Mailman 3 July 2016 - pypy-issue

Issue #2358: pip install matplotlib fails on PyPy 5.3 on OS X (pypy/pypy)
by Josiah Walker 31 Jul '16

31 Jul '16

New issue 2358: pip install matplotlib fails on PyPy 5.3 on OS X https://bitbucket.org/pypy/pypy/issues/2358/pip-install-matplotlib-fails-on… Josiah Walker: Numpy is installed from the main numpy repository (the fix for pypy has already been added to it). Scipy has installed successfully. Installing matplotlib gives the following error: UPDATING build/lib.macosx-10.11-x86_64-2.7/matplotlib/_version.py set build/lib.macosx-10.11-x86_64-2.7/matplotlib/_version.py to '1.5.1' running build_ext building 'matplotlib.ft2font' extension Traceback (most recent call last): File "<module>", line 1, in <module> File "/private/var/folders/fx/1bjz7pr54fj20zcwdkmhbl480000gn/T/pip-build-e5DbXy/matplotlib/setup.py", line 277, in <module> **extra_args File "/usr/local/Cellar/pypy/5.3.0/libexec/lib-python/2.7/distutils/core.py", line 151, in setup dist.run_commands() File "/usr/local/Cellar/pypy/5.3.0/libexec/lib-python/2.7/distutils/dist.py", line 953, in run_commands self.run_command(cmd) File "/usr/local/Cellar/pypy/5.3.0/libexec/lib-python/2.7/distutils/dist.py", line 972, in run_command cmd_obj.run() File "/usr/local/Cellar/pypy/5.3.0/libexec/site-packages/setuptools/command/install.py", line 61, in run return orig.install.run(self) File "/usr/local/Cellar/pypy/5.3.0/libexec/lib-python/2.7/distutils/command/install.py", line 573, in run self.run_command('build') File "/usr/local/Cellar/pypy/5.3.0/libexec/lib-python/2.7/distutils/cmd.py", line 334, in run_command self.distribution.run_command(command) File "/usr/local/Cellar/pypy/5.3.0/libexec/lib-python/2.7/distutils/dist.py", line 972, in run_command cmd_obj.run() File "/usr/local/Cellar/pypy/5.3.0/libexec/lib-python/2.7/distutils/command/build.py", line 127, in run self.run_command(cmd_name) File "/usr/local/Cellar/pypy/5.3.0/libexec/lib-python/2.7/distutils/cmd.py", line 334, in run_command self.distribution.run_command(command) File "/usr/local/Cellar/pypy/5.3.0/libexec/lib-python/2.7/distutils/dist.py", line 972, in run_command cmd_obj.run() File "/usr/local/Cellar/pypy/5.3.0/libexec/site-packages/setuptools/command/build_ext.py", line 68, in run _build_ext.run(self) File "/usr/local/Cellar/pypy/5.3.0/libexec/lib-python/2.7/distutils/command/build_ext.py", line 347, in run self.build_extensions() File "/usr/local/Cellar/pypy/5.3.0/libexec/lib-python/2.7/distutils/command/build_ext.py", line 456, in build_extensions self.build_extension(ext) File "/usr/local/Cellar/pypy/5.3.0/libexec/site-packages/setuptools/command/build_ext.py", line 180, in build_extension _build_ext.build_extension(self, ext) File "/usr/local/Cellar/pypy/5.3.0/libexec/lib-python/2.7/distutils/command/build_ext.py", line 503, in build_extension include_dirs=ext.include_dirs, File "setupext.py", line 785, in __get__ result = obj._hooks[self._name]() + result File "setupext.py", line 812, in include_dirs_hook reload(numpy) File "/usr/local/Cellar/pypy/5.3.0/libexec/site-packages/numpy/__init__.py", line 117, in <module> raise RuntimeError('Reloading numpy is not supported') RuntimeError: Reloading numpy is not supported ---------------------------------------- Command "/usr/local/Cellar/pypy/5.3.0/bin/pypy -u -c "import setuptools, tokenize;__file__='/private/var/folders/fx/1bjz7pr54fj20zcwdkmhbl480000gn/T/pip-build-e5DbXy/matplotlib/setup.py';exec(compile(getattr(tokenize, 'open', open)(__file__).read().replace('\r\n', '\n'), __file__, 'exec'))" install --record /var/folders/fx/1bjz7pr54fj20zcwdkmhbl480000gn/T/pip-p_3Kp6-record/install-record.txt --single-version-externally-managed --compile" failed with error code 1 in /private/var/folders/fx/1bjz7pr54fj20zcwdkmhbl480000gn/T/pip-build-e5DbXy/matplotlib/

