Mailman 3 January 2017 - Python-Dev

enable-framework Vs. Naught
by André Lemos 10 Jan '17

10 Jan '17

Hi, I have a C++ module that I am compiling to use inside of my Python installation under Mac OS. If I compile & link it against a Framework enabled Python installation, it works fine, but if I compile & link it against a *non* enabled Framework installation that we use for distribution, I simply get a non inspiring: Fatal Python error: PyThreadState_Get: no current thread I am using python-config to get my flags on both the examples, but I simply cannot get it to run (although it compiles fine) on a *non* enabled Framework installation. Thoughts/Help? -- André Lemos

3 2

Re: [Python-Dev] Adding bytes.frombuffer() constructor to PEP 467
by Alexander Belopolsky 10 Jan '17

10 Jan '17

On Wed, Oct 12, 2016 at 12:08 AM, INADA Naoki <songofacandy(a)gmail.com> wrote: > > Now I'm sure about bytes.frombuffer() is worth enough. I would like to revive this thread (taking a liberty to shorten the subject line.) The issue of how the bytes(x) constructor should behave when given objects of various types have come up recently in issue 29159 (Regression in bytes constructor). [1] The regression was introduced in issue 27704 (bytes(x) is slow when x is bytearray) which attempted to speed-up creating bytes and bytearray from byte-like objects. I think the core problem is that the bytes(x) constructor tries to be the Jack of All Trades. Here is how it is documented in the docstring: | Construct an immutable array of bytes from: | - an iterable yielding integers in range(256) | - a text string encoded using the specified encoding | - any object implementing the buffer API. | - an integer | On the other hand, the reference manual while not having this description in the bytes section, has a similar list in the bytearray section. [3] """ The optional source parameter can be used to initialize the array in a few different ways: * If it is a string, you must also give the encoding (and optionally, errors) parameters; bytearray() then converts the string to bytes using str.encode(). * If it is an integer, the array will have that size and will be initialized with null bytes. * If it is an object conforming to the buffer interface, a read-only buffer of the object will be used to initialize the bytes array. * If it is an iterable, it must be an iterable of integers in the range 0 <= x < 256, which are used as the initial contents of the array. Without an argument, an array of size 0 is created. """ Note that the integer case is listed before buffer interface. Neither document mentions the possibility that the source type has a __bytes__ method. This ambiguity between integer-like and buffer-like sources causes a problem in the case when a 3rd party type is both integer-like and buffer-like. This is what happens with numpy arrays: >>> bytes(numpy.array([2], 'i1')) b'\x00\x00' >>> bytes(numpy.array([2, 2], 'i1')) b'\x02\x02' For better or worse, single-element numpy arrays have a working __index__ methods >>> numpy.array([2], 'i1').__index__() 2 and are interpreted as integers by the bytes(X) constructor. I propose the following: 1. For 3.6, restore and document 3.5 behavior. Recommend that 3rd party types that are both integer-like and buffer-like implement their own __bytes__ method to resolve the bytes(x) ambiguity. 2. For 3.7, I would like to see a drastically simplified bytes(x): 2.1. Accept only objects with a __bytes__ method or a sequence of ints in range(256). 2.2. Expand __bytes__ definition to accept optional encoding and errors parameters. Implement str.__bytes__(self, [encoding[, errors]]). 2.3. Implement new specialized bytes.fromsize and bytes.frombuffer constructors as per PEP 467 and Inada Naoki proposals. 2.4. Implement memoryview.__bytes__ method so that bytes(memoryview(x)) works ad before. 2.5. Implement a fast bytearray.__bytes__ method. 3. Consider promoting __bytes__ to a tp_bytes type slot. [1]: http://bugs.python.org/issue29159 [2]: http://bugs.python.org/issue27704 [3]: https://docs.python.org/3/library/functions.html#bytearray

7 12

Imports with underscores
by Steve Holden 10 Jan '17

10 Jan '17

One of my developers recently submitted a pull request incuding a number of lines like import os as _os When I asked him why he suggested a) this would improve encapsulation, and b) the practice was supported in the stdlib. Further investigation reveals that some modules (e.g. argparse, crypt, difflib, random) do use this technique, but it is far from universal. So I thought it would be useful to get input from current devs about the value of this practice, since to me it seems somewhat anti-pythonic. What advantages does it confer? regards Steve Holden

