Mailman 3 February 2014 - pypy-issue

[issue1680] ast.NodeTransformer throws AttributeError in pypy only
by dmlockhart 07 Mar '14

07 Mar '14

New submission from dmlockhart <dmlockhart+pypy(a)gmail.com>: I have some code which performs transformations of python ast objects. The code executes as expected in CPython 2.7, but when run with the pypy interpreter, I receive the following traceback: Traceback (most recent call last): File "app_main.py", line 72, in run_toplevel File "ast_exception_bug.py", line 47, in <module> new_tree = RemoveSelf().visit( tree ) File "/Users/dmlockhart/venvs/pypy-2.0.2/lib-python/2.7/ast.py", line 241, in visit return visitor(node) File "/Users/dmlockhart/venvs/pypy-2.0.2/lib-python/2.7/ast.py", line 297, in generic_visit value = self.visit(value) File "/Users/dmlockhart/venvs/pypy-2.0.2/lib-python/2.7/ast.py", line 241, in visit return visitor(node) File "/Users/dmlockhart/venvs/pypy-2.0.2/lib-python/2.7/ast.py", line 297, in generic_visit value = self.visit(value) File "/Users/dmlockhart/venvs/pypy-2.0.2/lib-python/2.7/ast.py", line 241, in visit return visitor(node) File "/Users/dmlockhart/venvs/pypy-2.0.2/lib-python/2.7/ast.py", line 297, in generic_visit value = self.visit(value) File "/Users/dmlockhart/venvs/pypy-2.0.2/lib-python/2.7/ast.py", line 241, in visit return visitor(node) File "/Users/dmlockhart/venvs/pypy-2.0.2/lib-python/2.7/ast.py", line 308, in generic_visit delattr(node, field) AttributeError: cannot delete attribute Platform: OSX Mavericks Interpreters Tested: pypy-2.0.2 pypy-2.1 ---------- files: ast_exception_bug.py messages: 6505 nosy: dmlockhart, pypy-issue priority: bug status: unread title: ast.NodeTransformer throws AttributeError in pypy only ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1680> ________________________________________

1 1

[issue1630] Matrix dot product ~5x slower in PyPy
by Gareth Williams 05 Mar '14

05 Mar '14

New submission from Gareth Williams <gareth.gambit(a)gmail.com>: Running the following code in PyPy takes around 5 times longer than it does when using standard Numpy matrix dot products. import time try: import numpypy except ImportError: pass import numpy def get_matrix(): import random n = 502 x = numpy.zeros((n,n), dtype=numpy.float64) for i in range(n): for j in range(n): x[i][j] = random.random() return x def main(): x = get_matrix() y = get_matrix() a = time.time() z = numpy.dot(x, y) b = time.time() print '%.2f seconds' % (b-a) main() ---------- messages: 6299 nosy: garethgambit, pypy-issue priority: performance bug status: unread title: Matrix dot product ~5x slower in PyPy ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1630> ________________________________________

3 2

[issue1663] np.sum on arrays of bools doesn't count past 127 correctly
by Ian Ozsvald 19 Feb '14

19 Feb '14

New submission from Ian Ozsvald <ian(a)ianozsvald.com>: np.sum(np.array([<booleans>]) with over 127 True values results in an incorrect count. Using Python's sum instead produces the correct answer. np.sum gives the correct answer for smaller counts of True values. Two examples are below - the tipping point seems to be a list of True values that sum to 128 and multiples of 128. Version: PyPy Nightly for today: pypy-c-jit-68443-d5e489e07679-linux64 and numpy branch checked-out today. Python 2.7.3 (d5e489e07679, Dec 16 2013, 23:00:17) [PyPy 2.3.0-alpha0 with GCC 4.6.3] on linux2 Type "help", "copyright", "credits" or "license" for more information. And now for something completely different: ``"messy" is not a judgement, but just a fact of complicatedness'' >>>> import numpy as np >>>> np.version <module 'numpy.version' from '/home/ian/Downloads/pypy-c-jit-68443-d5e489e07679-linux64/env_numpy/site-packages/numpy/version.pyc'> >>>> np.sum(np.array([True,False])) 1 >>>> np.sum(np.array([True,False]*1000)) # Incorrect behaviour -24 >>>> sum(np.array([True,False])) 1 >>>> sum(np.array([True,False]*1000)) 1000 It feels as though the counter is a single byte (as suggested below using array.itemsize), it counts up to 127 and then counts down back to 0 (when it should be counting to 256 and beyond), then cycles again: >>>> np.sum(np.array([True,False])) 1 >>>> np.sum(np.array([True,False]*127)) 127 >>>> np.sum(np.array([True,False]*128)) -128 >>>> np.sum(np.array([True,False]*129)) -127 >>>> np.sum(np.array([True,False]*255)) -1 >>>> np.sum(np.array([True,False]*256)) 0 >>>> np.sum(np.array([True,False]*257)) 1 >>>> np.sum(np.array([True,False]*384)) -128 >>>> np.sum(np.array([True,False]*385)) -127 >>>> np.array([True,False]) array([ True, False], dtype=bool) >>>> arr=np.array([True,False]) >>>> arr.itemsize 1 ---------- nosy: +ianozsvald status: unread -> chatting ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1663> ________________________________________