1 0

Issue #2357: pypy code ~50x slower than python (pypy/pypy)
by oscardssmith 31 Jul '16

31 Jul '16

New issue 2357: pypy code ~50x slower than python https://bitbucket.org/pypy/pypy/issues/2357/pypy-code-50x-slower-than-python oscardssmith: I was making a sudoku solver using numpy (code below). When I run it on python2, it takes 160ms. On PyPy, it is 7.8s. Any ideas why it is so much slower using pypy? from time import time from itertools import product import numpy as np class board: def __init__(self, state): self.board = np.asarray(state,np.int) self.size = len(state) self.vals = set(range(self.size+1)) def square_vals(self, coord): x, y = coord[0]//3*3, coord[1]//3*3 return self.board[x:x+3, y:y+3] def empty(self): return list(map(tuple, np.transpose(np.nonzero((self.board == 0))))) def valid_move(self, coord, value): return value not in self.board[coord[0]] and value not in self.board[:, coord[1]] and value not in self.square_vals(coord) def valid_moves(self, coord): return self.vals.difference(np.concatenate((self.board[coord[0]], self.board[:, coord[1]], self.square_vals(coord).flatten()))) def valid_board(self): for row, col in product(range(self.size), range(self.size)): value = self.board[(row, col)] self.board[(row, col)] = 0 if not (value not in self.board[row]] and value not in self.board[:, col] and value not in self.square_vals((row,col)): return False self.board[(row, col)] = value return True def __repr__(self): return "".join(''.join(str(row)[1:-1]) + '\n' for row in self.board) def solve(board): change_num = 0 empty, possible = board.empty(), dict() while i < len(empty): possible[empty[i]] = board.valid_moves(empty[i]) if len(possible[empty[i]]): board.board[empty[i]] = min(possible[empty[i]]) change_num += 1 else: while len(possible[empty[i - 1]]) == 1: i -= 1 board.board[empty[i]] = 0 i -= 1 change_num += 1 possible[empty[i]].remove(min(possible[empty[i]])) board.board[empty[i]] = min(possible[empty[i]]) i+=1 print(board) print(change_num) t1 = time() trial = board([ [3, 0, 0, 0, 0, 7, 0, 0, 9], [0, 0, 8, 0, 0, 0, 5, 0, 7], [2, 7, 4, 0, 8, 0, 0, 0, 0], [0, 0, 0, 0, 2, 0, 0, 0, 4], [0, 0, 7, 4, 0, 3, 1, 0, 0], [5, 0, 0, 0, 7, 0, 0, 0, 0], [0, 0, 0, 0, 6, 0, 4, 9, 8], [4, 0, 9, 0, 0, 0, 3, 0, 0], [1, 0, 0, 8, 0, 0, 0, 0, 0], ]) print(trial) solve(trial) print(1000*(time() - t1)) print(trial.valid_board())

1 0

Issue #2356: Error when translating py3k on OSX 10.11.5 (pypy/pypy)
by David Terry 30 Jul '16

