From issues-reply at bitbucket.org Wed Jan 3 01:02:53 2018 From: issues-reply at bitbucket.org (Sarge Borsch) Date: Wed, 03 Jan 2018 06:02:53 +0000 (UTC) Subject: [pypy-issue] Issue #2724: some (purely maths) code is ~6 times slower in PyPy3 than in CPython 3.6 (pypy/pypy) Message-ID: <20180103060253.29518.14966@celery-worker-109.ash1.bb-inf.net> New issue 2724: some (purely maths) code is ~6 times slower in PyPy3 than in CPython 3.6 https://bitbucket.org/pypy/pypy/issues/2724/some-purely-maths-code-is-6-times-slower Sarge Borsch: (AFAIK PyPy is supposed to be good at optimizing exactly these kinds of computations, so:) The code is below. The computational part executes in **~330 ms with CPython 3.6.4** (on average) but it takes **~1979ms under PyPy3** (detailed version below) ``` #!python def measure(fun): # used to measure run time of the interesting part from time import time t = time() result = fun() elapsed_ms = int(10 ** 3 * (time() - t)) print(str(elapsed_ms) + ' ms') return result def sum_digits(n): s = 0 while n: n, r = divmod(n, 10) s += r return s print(measure(lambda: sum_digits(11 ** 20000))) ``` PyPy3 detailed version: ``` $ pypy3 --version Python 3.5.3 (09f9160b643e3f02ccb8c843b2fbb4e5cbf54082, Dec 26 2017, 09:50:12) [PyPy 5.10.0 with GCC 4.2.1 Compatible Apple LLVM 9.0.0 (clang-900.0.39.2)] ``` OS is macOS 10.12.6. both PyPy3 and CPython3 were installed from homebrew, with default options. For most other 'number-crunching' programs PyPy3 gives ~2-6 times speedup on my machine (as expected), so it's probably not a broken build. I tried to further isolate the cause but to no avail so far. I will probably try again in some time. Fortunately the code is not too big already. Now, should I also test this on some previous PyPy3/2 versions? From issues-reply at bitbucket.org Thu Jan 4 04:17:15 2018 From: issues-reply at bitbucket.org (flok420) Date: Thu, 04 Jan 2018 09:17:15 +0000 (UTC) Subject: [pypy-issue] Issue #2725: pypy on raspberry pi 2 slower than regular python version (pypy/pypy) Message-ID: <20180104091715.31609.97709@celery-worker-107.ash1.bb-inf.net> New issue 2725: pypy on raspberry pi 2 slower than regular python version https://bitbucket.org/pypy/pypy/issues/2725/pypy-on-raspberry-pi-2-slower-than-regular flok420: When running feeks (https://github.com/flok99/feeks a chess program, commit 70528c31374fe2863c9aec24a2cfbd95e2f22888) on a raspberry pi 2, the pypy version is quite a bit slower than the regular python version: ``` #!shell folkert at vardan ~/QueenBee/feeks-87 $ ./main.py go movetime 5000 info depth 1 score cp 62 time 144 nodes 21 pv e2e4 info depth 2 score cp -33 time 533 nodes 69 pv e2e4 e7e6 info depth 3 score cp 117 time 3381 nodes 378 pv e2e3 e7e6 bestmove e2e3 ^Cfolkert at vardan ~/QueenBee/feeks-87 $ pypy ./main.py go movetime 5000 info depth 1 score cp 62 time 1270 nodes 21 pv e2e4 info depth 2 score cp -33 time 3590 nodes 69 pv e2e4 e7e6 bestmove e2e4 ``` What you see here: the regular version performs 378 nodes in 3381ms (111.8/s) while the pypy version performs 69 nodes in 3590ms (19.2/s). From issues-reply at bitbucket.org Thu Jan 4 12:18:04 2018 From: issues-reply at bitbucket.org (Julian Berman) Date: Thu, 04 Jan 2018 17:18:04 +0000 (UTC) Subject: [pypy-issue] Issue #2726: gdbm module is broken under gdbm 1.14 (pypy/pypy) Message-ID: <20180104171804.9047.46411@celery-worker-107.ash1.bb-inf.net> New issue 2726: gdbm module is broken under gdbm 1.14 https://bitbucket.org/pypy/pypy/issues/2726/gdbm-module-is-broken-under-gdbm-114 Julian Berman: Importing gdbm produces: ``` ~ ? pypy -c 'import gdbm' Julian at Macnetic Traceback (most recent call last): File "", line 1, in