Mailman 3 July 2014 - pypy-issue

Issue #1831: py3k: incorrect parsing of kwargname in the presence of kwonlyargs (pypy/pypy)
by Martin Matusiak 30 Jul '14

30 Jul '14

New issue 1831: py3k: incorrect parsing of kwargname in the presence of kwonlyargs https://bitbucket.org/pypy/pypy/issue/1831/py3k-incorrect-parsing-of-kwargn… Martin Matusiak: Without kwonly: ``` #!python (Pdb) snippet = 'def f(aaa, bbb, mmm=1, nnn=2, **kwargs): pass' (Pdb) self.compiler.compile(snippet, '<string>', 'single', 0) Signature(['aaa', 'bbb', 'mmm', 'nnn'], None, 'kwargs', []) ``` With kwonly: ``` #!python (Pdb) snippet = 'def f(aaa, bbb, *, mmm=1, nnn=2, **kwargs): pass' (Pdb) self.compiler.compile(snippet, '<string>', 'single', 0) Signature(['aaa', 'bbb'], None, 'mmm', ['mmm', 'nnn']) # expected: Signature(['aaa', 'bbb'], None, 'kwargs', ['mmm', 'nnn']) ``` (Here, Signature.__init__ was instrumented with: ``` #!python print self.__repr__() ``` I was trying to write a test for it: ``` #!python def test_kwonlyargs_signature(self): from pypy.interpreter.pycode import cpython_code_signature snippet = 'def f(aaa, bbb, *, mmm=1, nnn=2, **kwargs): pass' co = self.compiler.compile(snippet, '<string>', 'single', 0) sig = cpython_code_signature(co) # assert sig == .... ``` ...but I can't figure out why the code object returned isn't the one I'm trying to compile: ``` #!python (Pdb) co.co_argcount 0 ``` Any hints?

1 0

Issue #1830: PyPy doesn't implement sys.{get, set}dlopenflags (pypy/pypy)
by Romain Guillebert 30 Jul '14

30 Jul '14

New issue 1830: PyPy doesn't implement sys.{get,set}dlopenflags https://bitbucket.org/pypy/pypy/issue/1830/pypy-doesnt-implement-sys-get-set Romain Guillebert: PyPy doesn't have sys.{get,set}dlopenflags which seems to be needed if we want to embed cpython to make scipy work, because CPython needs to be loaded with RTLD_GLOBAL (I tried importing sqlite and it fails if I don't specify this) and cffi relies on the host's import system in API mode.

1 0

Issue #1829: Misleading PyPy/RPython translation error? (pypy/pypy)
by Edd Barrett 29 Jul '14

29 Jul '14

New issue 1829: Misleading PyPy/RPython translation error? https://bitbucket.org/pypy/pypy/issue/1829/misleading-pypy-rpython-translat… Edd Barrett: Hi, I got the following error when compiling teh pypy_bridge: ``` [translation:info] File "/home/vext01/research/pypy/rpython/rtyper/normalizec alls.py", line 96, in normalize_calltable_row_signature [translation:info] assert not shape_star, "XXX not implemented" [translation:ERROR] AssertionError: XXX not implemented ``` Some prodding: ``` (Pdb+ on Pdb++) ll 82 def normalize_calltable_row_signature(annotator, shape, row): 83 graphs = row.values() 84 assert graphs, "no graph??" 85 sig0 = graphs[0].signature 86 defaults0 = graphs[0].defaults 87 for graph in graphs[1:]: 88 if graph.signature != sig0: 89 break 90 if graph.defaults != defaults0: 91 break 92 else: 93 return False # nothing to do, all signatures already match 94 95 shape_cnt, shape_keys, shape_star = shape 96 -> assert not shape_star, "XXX not implemented" ... (Pdb+ on Pdb++) sig0 Signature(['self', 'w_dict'], None, None) (Pdb+ on Pdb++) graphs[7].signature Signature(['self', 'wpy_dict'], None, None) ``` Because the name of the argument differs, we end up breaking from the loop and fire the assertion. By renaming the `wpy_dict` argument to `w_dict` in the sources, translation succeeds. I'm not quite sure what to make of this, but if the argument names all need to match(?), then I would expect at least a better error message. Thanks

1 0

Issue #1828: py3k list don't have clear() method (pypy/pypy)
by Hyun Woo Park 29 Jul '14

29 Jul '14

New issue 1828: py3k list don't have clear() method https://bitbucket.org/pypy/pypy/issue/1828/py3k-list-dont-have-clear-method Hyun Woo Park: >>>> a = [1, 2, 3] >>>> a.clear() Traceback (most recent call last): File "<stdin>", line 1, in <module> AttributeError: 'list' object has no attribute 'clear'

1 0

Issue #1827: py3k: implement PyCode.dump() properly (pypy/pypy)
by Ronan Lamy 27 Jul '14

27 Jul '14

New issue 1827: py3k: implement PyCode.dump() properly https://bitbucket.org/pypy/pypy/issue/1827/py3k-implement-pycodedump-proper… Ronan Lamy: In py3k, `PyCode.dump()` is a hack that just calls the host (Python2) interpreter's `dis.dis`. This obviously cannot work correctly with Python3 opcodes, so we need to reimplement it in a Python3-aware way.