30 Jul '16

New issue 2356: Error when translating py3k on OSX 10.11.5 https://bitbucket.org/pypy/pypy/issues/2356/error-when-translating-py3k-on-… David Terry: `cd pypy/goal && pypy ../../rpython/bin/rpython --opt=jit` on OSX 10.11.5 gives the following error: ``` [rtyper] specializing: 41500 / 138243 blocks (30%) ++++++++++++++++++++++++******************%%%%%%%[f67e] translation-task} [Timer] Timings: [Timer] annotate --- 432.4 s [Timer] rtype_lltype --- 191.6 s [Timer] ========================================== [Timer] Total: --- 623.9 s [translation:info] Error: File "/Users/david/Projects/pypy/rpython/translator/goal/translate.py", line 317, in main drv.proceed(goals) File "/Users/david/Projects/pypy/rpython/translator/driver.py", line 551, in proceed result = self._execute(goals, task_skip = self._maybe_skip()) File "/Users/david/Projects/pypy/rpython/translator/tool/taskengine.py", line 114, in _execute res = self._do(goal, taskcallable, *args, **kwds) File "/Users/david/Projects/pypy/rpython/translator/driver.py", line 278, in _do res = func() File "/Users/david/Projects/pypy/rpython/translator/driver.py", line 345, in task_rtype_lltype rtyper.specialize(dont_simplify_again=True) File "/Users/david/Projects/pypy/rpython/rtyper/rtyper.py", line 188, in specialize self.specialize_more_blocks() File "/Users/david/Projects/pypy/rpython/rtyper/rtyper.py", line 231, in specialize_more_blocks self.specialize_block(block) File "/Users/david/Projects/pypy/rpython/rtyper/rtyper.py", line 317, in specialize_block self.translate_hl_to_ll(hop, varmapping) File "/Users/david/Projects/pypy/rpython/rtyper/rtyper.py", line 443, in translate_hl_to_ll resultvar = hop.dispatch() File "/Users/david/Projects/pypy/rpython/rtyper/rtyper.py", line 660, in dispatch return translate_meth(self) File "<33181-codegen /Users/david/Projects/pypy/rpython/rtyper/rtyper.py:523>", line 5, in translate_op_mul return pair(r_arg1, r_arg2).rtype_mul(hop) File "/Users/david/Projects/pypy/rpython/rtyper/rfloat.py", line 87, in rtype_mul return _rtype_template(hop, 'mul') File "/Users/david/Projects/pypy/rpython/rtyper/rfloat.py", line 127, in _rtype_template vlist = hop.inputargs(Float, Float) File "/Users/david/Projects/pypy/rpython/rtyper/rtyper.py", line 691, in inputargs vars.append(self.inputarg(converted_to[i], i)) File "/Users/david/Projects/pypy/rpython/rtyper/rtyper.py", line 680, in inputarg return self.llops.convertvar(v, r_binding, converted_to) File "/Users/david/Projects/pypy/rpython/rtyper/rtyper.py", line 823, in convertvar (r_from, r_to)) [translation:ERROR] TyperError: don't know how to convert from <IntegerRepr INT> to <FloatRepr Float> .. (pypy.module.time.interp_time:284)gettimeofday .. block@151 with 2 exits(last_exception) .. v2513 = mul(v2512, (1e-06)) [translation] start debugger... > /Users/david/Projects/pypy/rpython/rtyper/rtyper.py(823)convertvar() -> (r_from, r_to)) ```

1 0

Issue #2355: py3k: kill ctypes.pythonapi (pypy/pypy)
by Philip Jenvey 30 Jul '16

30 Jul '16

New issue 2355: py3k: kill ctypes.pythonapi https://bitbucket.org/pypy/pypy/issues/2355/py3k-kill-ctypespythonapi Philip Jenvey: This was killed in pypy2 in f65f69804dfc. This needs to be merged into py3k (probably just the lib-python/2.7 bits into lib-python/3) and could likely require further adjustments to tests.

1 0

Issue #2354: Confusing type message: do_stringformat()-> assert s_str.is_constant() (pypy/pypy)
by Sarah Mount 25 Jul '16