File "/usr/local/Cellar/pypy/5.10.0/libexec/lib_pypy/gdbm.py", line 1, in from _gdbm_cffi import ffi, lib # generated by _gdbm_build.py ImportError: unable to load extension module '/usr/local/Cellar/pypy/5.10.0/libexec/lib_pypy/_gdbm_cffi.pypy-41.so': dlopen(/usr/local/Cellar/pypy/5.10.0/libexec/lib_pypy/_gdbm_cffi.pypy-41.so, 6): Symbol not found: _gdbm_errno Referenced from: /usr/local/Cellar/pypy/5.10.0/libexec/lib_pypy/_gdbm_cffi.pypy-41.so Expected in: /usr/local/opt/gdbm/lib/libgdbm.4.dylib in /usr/local/Cellar/pypy/5.10.0/libexec/lib_pypy/_gdbm_cffi.pypy-41.so ? pypy --version Julian at Macnetic Python 2.7.13 (0e7ea4fe15e82d5124e805e2e4a37cae1a402d4b, Dec 26 2017, 08:20:00) [PyPy 5.10.0 with GCC 4.2.1 Compatible Apple LLVM 9.0.0 (clang-900.0.39.2)] ``` https://bugs.python.org/msg309392 looks like an upstream CPython equivalent issue. From issues-reply at bitbucket.org Sat Jan 6 08:40:15 2018 From: issues-reply at bitbucket.org (Dingyuan Wang) Date: Sat, 06 Jan 2018 13:40:15 +0000 (UTC) Subject: [pypy-issue] Issue #2727: PGO build remains profile code (pypy/pypy) Message-ID: <20180106134014.40445.44416@celery-worker-107.ash1.bb-inf.net> New issue 2727: PGO build remains profile code https://bitbucket.org/pypy/pypy/issues/2727/pgo-build-remains-profile-code Dingyuan Wang: When doing an PGO build, `pypy ../../rpython/bin/rpython -Ojit --profopt targetpypystandalone`, the resulting `pypy(3)` binary still contains code for generating profiles. It will write `/tmp/usession-whatever-0/testing_1/main.gcda` file depending on the build environment. From issues-reply at bitbucket.org Sat Jan 6 22:15:17 2018 From: issues-reply at bitbucket.org (biblicabeebli) Date: Sun, 07 Jan 2018 03:15:17 +0000 (UTC) Subject: [pypy-issue] Issue #2728: performance collapse with setattr in a large for-loop (pypy/pypy) Message-ID: <20180107031517.35968.69154@celery-worker-110.ash1.bb-inf.net> New issue 2728: performance collapse with setattr in a large for-loop https://bitbucket.org/pypy/pypy/issues/2728/performance-collapse-with-setattr-in-a biblicabeebli: Hi... I was interested in testing the performance difference between CPython's attribute assignment on class objects and dictionary key value assignment. I wrote a (bad) script to test this, ensured it was compatible and equivalent between Python 2 and 3, and ran it on CPython 2.7 and 3.6. Then, for goofs, I thought I would test it on pypy. pypy version info: pypy3.5-5.10 compiled using pyenv 1.2.0-8-g785738d running on mac os 10.11.6 clang version: Apple LLVM version 8.0.0 (clang-800.0.42.1) Target: x86_64-apple-darwin15.6.0 Thread model: posix Well, it either never finishes execution, or it takes more than 1 minute vs CPython's 0.057096 execution of that problematic for-loop on my computer. The eventual stack trace when after I hit ^C is pasted below, I therefore assume it is setattr that is causing the issue. Note: I tested both the class's __setattr__ method and the built in setattr() function, both have this problem. ^C Traceback (most recent call last): File "dict_class_test.py", line 104, in do(100000) # <-- takes forever??? File "dict_class_test.py", line 41, in do x.