3 7

[issue1694] Fatal RPython error: KeyError while doing pip install
by kostia.lopuhin 15 Feb '14

15 Feb '14

New submission from kostia.lopuhin <kostia.lopuhin(a)gmail.com>: I think this is the same issue reported here https://github.com/neithere/monk/issues/12, I can reproduce it while doing pip install of a large number of packages on pypy 2.2.1 both under OS X (like 5 times in a row) and Ubuntu 12.04 (just tried once). Is is a known issue? Or maybe it is fixed on default? I can come up with steps to reproduce it reliably, or do anything else I can to help. ---------- messages: 6549 nosy: kostia.lopuhin, pypy-issue priority: critical release: 2.2 status: unread title: Fatal RPython error: KeyError while doing pip install ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1694> ________________________________________

2 5

[issue1693] pickle.dumps results in different raw strings in PyPy vs CPython2.7
by Daniel Axtens 14 Feb '14

14 Feb '14

New submission from Daniel Axtens <daniel(a)axtens.net>: PyPy 2.2.1 is generating different pickles of instances of new-style classes to CPython 2.7. It's also not generating stable pickles: pickling an instance new-style class, restoring it and then pickling it again gives a different result. The objects are being restored fine, so it's a very minor issue - it's just triggering some tests in a project I'm working on that are comparing raw pickles. Test code is attached. Test setup: Mac OS 10.8 on a 2GHz MacBook5,1 with 8GB ram uname -a: Darwin Daniels-MacBook.local 12.5.0 Darwin Kernel Version 12.5.0: Sun Sep 29 13:33:47 PDT 2013; root:xnu-2050.48.12~1/RELEASE_X86_64 x86_64 pypy --version: Python 2.7.3 (87aa9de10f9c, Nov 24 2013, 20:57:21) [PyPy 2.2.1 with GCC 4.2.1 Compatible Apple LLVM 5.0 (clang-500.2.79)] python --version: Python 2.7.2 python testpypypickle.py: "ccopy_reg\n_reconstructor\np0\n(c__main__\nA\np1\nc__builtin__\nobject\np2\nNtp3\nRp4\n(dp5\nS'a'\np6\nI1\nsS'c'\np7\nI3\nsS'a_a'\np8\nI2\ns b." "ccopy_reg\n_reconstructor\np0\n(c__main__\nA\np1\nc__builtin__\nobject\np2\nNtp3\nRp4\n(dp5\nS'a'\np6\nI1\nsS'c'\np7\nI3\nsS'a_a'\np8\nI2\ns b." '\x80\x02c__main__\nA\nq\x00)\x81q\x01}q\x02(U\x01aq\x03K\x01U\x01cq\x04K\x03U\x03a_aq\x05K\x02ub.' '\x80\x02c__main__\nA\nq\x00)\x81q\x01}q\x02(U\x01aq\x03K\x01U\x01cq\x04K\x03U\x03a_aq\x05K\x02ub.' pypy testpypypickle.py: "ccopy_reg\n_reconstructor\np0\n(c__main__\nA\np1\nc__builtin__\nobject\np2\nNtp3\nRp4\n(dp5\nS'c'\np6\nI3\nsS'a_a'\np7\nI2\nsS'a'\np8\nI1\ns b." "ccopy_reg\n_reconstructor\np0\n(c__main__\nA\np1\nc__builtin__\nobject\np2\nNtp3\nRp4\n(dp5\nS'a_a'\np6\nI2\nsS'c'\np7\nI3\nsS'a'\np8\nI1\ns b." '\x80\x02c__main__\nA\nq\x00)\x81q\x01}q\x02(U\x01cq\x03K\x03U\x03a_aq\x04K\x02U\x01aq\x05K\x01ub.' '\x80\x02c__main__\nA\nq\x00)\x81q\x01}q\x02(U\x03a_aq\x03K\x02U\x01cq\x04K\x03U\x01aq\x05K\x01ub.' P.S. Does the pickle format actually makes any promises about the ordering of things in raw pickles? If not, this is not actually a bug and I'll fix the tests instead. ---------- files: testpypypickle.py messages: 6543 nosy: dja, pypy-issue priority: bug release: 2.2 status: unread title: pickle.dumps results in different raw strings in PyPy vs CPython2.7 ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1693> ________________________________________