25 Jul '16

New issue 2354: Confusing type message: do_stringformat()-> assert s_str.is_constant() https://bitbucket.org/pypy/pypy/issues/2354/confusing-type-message-do_strin… Sarah Mount: In the example interpreter below, there is some confusion between class attributes and object attributes. The "bug" here is that the error message that the RPython annotator gives is confusing. ``` #!python $ cat example2.py class Interpreter(object): help_message = 'usage: %s' def __init__(self, name): self.name = name def get_entry_point(self): def entry_point(argv): print self.help_message % self.name return 0 return entry_point class MyInterpreter(Interpreter): help_message = 'I WANT MY OWN HELP MESSAGE!!! usage: %s' def __init__(self): Interpreter.__init__(self, 'myinterp') # Fix type error by removing MyInterpreter.help_message # and writing: # # Interpreter.help_message = 'I WANT MY OWN HELP MESSAGE!!! usage: %s' # # in this constructor instead. def get_entry_point(self): def entry_point(argv): print self.help_message % self.name return 0 return entry_point def target(driver, args): return MyInterpreter().get_entry_point(), None $ ../pypy/rpython/bin/rpython -Ojit example2.py [translation:info] Error: File "/home/snim2/Desktop/working/pypy/rpython/translator/goal/translate.py", line 317, in main drv.proceed(goals) File "/home/snim2/Desktop/working/pypy/rpython/translator/driver.py", line 551, in proceed result = self._execute(goals, task_skip = self._maybe_skip()) File "/home/snim2/Desktop/working/pypy/rpython/translator/tool/taskengine.py", line 114, in _execute res = self._do(goal, taskcallable, *args, **kwds) File "/home/snim2/Desktop/working/pypy/rpython/translator/driver.py", line 278, in _do res = func() File "/home/snim2/Desktop/working/pypy/rpython/translator/driver.py", line 345, in task_rtype_lltype rtyper.specialize(dont_simplify_again=True) File "/home/snim2/Desktop/working/pypy/rpython/rtyper/rtyper.py", line 188, in specialize self.specialize_more_blocks() File "/home/snim2/Desktop/working/pypy/rpython/rtyper/rtyper.py", line 231, in specialize_more_blocks self.specialize_block(block) File "/home/snim2/Desktop/working/pypy/rpython/rtyper/rtyper.py", line 317, in specialize_block self.translate_hl_to_ll(hop, varmapping) File "/home/snim2/Desktop/working/pypy/rpython/rtyper/rtyper.py", line 443, in translate_hl_to_ll resultvar = hop.dispatch() File "/home/snim2/Desktop/working/pypy/rpython/rtyper/rtyper.py", line 660, in dispatch return translate_meth(self) File "<259-codegen /home/snim2/Desktop/working/pypy/rpython/rtyper/rtyper.py:523>", line 5, in translate_op_mod return pair(r_arg1, r_arg2).rtype_mod(hop) File "/home/snim2/Desktop/working/pypy/rpython/rtyper/rstr.py", line 554, in rtype_mod return r_str.ll.do_stringformat(hop, [(hop.args_v[1], hop.args_r[1])]) File "/home/snim2/Desktop/working/pypy/rpython/rtyper/lltypesystem/rstr.py", line 1151, in do_stringformat assert s_str.is_constant() [translation:ERROR] AssertionError [translation] start debugger... > /home/snim2/Desktop/working/pypy/rpython/rtyper/lltypesystem/rstr.py(1151)do_stringformat() -> assert s_str.is_constant() ```

1 0

Issue #2353: Confusing type error: AttributeError: RangeIteratorRepr instance has no attribute ll_getnextindex (pypy/pypy)
by Sarah Mount 25 Jul '16

25 Jul '16

New issue 2353: Confusing type error: AttributeError: RangeIteratorRepr instance has no attribute ll_getnextindex https://bitbucket.org/pypy/pypy/issues/2353/confusing-type-error-attributee… Sarah Mount: The call to **enumerate** in the example interpreter below produces the confusing error message also shown. ``` #!python $ cat example.py def entry_point(argv): for x, y in enumerate(xrange(10)): print x, y return 0 def target(driver, args): return entry_point, None $ .../pypy/rpython/bin/rpython -Ojit example.py ... [translation:info] Error: File "/home/snim2/Desktop/working/pypy/rpython/translator/goal/translate.py", line 317, in main drv.proceed(goals) File "/home/snim2/Desktop/working/pypy/rpython/translator/driver.py", line 551, in proceed result = self._execute(goals, task_skip = self._maybe_skip()) File "/home/snim2/Desktop/working/pypy/rpython/translator/tool/taskengine.py", line 114, in _execute res = self._do(goal, taskcallable, *args, **kwds) File "/home/snim2/Desktop/working/pypy/rpython/translator/driver.py", line 278, in _do res = func() File "/home/snim2/Desktop/working/pypy/rpython/translator/driver.py", line 345, in task_rtype_lltype rtyper.specialize(dont_simplify_again=True) File "/home/snim2/Desktop/working/pypy/rpython/rtyper/rtyper.py", line 188, in specialize self.specialize_more_blocks() File "/home/snim2/Desktop/working/pypy/rpython/rtyper/rtyper.py", line 231, in specialize_more_blocks self.specialize_block(block) File "/home/snim2/Desktop/working/pypy/rpython/rtyper/rtyper.py", line 316, in specialize_block hop.setup() # this is called from here to catch TyperErrors... File "/home/snim2/Desktop/working/pypy/rpython/rtyper/rtyper.py", line 639, in setup self.r_result = rtyper.getrepr(self.s_result) File "/home/snim2/Desktop/working/pypy/rpython/rtyper/rtyper.py", line 158, in getrepr result = s_obj.rtyper_makerepr(self) File "/home/snim2/Desktop/working/pypy/rpython/rtyper/rmodel.py", line 282, in rtyper_makerepr return EnumerateIteratorRepr(r_baseiter) File "/home/snim2/Desktop/working/pypy/rpython/rtyper/rrange.py", line 204, in __init__ self.ll_getnextindex = r_baseiter.ll_getnextindex File "/home/snim2/Desktop/working/pypy/rpython/rtyper/rmodel.py", line 107, in __getattr__ self.__class__.__name__, name)) [translation:ERROR] AttributeError: RangeIteratorRepr instance has no attribute ll_getnextindex [translation] start debugger... > /home/snim2/Desktop/working/pypy/rpython/rtyper/rmodel.py(107)__getattr__() -> self.__class__.__name__, name)) ```