__setattr__(i_string, i) From issues-reply at bitbucket.org Sun Jan 7 06:43:32 2018 From: issues-reply at bitbucket.org (Ned Batchelder) Date: Sun, 07 Jan 2018 11:43:32 +0000 (UTC) Subject: [pypy-issue] Issue #2729: PyPy3-5.10 incorrectly decodes astral plane JSON characters (pypy/pypy) Message-ID: <20180107114332.32105.80529@celery-worker-108.ash1.bb-inf.net> New issue 2729: PyPy3-5.10 incorrectly decodes astral plane JSON characters https://bitbucket.org/pypy/pypy/issues/2729/pypy3-510-incorrectly-decodes-astral-plane Ned Batchelder: This is a regression from pypy3-5.9 ``` note = u"a\xa0\u266b\U0001d157" round_tripped = json.loads(json.dumps({"note": note}))['note'] print(ascii(note)) print(ascii(round_tripped)) ``` PyPy3-5.9 printed: ``` 'a\xa0\u266b\U0001d157' 'a\xa0\u266b\U0001d157' ``` (as does every CPython 3.x) PyPy3-5.10 prints: ``` 'a\xa0\u266b\U0001d157' 'a\xa0\u266b\ud834\udd57' ``` This is on a Mac. The PyPy2-5.10 binary didn't work at all for me, so I'm not sure what it produces. From issues-reply at bitbucket.org Sun Jan 7 16:21:44 2018 From: issues-reply at bitbucket.org (mattip) Date: Sun, 07 Jan 2018 21:21:44 +0000 (UTC) Subject: [pypy-issue] Issue #2730: non-ascii in filename confuses pypy3 (pypy/pypy) Message-ID: <20180107212144.29911.87502@celery-worker-109.ash1.bb-inf.net> New issue 2730: non-ascii in filename confuses pypy3 https://bitbucket.org/pypy/pypy/issues/2730/non-ascii-in-filename-confuses-pypy3 mattip: the following sequence (from a DOS terminal) shows how PyPy3 v5.10 handles non-ascii in filenames improperly: ``` chcp 869 mkdir abc??? echo text > abc???\test.txt python3.exe -c "import os; print(os.path.exists(r'abc???\test.txt'))" pypy3-c.exe -c "import os; print(os.path.exists(r'abc???\test.txt'))" ``` prints True for CPython, False for PyPY From issues-reply at bitbucket.org Mon Jan 8 15:32:59 2018 From: issues-reply at bitbucket.org (rndblnch) Date: Mon, 08 Jan 2018 20:32:59 +0000 (UTC) Subject: [pypy-issue] Issue #2731: pypy and pypy3 v5.10 binaries for Mac OS X are linked against third party dynamic libraries (pypy/pypy) Message-ID: <20180108203259.21404.92638@celery-worker-107.ash1.bb-inf.net> New issue 2731: pypy and pypy3 v5.10 binaries for Mac OS X are linked against third party dynamic libraries https://bitbucket.org/pypy/pypy/issues/2731/pypy-and-pypy3-v510-binaries-for-mac-os-x rndblnch: The binary distributions made available for Mac OS X for the 5.10 release are all linked to libraries located in /usr/local/opt/, which does not exist by default on the mac. As a result the binaries do not run. The machine that produced the builds may have a third party package manager installed that provides those libraries. for pypy2, the libraries linked here are: libssl, libcrypto, libffi and libncursesw. for pypy3, the libraries linked here are: libffi and libncursesw. for pypy3-2 (high sierra-compatible): libffi ### pypy2 ### [esperluet:/tmp] blanch% tar xjf ~/Downloads/pypy2-v5.10.0-osx64.tar.bz2 [esperluet:/tmp] blanch% cd pypy2-v5.10.0-osx64/bin/ [esperluet:/tmp/pypy2-v5.10.0-osx64/bin] blanch% otool -L libpypy-c.dylib libpypy-c.dylib: @rpath/libpypy-c.dylib (compatibility version 0.0.0, current version 0.0.0) /usr/lib/libutil.dylib (compatibility version 1.0.0, current version 1.0.0) /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1252.0.0) /usr/lib/libbz2.1.0.dylib (compatibility version 1.0.0, current version 1.0.5) /usr/lib/libz.1.dylib (compatibility version 1.0.0, current version 1.2.11) **/usr/local/opt/openssl/lib/libssl.1.0.0.dylib (compatibility version 1.0.0, current version 1.0.0)** **/usr/local/opt/openssl/lib/libcrypto.1.0.0.dylib (compatibility version 1.0.0, current version 1.0.0)** /usr/lib/libexpat.1.dylib (compatibility version 7.0.0, current version 8.0.0) **/usr/local/opt/libffi/lib/libffi.6.dylib (compatibility version 7.0.0, current version 7.4.0)** **/usr/local/opt/ncurses/lib/libncursesw.6.dylib (compatibility version 6.0.0, current version 6.0.0)** ### pypy3 ### [esperluet:/tmp] blanch% tar xjf ~/Downloads/pypy3-v5.10.0-osx64.tar.bz2 [esperluet:/tmp] blanch% otool -L pypy3-v5.10.0-osx64/bin/libpypy3-c.dylib pypy3-v5.10.0-osx64/bin/libpypy3-c.dylib: @rpath/libpypy3-c.dylib (compatibility version 0.0.0, current version 0.0.0) /usr/lib/libutil.dylib (compatibility version 1.0.0, current version 1.0.0) /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1252.0.0) /usr/lib/libbz2.1.0.dylib (compatibility version 1.0.0, current version 1.0.5) /usr/lib/libexpat.1.dylib (compatibility version 7.0.0, current version 8.0.0) /usr/lib/libz.1.dylib (compatibility version 1.0.0, current version 1.2.11) **/usr/local/opt/libffi/lib/libffi.6.dylib (compatibility version 7.0.0, current version 7.4.0)** **/usr/local/opt/ncurses/lib/libncursesw.6.dylib (compatibility version 6.0.0, current version 6.0.0)** ### pypy3-2 (high-sierra) ### [esperluet:/tmp] blanch% tar xjf ~/Downloads/pypy3-v5.10.0-osx64-2.tar.bz2 [esperluet:/tmp] blanch% otool -L pypy3-v5.10.0-osx64-2/bin/libpypy3-c.dylib pypy3-v5.10.0-osx64-2/bin/libpypy3-c.dylib: @rpath/libpypy3-c.dylib (compatibility version 0.0.0, current version 0.0.0) /usr/lib/libutil.dylib (compatibility version 1.0.0, current version 1.0.0) /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1238.0.0) /usr/lib/libexpat.1.dylib (compatibility version 7.0.0, current version 8.0.0) /usr/lib/libbz2.1.0.dylib (compatibility version 1.0.0, current version 1.0.5) /usr/lib/libz.1.dylib (compatibility version 1.0.0, current version 1.2.8) **/usr/local/opt/libffi/lib/libffi.6.dylib (compatibility version 7.0.0, current version 7.1.0)** /usr/lib/libncurses.5.4.dylib (compatibility version 5.4.0, current version 5.4.0) From issues-reply at bitbucket.org Sun Jan 14 09:10:09 2018 From: issues-reply at bitbucket.org (squeaky) Date: Sun, 14 Jan 2018 14:10:09 +0000 (UTC) Subject: [pypy-issue] Issue #2732: [PyPy 3.5 5.10.1] Hitting a sequence that generates bell crashes REPL (pypy/pypy) Message-ID: <20180114141009.24060.37764@celery-worker-110.ash1.bb-inf.net> New issue 2732: [PyPy 3.5 5.10.1] Hitting a sequence that generates bell crashes REPL https://bitbucket.org/pypy/pypy/issues/2732/pypy-35-5101-hitting-a-sequence-that squeaky: Might be related to: https://bitbucket.org/pypy/pypy/issues/2718/pypy35-traceback-when-hitting-backspace-in I am running gnome-terminal 3.24.2. When I hit for example 'Ctrl + Alt + -' in the REPL I get the following stacktrace: ``` Python 3.5.3 (3f6eaa010fce78cc7973bdc1dfdb95970f08fed2, Jan 14 2018, 12:54:44) [PyPy 5.10.1 with GCC 7.2.0] on linux Type "help", "copyright", "credits" or "license" for more information. >>>> Traceback (most recent call last): File "/home/squeaky/workspace/portable-pypy/pypy3.5-5.10.1-linux_x86_64-portable/lib_pypy/_pypy_interact.py", line 45, in interactive_console run_interactive(mainmodule) File "/home/squeaky/workspace/portable-pypy/pypy3.5-5.10.1-linux_x86_64-portable/lib_pypy/pyrepl/simple_interact.py", line 73, in run_multiline_interactive_console returns_unicode=True) File "/home/squeaky/workspace/portable-pypy/pypy3.5-5.10.1-linux_x86_64-portable/lib_pypy/pyrepl/readline.py", line 275, in multiline_input return reader.readline(returns_unicode=returns_unicode) File "/home/squeaky/workspace/portable-pypy/pypy3.5-5.10.1-linux_x86_64-portable/lib_pypy/pyrepl/reader.py", line 620, in readline self.handle1() File "/home/squeaky/workspace/portable-pypy/pypy3.5-5.10.1-linux_x86_64-portable/lib_pypy/pyrepl/reader.py", line 603, in handle1 self.do_cmd(cmd) File "/home/squeaky/workspace/portable-pypy/pypy3.5-5.10.1-linux_x86_64-portable/lib_pypy/pyrepl/reader.py", line 555, in do_cmd self.refresh() File "/home/squeaky/workspace/portable-pypy/pypy3.5-5.10.1-linux_x86_64-portable/lib_pypy/pyrepl/reader.py", line 536, in refresh screen = self.calc_screen() File "/home/squeaky/workspace/portable-pypy/pypy3.5-5.10.1-linux_x86_64-portable/lib_pypy/pyrepl/completing_reader.py", line 230, in calc_screen screen = super(CompletingReader, self).calc_screen() File "/home/squeaky/workspace/portable-pypy/pypy3.5-5.10.1-linux_x86_64-portable/lib_pypy/pyrepl/reader.py", line 299, in calc_screen l, l2 = disp_str(line) File "/home/squeaky/workspace/portable-pypy/pypy3.5-5.10.1-linux_x86_64-portable/lib_pypy/pyrepl/reader.py", line 86, in disp_str return join(s), b TypeError: sequence item 0: expected str, bytes found ``` From issues-reply at bitbucket.org Sun Jan 14 17:31:27 2018 From: issues-reply at bitbucket.org (mattip) Date: Sun, 14 Jan 2018 22:31:27 +0000 (UTC) Subject: [pypy-issue] Issue #2733: pypy3 slowdown for json_dumps benchmark (pypy/pypy) Message-ID: <20180114223126.29419.59777@celery-worker-109.ash1.bb-inf.net> New issue 2733: pypy3 slowdown for json_dumps benchmark https://bitbucket.org/pypy/pypy/issues/2733/pypy3-slowdown-for-json_dumps-benchmark mattip: Using the cpython performance set of benchmarks, pypy3-v5.10.1 is very slow as if the JIT is off. ``` #setup python -mpip install performance; python -mperformance venv create pypy2 -mpip install performance; pypy2 -mperformance venv create pypy3 -mpip install performance; pypy3 -mperformance venv create # runs the bnechmark python -mperformance run -b json_dumps venv venv/pypy3* python -mperformance run -b json_dumps venv venv/pypy2* python -mperformance run -b json_dumps venv venv/cpython* ``` Gives me ~5ms for pypy2, ~17 ms for cpython2, ~50 ms for pypy3 From issues-reply at bitbucket.org Mon Jan 15 12:02:58 2018 From: issues-reply at bitbucket.org (Yaroslav Fedevych) Date: Mon, 15 Jan 2018 17:02:58 +0000 (UTC) Subject: [pypy-issue] Issue #2734: Establish a build and release pipeline for macOS (pypy/pypy) Message-ID: <20180115170258.5877.50610@celery-worker-105.ash1.bb-inf.net> New issue 2734: Establish a build and release pipeline for macOS https://bitbucket.org/pypy/pypy/issues/2734/establish-a-build-and-release-pipeline-for Yaroslav Fedevych: This is a ticket that would track progress on making a build tooling which would enable reliable building of PyPy for macOS. One of the objectives is an ability to produce a standalone binary which can run even on a freshly-installed OS, without xcode-tools or Homebrew. Another is to eliminate the unfortunate cases when an officially-distributed binary is tied to the OS version so much that it fails to launch anywhere else. When building with Homebrew, the tooling must use as many Homebrew-provided libraries as possible. And of course, library lookups must remain cross-platform and not break any other builds. The tooling itself must be reproducible enough so it should be easy to replicate by anyone else, without being tied to specific binaries versions, and it shouldn't be sensitive to whatever else the user might have installed. From issues-reply at bitbucket.org Mon Jan 15 12:25:35 2018 From: issues-reply at bitbucket.org (William Minchin) Date: Mon, 15 Jan 2018 17:25:35 +0000 (UTC) Subject: [pypy-issue] Issue #2735: `os.readlink` missing in PyPy3 5.10.1 on Windows (pypy/pypy) Message-ID: <20180115172535.413.44501@celery-worker-105.ash1.bb-inf.net> New issue 2735: `os.readlink` missing in PyPy3 5.10.1 on Windows https://bitbucket.org/pypy/pypy/issues/2735/osreadlink-missing-in-pypy3-5101-on William Minchin: I'm trying to create a virtual environment using the `venv` module. ``` > C:\Programs\pypy3-v5.10.1-win32\pypy3.exe -m venv environment-name Error: module 'os' has no attribute 'readlink' ``` Comparing the "regular" Python documentation for Python 2 and 3, it looks like this functionality was added to the