2 2

[issue1687] WeakValueDictionary and __del__
by Alexey 13 Feb '14

13 Feb '14

New submission from Alexey <hellman1908(a)gmail.com>: When storing objects with defined __del__ method (even with only 'pass') in WeakValueDictionary, sometimes they disappear suddenly from dict. Attached script raises assertion error when bug is catched. Run it several times and wait 5-10 seconds (well, it's rather stable for me in 2-5 seconds). You can remove __del__ method and bug disappears. Checked on pypy2.2.0 and 2.2.1 (both are buggy). CPython2.7 and 3.3 checked also - no bugs. Also it seems the but is not in WeakValueDictionary, because I implemented it's variant with a couple of needed methods using just dict and weakref.ref's and the bug still applies. ---------- files: bug.py messages: 6523 nosy: hellman, pypy-issue priority: bug status: unread title: WeakValueDictionary and __del__ ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1687> ________________________________________

2 3

[issue1692] Expose PyByteArray_AS_STRING in cpyext
by Alex Gaynor 12 Feb '14

12 Feb '14

New submission from Alex Gaynor <alex.gaynor(a)gmail.com>: This is needed for a C-extension I came across. ---------- messages: 6539 nosy: agaynor, pypy-issue priority: feature status: unread title: Expose PyByteArray_AS_STRING in cpyext ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1692> ________________________________________

2 2

[issue1685] broken numpypy installation on osx
by mikefc 12 Feb '14

12 Feb '14

New submission from mikefc <coolbutuseless(a)gmail.com>: Running latest pypy nightly on OSX Mavericks. Installed numpypy via: pip install git+https://bitbucket.org/pypy/numpy.git (Install log attached.) Seg fault on import: ============================================================== (numpypy)mikes-iMac:python mike$ python Python 2.7.3 (001e11396293, Jan 31 2014, 23:01:03) [PyPy 2.3.0-alpha0 with GCC 4.2.1 Compatible Clang Compiler] on darwin Type "help", "copyright", "credits" or "license" for more information. And now for something completely different: ``it's nonsense all the way down, and also all the way up'' >>>> import numpy Segmentation fault: 11 ============================================================== Compiler is Apple's default LLVM/Clang that comes with LLVM/Clang. Note: I haven't tried running numpypy on OSX since it was split off into its own repo. ---------- files: numpypy_installlog.txt messages: 6516 nosy: mikefc, pypy-issue priority: bug status: unread title: broken numpypy installation on osx ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1685> ________________________________________

3 2

[issue1691] numpy nansum returns 0 when nan in list
by Ryan 12 Feb '14

12 Feb '14

New submission from Ryan <welch.ryan(a)gmail.com>: Seems like nansum is returning 0 whenever np.nan is present in a list: python nansum_bug.py Answer should be 3.0, got: 3.0 pypy nansum_bug.py Answer should be 3.0, got: 0.0 I'm using pypy 2.2.1 with a fresh pull from the numpy bitbucket repo: >>>> np.version.full_version '1.8.0.dev-20902db' >>>> np.version.git_revision '20902db0f22b6681fda533d2cb336c7af081ed22' ---------- files: nansum_bug.py messages: 6537 nosy: pypy-issue, welchr priority: bug release: 2.2 status: unread title: numpy nansum returns 0 when nan in list ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1691> ________________________________________

2 1

[issue1682] changes in cpyext broke several lib-python tests on mac64
by Brian Kearns 11 Feb '14

11 Feb '14

New submission from Brian Kearns <bdkearns(a)gmail.com>: I suspect 49c649807414 and related changesets. * thread #1: tid = 0xc93b3, 0x00000001007272eb pypy-c1`pypy_g_init_types + 11, queue = 'com.apple.main-thread, stop reason = EXC_BAD_ACCESS (code=1, address=0x2623461) frame #0: 0x00000001007272eb pypy-c1`pypy_g_init_types + 11 pypy-c1`pypy_g_init_types + 11: -> 0x1007272eb: testb $16, 169(%rdi) 0x1007272f2: jne 0x100727320 ; pypy_g_init_types + 64 0x1007272f4: callq 0x10077ad10 ; pypy_g_type_realize 0x1007272f9: leaq 32528048(%rip), %rax ; pypy_g_ExcData ---------- messages: 6509 nosy: bdk, pypy-issue priority: critical status: unread title: changes in cpyext broke several lib-python tests on mac64 ________________________________________ PyPy bug tracker <tracker(a)bugs.pypy.org> <https://bugs.pypy.org/issue1682> ________________________________________

2 2