1 0

Issue #2352: list.__contains__ compares objects in wrong order (pypy/pypy)
by Vasiliy Faronov 23 Jul '16

23 Jul '16

New issue 2352: list.__contains__ compares objects in wrong order https://bitbucket.org/pypy/pypy/issues/2352/list__contains__-compares-objec… Vasiliy Faronov: Consider the following program: ``` #!python class Foo(object): def __init__(self, bar): self.bar = bar def __repr__(self): return 'Foo(%r)' % (self.bar,) def __eq__(self, other): print('Foo.__eq__(%r, %r)' % (self, other)) return self.bar == other foo1 = Foo('A') foo2 = Foo('B') assert foo1 not in [foo2] ``` Under CPython 2.7.11 and 3.5.1, it prints: ``` #!python Foo.__eq__(Foo('A'), Foo('B')) Foo.__eq__(Foo('B'), 'A') ``` But under PyPy 5.3.1 (2.7), it prints: ``` #!python Foo.__eq__(Foo('B'), Foo('A')) Foo.__eq__(Foo('A'), 'B') ``` PyPy’s behavior contradicts the [Python 2 language reference](https://docs.python.org/2/reference/expressions.html#membership-…: > For the list and tuple types, `x in y` is true if and only if there exists an index *i* such that either `x is y[i]` or `x == y[i]` is true. According to the [Python 3 language reference](https://docs.python.org/3/reference/expressions.html#value-compa… (can’t find anything about this in Python 2), equality does not have to be symmetric (emphasis mine): > User-defined classes that customize their comparison behavior should follow some consistency rules, **if possible**: > > . . . > > * Comparison should be symmetric > > . . . > > Python does not enforce these consistency rules. In fact, the not-a-number values are an example for not following these rules. In my case, `x` and `y` are instances of different classes with completely different, magical `__eq__`, so this becomes important.

1 0

Issue #2351: Creating and overwriting tuples 4x slower on PyPy (pypy/pypy)
by Eric Astor 22 Jul '16

22 Jul '16

New issue 2351: Creating and overwriting tuples 4x slower on PyPy https://bitbucket.org/pypy/pypy/issues/2351/creating-and-overwriting-tuples… Eric Astor: Using PyPy 5.3.1 (on Windows): ``` #!bash pypy -m timeit -n 10000 -s "a = range(10000)" "tuple(a)" ``` takes (on my system) ~67 usec per loop, while ``` #!bash python27 -m timeit -n 10000 -s "a = range(10000)" "tuple(a)" ``` takes only ~25.5 usec per loop. Furthermore, assigning the tuple to a variable ``` #!bash pypy -m timeit -n 10000 -s "a = range(10000)" "x = tuple(a)" ``` now takes ~109 usec per loop, but does not impact CPython's performance at all. However, if we clear the variable after each assignment, performance recovers: ``` #!bash pypy -m timeit -n 10000 -s "a = range(10000)" "x = tuple(a); del x" ``` returns to taking ~67 usec per loop.

1 0

Issue #2350: Windows PyPy2 sys.stderr emitting \r\r\n line separators (pypy/pypy)
by John Vandenberg 21 Jul '16

21 Jul '16

New issue 2350: Windows PyPy2 sys.stderr emitting \r\r\n line separators https://bitbucket.org/pypy/pypy/issues/2350/windows-pypy2-sysstderr-emittin… John Vandenberg: If a process prints a line to stderr, pypy replaces the newline with `\r\r\n`. This happens with any of these approaches: ``` from __future__ import print_function print('foo', file=sys.stderr) sys.stderr.write('foo\n') sys.stderr.writelines(['foo\n']) ``` This doesnt occur for stdout, which emits a normal windows line ending of `\r\n`.

1 0

Issue #2349: [patch] fix build on DragonFly (pypy/pypy)
by David Naylor 21 Jul '16

21 Jul '16

New issue 2349: [patch] fix build on DragonFly https://bitbucket.org/pypy/pypy/issues/2349/patch-fix-build-on-dragonfly David Naylor: DragonFly needs the same fix for vmprof as FreeBSD. The attached patch fixes vmprof_config.h.

1 0