Windows version of Python 3. - Python 2 os.readlink() --> https://docs.python.org/2/library/os.html#os.readlink - Python 3 os.readlink() --> https://docs.python.org/3/library/os.html#os.readlink I'm on Windows 10 x64, running PyPy3 v5.10.1 through Powershell. From issues-reply at bitbucket.org Mon Jan 15 20:52:43 2018 From: issues-reply at bitbucket.org (Yeo Han, Jordan) Date: Tue, 16 Jan 2018 01:52:43 +0000 (UTC) Subject: [pypy-issue] Issue #2736: ImportError: cannot import name '_remove_dead_weakref' on Pypy Windows (pypy/pypy) Message-ID: <20180116015243.10745.40844@celery-worker-109.ash1.bb-inf.net> New issue 2736: ImportError: cannot import name '_remove_dead_weakref' on Pypy Windows https://bitbucket.org/pypy/pypy/issues/2736/importerror-cannot-import-name Yeo Han, Jordan: There is an error with the latest Pypy 5.10.0 for windows in which Pypy cannot import the library weakref. If Pypy is used alone without any file, to replicate the error, just type. ``` #!python import weakref ``` and the error ImportError: cannot import name '_remove_dead_weakref' will appear. This restricts using the Multiprocessing library as well as Multiprocessing uses the weakref library. Any way to solve this? From issues-reply at bitbucket.org Fri Jan 19 01:43:59 2018 From: issues-reply at bitbucket.org (Corentin Chary) Date: Fri, 19 Jan 2018 06:43:59 +0000 (UTC) Subject: [pypy-issue] Issue #2737: pyhash fails to build with pypy (pypy/pypy) Message-ID: <20180119064358.17610.40408@celery-worker-107.ash1.bb-inf.net> New issue 2737: pyhash fails to build with pypy https://bitbucket.org/pypy/pypy/issues/2737/pyhash-fails-to-build-with-pypy Corentin Chary: See https://github.com/flier/pyfasthash/issues/20#issuecomment-358857011 ``` $ pip install pyhash Collecting pyhash Using cached pyhash-0.8.2.tar.gz Building wheels for collected packages: pyhash Running setup.py bdist_wheel for pyhash ... error Complete output from command /home/cchary/dev/graphite/carbon/venv/bin/pypy -u -c "import setuptools, tokenize;__file__='/tmp/pip-build-MsOBgl/pyhash/setup.py';f=getattr(tokenize, 'open', open)(__file__);code=f.read().replace('\r\n', '\n');f.close();exec(compile(code, __file__, 'exec'))" bdist_wheel -d /tmp/tmpInZIz0pip-wheel- --python-tag pp258: running bdist_wheel running build running build_py creating build creating build/lib.linux-x86_64-2.7 copying pyhash.py -> build/lib.linux-x86_64-2.7 running build_ext building '_pyhash' extension creating build/temp.linux-x86_64-2.7 creating build/temp.linux-x86_64-2.7/src creating build/temp.linux-x86_64-2.7/src/fnv creating build/temp.linux-x86_64-2.7/src/smhasher creating build/temp.linux-x86_64-2.7/src/lookup3 creating build/temp.linux-x86_64-2.7/src/SuperFastHash cc -pthread -DNDEBUG -O2 -fPIC -DBOOST_PYTHON_STATIC_LIB -I/home/cchary/dev/graphite/carbon/venv/include -c src/Hash.cpp -o build/temp.linux-x86_64-2.7/src/Hash.o -DSUPPORT_INT128=1 -msse4.2 -maes In file included from src/Hash.cpp:1:0: src/Hash.h: In function ?PyObject* internal::convert(const T&) [with T = __int128 unsigned; PyObject = _object]?: src/Hash.h:72:36: error: invalid conversion from ?const unsigned char*? to ?unsigned char*? [-fpermissive] return ::_PyLong_FromByteArray((const unsigned char*) &value, sizeof(uint128_t), /*little_endian*/ 1, /*is_signed*/ 0); ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~ In file included from /home/cchary/dev/graphite/carbon/venv/include/Python.h:141:0, from /usr/include/boost/python/detail/wrap_python.hpp:151, from /usr/include/boost/python/detail/prefix.hpp:13, from /usr/include/boost/python/args.hpp:8, from /usr/include/boost/python.hpp:11, from src/Hash.h:3, from src/Hash.cpp:1: /usr/lib/pypy/include/pypy_decl.h:1047:24: note: initializing argument 1 of ?PyObject* _PyPyLong_FromByteArray(unsigned char*, long unsigned int, int, int)? PyAPI_FUNC(PyObject *) _PyLong_FromByteArray(unsigned char *arg0, Unsigned arg1, int arg2, int arg3); ^ In file included from src/Hash.cpp:1:0: src/Hash.h: In function ?T extract_hash_value(PyObject*) [with T = __int128 unsigned; PyObject = _object]?: src/Hash.h:152:5: error: ?