission from Tobias Oberstein <tobias.oberstein(a)gmail.com>: The ftime() function which is declared in <sys/timeb.h> which is only included in rpython/rtyper/module/ll_time.py is deprecated: http://linux.die.net/man/3/ftime The inclusion of the respective header produces lots of warnings on FreeBSD/Clang and clutters the log. The Linux man page suggests: "This function is obsolete. Don't use it. If the time in seconds suffices, time(2) can be used; gettimeofday(2) gives microseconds; clock_gettime(2) gives nanoseconds but is not as widely available." clock_gettime is available on both Linux and FreeBSD and provides best-resolution time: http://linux.die.net/man/2/clock_gettime http://www.freebsd.org/cgi/man.cgi?query=clock_gettime http://stackoverflow.com/questions/275004/c-timer-function-to-provide-time-… However, it's not there on OSX: http://stackoverflow.com/questions/5167269/clock-gettime-alternative-in-mac… So we could adjust ll_time.py to use clock_gettime() instead of ftime() on Linux/FreeBSD and do a workaround for OSX. Not sure if that would be acceptable .. ---------- messages: 6350 nosy: oberstet, pypy-issue priority: bug status: unread title: Posix 2001 ftime() is deprecated ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1640> ________________________________________

3 5

[issue1086] cpyext: calling PyIter_Next() on a "dictionaryiterator" object crashes when the dict was modified
by Stefan Behnel 22 Dec '13

22 Dec '13

New submission from Stefan Behnel <stefan_ml(a)behnel.de>: The iterdict.pyx test crashes in the PyPy adapted Cython. https://github.com/scoder/cython/blob/master/tests/run/iterdict.pyx#L362 cdef int count = 0 for i in d.iterkeys(): d[i+1] = 5 count += 1 if count > 5: break # safety Readable parts of the GDB stack trace: #2 0x0000000000f2afbc in PyIter_Next () #3 0x00007ffff4c6ad72 in __Pyx_dict_iter_next (dict_or_iter=0x7ffff7e45350, orig_length=140737315342408, ppos=0x7fffffffb428, pkey=0x7fffffffb418, pvalue=0x0, pitem=0x0, is_dict=0) at iterdict.c:5714 #4 0x00007ffff4c651f6 in __pyx_pf_8iterdict_40iterdict_change_size (__pyx_self=0x0, __pyx_v_d=0x7ffff7e45320) at iterdict.c:4007 #5 0x00007ffff4c65079 in __pyx_pw_8iterdict_41iterdict_change_size (__pyx_self=0x0, __pyx_v_d=0x7ffff7e45320) at iterdict.c:3929 ---------- messages: 4063 nosy: pypy-issue, sbehnel priority: bug status: unread title: cpyext: calling PyIter_Next() on a "dictionaryiterator" object crashes when the dict was modified ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1086> ________________________________________

2 1

[issue1008] numpypy: commit the code for empty_like, zeros_like, ones_like
by Dmitrey 18 Dec '13

18 Dec '13

New submission from Dmitrey <dmitrey15(a)ukr.net>: Code in the attached file. I have wrote it without looking into numpy sources - seems like the funcs are implemented somewhere in C or Cython code. ---------- messages: 3719 nosy: Dmitrey, pypy-issue priority: feature status: unread title: numpypy: commit the code for empty_like, zeros_like, ones_like ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1008> ________________________________________

6 11

[issue1581] strptime does not like string_
by Wouter van Heyst 18 Dec '13

18 Dec '13

New submission from Wouter van Heyst <wouter.pypy(a)richtlijn.be>: The attached code works with cpython but fails with pypy b9699dee6f10. Traceback (most recent call last): File "app_main.py", line 72, in run_toplevel File "/tmp/bug.py", line 8, in <module> datetime.strptime(a[0], '%Y-%m-%d') File "/home/larstiq/src/pypy/lib_pypy/datetime.py", line 1701, in strptime struct, micros = _strptime(date_string, format) File "/home/larstiq/src/pypy/lib-python/2.7/_strptime.py", line 322, in _strptime found = format_regex.match(data_string) TypeError: unsupported operand type for buffer: 'string_' ---------- messages: 6054 nosy: larstiq, pypy-issue priority: bug status: unread title: strptime does not like string_ ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1581> ________________________________________

2 1

[issue1466] for slices, __array__interface['data'][0] yields adress of base object's data
by kfj 18 Dec '13

18 Dec '13

New submission from kfj <_kfj(a)yahoo.com>: Here's what I observed: $ pypy Python 2.7.3 (aa7b56060ddd, Apr 22 2013, 22:03:12) [PyPy 2.0.0-beta2 with GCC 4.4.3] on linux2 Type "help", "copyright", "credits" or "license" for more information. And now for something completely different: ``PyPy 1.2 released - http://pypy.org/'' >>>> import numpypy as np >>>> a=np.zeros(10) >>>> a.__array_interface__['data'][0] -1220095968 >>>> a[2:].__array_interface__['data'][0] -1220095968 <<<<<<<< same value $ python Python 2.7.3 (default, Apr 10 2013, 05:46:21) [GCC 4.6.3] on linux2 Type "help", "copyright", "credits" or "license" for more information. >>> import numpy as np >>> a=np.zeros(10) >>> a.__array_interface__['data'][0] 162282152 >>> a[2:].__array_interface__['data'][0] 162282168 <<<<<<<<<<< different value ---------- nosy: +kfj status: unread -> chatting ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1466> ________________________________________

2 1

[issue918] implement numpy.matrix in micronumpy
by Peter 18 Dec '13

18 Dec '13

New submission from Peter <p.j.a.cock(a)googlemail.com>: Following should work on any version of bumpy, here using 1.5.1 on Python 2.6 on Mac OS X, >>> import numpy >>> m = numpy.matrix([[1,0],[0,1]]) >>> m matrix([[1, 0], [0, 1]]) Fails on pypy 1.6, $ pypy Python 2.7.1 (dcae7aed462b, Aug 17 2011, 09:46:15) [PyPy 1.6.0 with GCC 4.0.1] on darwin Type "help", "copyright", "credits" or "license" for more information. And now for something completely different: ``it's not a hack, it's a workaround'' >>>> import numpy >>>> m = numpy.matrix([[1,0],[0,1]]) Traceback (most recent call last): File "<console>", line 1, in <module> AttributeError: 'module' object has no attribute 'matrix' The matrix class is generally similar to a 2 dimensional array in numpy except that the * operator does matrix multiplication rather than element wise operation. This is particularly useful in teaching classes, even if many serious bumpy users prefer to avoid the matrix class and work with raw arrays and the dot method/function. ---------- messages: 3373 nosy: pypy-issue priority: bug release: 1.6 status: unread title: implement numpy.matrix in micronumpy ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue918> ________________________________________

3 3