1 0

Issue #1826: The stackless implementation of PyPy is incomplete (pypy/pypy)
by Anselm Kruis 26 Jul '14

26 Jul '14

New issue 1826: The stackless implementation of PyPy is incomplete https://bitbucket.org/pypy/pypy/issue/1826/the-stackless-implementation-of-… Anselm Kruis: Hi, the stackless implementations of PyPy and Stackless-Python differ. Stackless added quite a few new functions and properties and fixed shortcomings during the past year. These changes didn't make it to PyPy. I discussed this situation at Europython 2014 with Armin. As a first step to improve the situation, I created a test suite for the Stackless API: https://bitbucket.org/stackless-dev/stackless-testsuite/overview This test-suite is a pure Python and runs unmodified on Python 2 or 3. It tests the API that every implementation of the module "stackless" should provide. Currently, the test-suite is not yet complete. It only checks, if expected functions/methods/attributes are present, but it does not yet check the functionality. To run the test suite simply execute ``` #!sh $ hg clone https://bitbucket.org/stackless-dev/stackless-testsuite $ cd stackless-testsuite $ pypy setup.py test ``` Any comments on the test suite, the stackless API and any help to improve the situation are highly welcome.

1 0

Issue #1825: function has no __qualname__ (pypy/pypy)
by Ian Foote 26 Jul '14

26 Jul '14

New issue 1825: function has no __qualname__ https://bitbucket.org/pypy/pypy/issue/1825/function-has-no-__qualname__ Ian Foote: Python 3.3 added the `__qualname__` attribute to functions and classes. (http://legacy.python.org/dev/peps/pep-3155/)

1 0

Issue #1824: test_csv failing with exception message comparison (pypy/pypy)
by Martin Matusiak 26 Jul '14

26 Jul '14

New issue 1824: test_csv failing with exception message comparison https://bitbucket.org/pypy/pypy/issue/1824/test_csv-failing-with-exception-… Martin Matusiak: See http://bugs.python.org/issue22076 The csv module has an exception message with bad grammar: - delimiter" must be an 1-character string "an" should be "a" Needs to be fixed in pypy 3.3 branch for now.

1 0

Issue #1823: curses module noutrefresh call fails when called on a pad object in pypy 2.3.1 (pypy/pypy)
by Tapani Kiiskinen 23 Jul '14

23 Jul '14

New issue 1823: curses module noutrefresh call fails when called on a pad object in pypy 2.3.1 https://bitbucket.org/pypy/pypy/issue/1823/curses-module-noutrefresh-call-f… Tapani Kiiskinen: I'm doing a noutrefresh call on a pad which i've instanciated with curses.newpad: pad.noutrefresh(0, 0, y, x, y + rows - 1, x + cols - 1) which fails with error: File "/home/tappi/programs/pypy-2.3.1-linux64/lib_pypy/_curses.py", line 840, in noutrefresh return _check_ERR(lib.wnoutrefresh(self._win, *args), TypeError: 'int(*)(WINDOW *)' expects 1 arguments, got 7 I checked the _curses.py file and lib._m_ispad(self.win) returns false which results in wnoutrefresh being called instead of pnoutrefresh, where pnoutrefresh would be the right one to call when using pads. The definition of _m_ispad depends on the existence of WINDOW_HAS_FLAGS definition: int _m_ispad(WINDOW *win) { #if defined WINDOW_HAS_FLAGS return (win->_flags & _ISPAD); #else return 0; #endif } If I hack this piece of code to not check for WINDOW_HAS_FLAGS my code works. Supposedly that variable should be defined somewhere but it's not being defined. Context: the same code works with cpython. I'm running an ubuntu 13.10 PyPy version: Python 2.7.6 (32f35069a16d, Jun 06 2014, 11:20:40) [PyPy 2.3.1 with GCC 4.6.3] on linux2

1 0

Issue #1822: JIT compiles tons of bridges (pypy/pypy)
by Remi Meier 22 Jul '14

22 Jul '14

New issue 1822: JIT compiles tons of bridges https://bitbucket.org/pypy/pypy/issue/1822/jit-compiles-tons-of-bridges Remi Meier: code from http://rosettacode.org/wiki/Perlin_noise#Python see attached file Nightly pypy (pypy-c-jit-72476-dbdcc057562b-linux64) is slower with jit than with "--jit off". Also, current STM builds don't have this problem. **Look at number of bridges:** JIT summary of PyPy-STM: ``` 0# [46643b20651b8] {jit-summary 0# Tracing: 29 0.078599 0# Backend: 27 0.019512 0# TOTAL: 0.456509 0# ops: 50153 0# recorded ops: 11775 0# calls: 885 0# guards: 6755 0# opt ops: 2654 0# opt guards: 637 0# forcings: 96 0# abort: trace too long: 2 0# abort: compiling: 0 0# abort: vable escape: 0 0# abort: bad loop: 0 0# abort: force quasi-immut: 0 0# nvirtuals: 4165 0# nvholes: 614 0# nvreused: 2888 0# Total # of loops: 10 0# Total # of bridges: 18 0# Freed # of loops: 0 0# Freed # of bridges: 0 0# [46643b20f1765] jit-summary} ``` JIT summary of pypy nightly: ``` [466341fc5ad18] {jit-summary Tracing: 921 4.593343 Backend: 920 0.672270 TOTAL: 9.669084 ops: 6136322 recorded ops: 1571845 calls: 56491 guards: 196164 opt ops: 397249 opt guards: 57832 forcings: 942 abort: trace too long: 1 abort: compiling: 0 abort: vable escape: 0 abort: bad loop: 0 abort: force quasi-immut: 0 nvirtuals: 1278724 nvholes: 591642 nvreused: 556151 Total # of loops: 19 Total # of bridges: 902 Freed # of loops: 0 Freed # of bridges: 0 [466341fcecc7b] jit-summary} ```

1 0