_PyLong_AsByteArray? was not declared in this scope _PyLong_AsByteArray((PyLongObject *)obj, (unsigned char *) &value, sizeof(uint128_t), /*little_endian*/ 1, /*is_signed*/ 0); ^~~~~~~~~~~~~~~~~~~ src/Hash.h:152:5: note: suggested alternative: ?_PyLong_FromByteArray? _PyLong_AsByteArray((PyLongObject *)obj, (unsigned char *) &value, sizeof(uint128_t), /*little_endian*/ 1, /*is_signed*/ 0); ^~~~~~~~~~~~~~~~~~~ _PyLong_FromByteArray In file included from src/Hash.cpp:1:0: src/Hash.h: In static member function ?static boost::python::api::object Hasher::CallWithArgs(boost::python::tuple, boost::python::dict)?: src/Hash.h:237:50: error: there are no arguments to ?PyUnicode_AsUTF16String? that depend on a template parameter, so a declaration of ?PyUnicode_AsUTF16String? must be available [-fpermissive] py::object utf16 = py::object(py::handle<>(PyUnicode_AsUTF16String(arg.ptr()))); ^~~~~~~~~~~~~~~~~~~~~~~ src/Hash.h:237:50: note: (if you use ?-fpermissive?, G++ will accept your code, but allowing the use of an undeclared name is deprecated) src/Hash.h: In instantiation of ?static boost::python::api::object Hasher::CallWithArgs(boost::python::tuple, boost::python::dict) [with T = city_hash_t<__int128 unsigned>]?: src/CityHash.h:155:57: required from here src/Hash.h:237:73: error: ?PyUnicode_AsUTF16String? was not declared in this scope py::object utf16 = py::object(py::handle<>(PyUnicode_AsUTF16String(arg.ptr()))); ~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~ src/Hash.h:237:73: note: suggested alternative: ?PyUnicode_AsUTF8String? ``` Looks like the first error could be fixed in pypy (the missing `const`). The second comes from the fact that `Py_UNICODE_WIDE` is defined even if UTF16 function do not seem to be implemented From issues-reply at bitbucket.org Tue Jan 23 01:40:51 2018 From: issues-reply at bitbucket.org (Ruud van der Ham) Date: Tue, 23 Jan 2018 06:40:51 +0000 (UTC) Subject: [pypy-issue] Issue #2738: Windows 3.5 distribution misses \bin directory (pypy/pypy) Message-ID: <20180123064051.31340.50573@celery-worker-108.ash1.bb-inf.net> New issue 2738: Windows 3.5 distribution misses \bin directory https://bitbucket.org/pypy/pypy/issues/2738/windows-35-distribution-misses-bin Ruud van der Ham: I downloaded and installed the latest Windows 3.5 distribution, but that seems to miss the bin directory completely. Therefore, there?s no pip.exe and I am not able to install any packages. From issues-reply at bitbucket.org Tue Jan 23 03:54:36 2018 From: issues-reply at bitbucket.org (Ruud van der Ham) Date: Tue, 23 Jan 2018 08:54:36 +0000 (UTC) Subject: [pypy-issue] Issue #2739: Numpy install problem under PyPy 3.5 (Windows) (pypy/pypy) Message-ID: <20180123085435.16289.37922@celery-worker-110.ash1.bb-inf.net> New issue 2739: Numpy install problem under PyPy 3.5 (Windows) https://bitbucket.org/pypy/pypy/issues/2739/numpy-install-problem-under-pypy-35 Ruud van der Ham: I am trying to install numpy under PyPy 3.5 on Windows with **pypy3 -m pip install cython numpy** Everything seems to install but the setup of numpy ends with an error message.![PyPy numpy problem.png](https://bitbucket.org/repo/R7AbB/images/2412465912-PyPy%20numpy%20problem.png) FYI: I am using Windows 10. From issues-reply at bitbucket.org Mon Jan 29 18:03:06 2018 From: issues-reply at bitbucket.org (Christoph Reiter) Date: Mon, 29 Jan 2018 23:03:06 +0000 (UTC) Subject: [pypy-issue] Issue #2740: hash() does not raise TypeError when tp_hash=0 and tp_richcompare!=0 (pypy/pypy) Message-ID: <20180129230306.9353.6409@celery-worker-105.ash1.bb-inf.net> New issue 2740: hash() does not raise TypeError when tp_hash=0 and tp_richcompare!=0 https://bitbucket.org/pypy/pypy/issues/2740/hash-does-not-raise-typeerror-when-tp_hash Christoph Reiter: Using pypy 5.10 + cpyext The Python docs state: "When this field is not set, two possibilities exist: if the tp_compare and tp_richcompare fields are both NULL, a default hash value based on the object?s address is returned; otherwise, a TypeError is raised." https://docs.python.org/2/c-api/typeobj.html#c.PyTypeObject.tp_hash I have a class which has tp_hash=0 and tp_richcompare!=0, and doing hash(instance) returns a hash and does not raise TypeError. From issues-reply at bitbucket.org Mon Jan 29 18:22:22 2018 From: issues-reply at bitbucket.org (Christoph Reiter) Date: Mon, 29 Jan 2018 23:22:22 +0000 (UTC) Subject: [pypy-issue] Issue #2741: PyArg_ParseTuple: "et" allows passing in None instead of raising (pypy/pypy) Message-ID: <20180129232222.488.19112@celery-worker-107.ash1.bb-inf.net> New issue 2741: PyArg_ParseTuple: "et" allows passing in None instead of raising https://bitbucket.org/pypy/pypy/issues/2741/pyarg_parsetuple-et-allows-passing-in-none Christoph Reiter: Using PyPy 5.10: Having a function which does PyArg_ParseTuple("et", "utf-8", &utf8) Expected: Calling func(None) should raise Actual: No error raised From issues-reply at bitbucket.org Tue Jan 30 04:10:23 2018 From: issues-reply at bitbucket.org (Christoph Reiter) Date: Tue, 30 Jan 2018 09:10:23 +0000 (UTC) Subject: [pypy-issue] Issue #2742: cpyext: class subclassable even if Py_TPFLAGS_BASETYPE not set (pypy/pypy) Message-ID: <20180130091022.23004.92815@celery-worker-106.ash1.bb-inf.net> New issue 2742: cpyext: class subclassable even if Py_TPFLAGS_BASETYPE not set https://bitbucket.org/pypy/pypy/issues/2742/cpyext-class-subclassable-even-if Christoph Reiter: I have a base class which has Py_TPFLAGS_BASETYPE set, then on subclass which has it cleared. In PyPy the resulting type can be subclassed, in CPython TypeError is raised. From issues-reply at bitbucket.org Tue Jan 30 04:16:17 2018 From: issues-reply at bitbucket.org (Christoph Reiter) Date: Tue, 30 Jan 2018 09:16:17 +0000 (UTC) Subject: [pypy-issue] Issue #2743: pypy3/cpyext: PyState_FindModule missing (pypy/pypy) Message-ID: <20180130091616.39773.22601@celery-worker-108.ash1.bb-inf.net> New issue 2743: pypy3/cpyext: PyState_FindModule missing https://bitbucket.org/pypy/pypy/issues/2743/pypy3-cpyext-pystate_findmodule-missing Christoph Reiter: ``` implicit declaration of function ?PyState_FindModule?; did you mean ?PyImport_AddModule?? [-Wimplicit-function-declaration] PyObject *cairo_module = PyState_FindModule(&cairomoduledef); ``` From issues-reply at bitbucket.org Tue Jan 30 17:13:17 2018 From: issues-reply at bitbucket.org (mattip) Date: Tue, 30 Jan 2018 22:13:17 +0000 (UTC) Subject: [pypy-issue] Issue #2744: missing win32 functions in _multiprocessing (pypy/pypy) Message-ID: <20180130221317.16291.97211@celery-worker-109.ash1.bb-inf.net> New issue 2744: missing win32 functions in _multiprocessing https://bitbucket.org/pypy/pypy/issues/2744/missing-win32-functions-in mattip: send, recv, closesocket are missing from the module simple test: ``` from multiprocessing import Pool def f(x): return x*x if __name__ == '__main__': with Pool(5) as p: print(p.map(f, [1, 2, 3])) ```