10 11

ctypes, memory mapped files and context manager
by Hans-Peter Jansen 08 Jan '17

08 Jan '17

Hi, first of all, sorry for being such a pest, but all former attempts to solve this issue on other fora has been grinding to a halt. In short: I try to combine a context manager with ctypes structures on memory mapped files in order to manage huge binary files. (An approach, that performs great, while easy to use and keeping the resource usage low). FWIW, the code is targeted for Linux environments running Python3. The smallest script demonstrating the issue (thanks to Peter Otten): import ctypes import mmap from contextlib import contextmanager class T(ctypes.Structure): _fields = [("foo", ctypes.c_uint32)] @contextmanager def map_struct(m, n): m.resize(n * mmap.PAGESIZE) yield T.from_buffer(m) SIZE = mmap.PAGESIZE * 2 f = open("tmp.dat", "w+b") f.write(b"\0" * SIZE) f.seek(0) m = mmap.mmap(f.fileno(), mmap.PAGESIZE) with map_struct(m, 1) as a: a.foo = 1 with map_struct(m, 2) as b: b.foo = 2 resulting in: $ python3 mmap_test.py Traceback (most recent call last): File "mmap_test.py", line 23, in <module> with map_struct(m, 2) as b: File "/usr/lib64/python3.4/contextlib.py", line 59, in __enter__ return next(self.gen) File "mmap_test.py", line 12, in map_struct m.resize(n * mmap.PAGESIZE) BufferError: mmap can't resize with extant buffers exported. Python2 does not crash, but that's a different story. What happens here is: the context manager variable "a" keeps a reference to a memory mapped area alive, that results in a unresizable and not properly closable mmap. Right now, this rather ugly and error prone workaround must be used, that renders the purpose of the context manager ad absurdum: with map_struct(m, 1) as a: a.foo = 1 del a with map_struct(m, 2) as b: b.foo = 2 del b In order to get this working properly, the ctypes mapping needs a method to free the mapping actively. E.g.: @contextmanager def map_struct(m, n): m.resize(n * mmap.PAGESIZE) yield T.from_buffer(m) T.unmap_buffer(m) Other attempts with weakref and the like do not work due to the nature of the ctypes types. My own investigations in the _ctypes module were unsuccessful so far. Hopefully, somebody in the audience cares enough for this module in order to get this fixed up (or probably I'm missing something obvious..). Any ideas are very much appreciated. Pete

4 8

Summary of Python tracker Issues
by Python tracker 06 Jan '17

06 Jan '17

ACTIVITY SUMMARY (2016-12-30 - 2017-01-06) Python tracker at http://bugs.python.org/ To view or respond to any of the issues listed below, click on the issue. Do NOT respond to this message. Issues counts and deltas: open 5685 (+22) closed 35253 (+45) total 40938 (+67) Open issues with patches: 2460 Issues opened (44) ================== #24725: test_socket testFDPassEmpty fails on OS X 10.11+ with "Cannot http://bugs.python.org/issue24725 reopened by ned.deily #29013: zipfile: inconsistent doc for ZIP64 file size http://bugs.python.org/issue29013 reopened by serhiy.storchaka #29116: Make str and bytes error messages on concatenation conform wit http://bugs.python.org/issue29116 opened by Jim Fasarakis-Hilliard #29117: dir(<bound method>) should include dunder attributes of the un http://bugs.python.org/issue29117 opened by Antony.Lee #29120: Move hash randomisation initialisation out of Python/random.c http://bugs.python.org/issue29120 opened by ncoghlan #29121: sqlite3 Controlling Transactions documentation not updated http://bugs.python.org/issue29121 opened by palaviv #29124: Freeze fails to compile on windows http://bugs.python.org/issue29124 opened by Liran Ringel #29125: Shell injection via TIX_LIBRARY when using tkinter.tix http://bugs.python.org/issue29125 opened by symphorien #29126: Fix comments about _PyThreadState_Current http://bugs.python.org/issue29126 opened by Jim Nasby #29127: Incorrect reference names in asyncio.subprocess documentation http://bugs.python.org/issue29127 opened by Eric Ahn #29132: shlex.shlex with punctuation_chars and posix doesn't handle pu http://bugs.python.org/issue29132 opened by evan_ #29133: Minor inaccuracy in shlex.shlex punctuation_chars example http://bugs.python.org/issue29133 opened by evan_ #29136: Add OP_NO_TLSv1_3 http://bugs.python.org/issue29136 opened by christian.heimes #29137: Fix fpectl-induced ABI breakage http://bugs.python.org/issue29137 opened by njs #29139: operator.concat/iconcat could only work if left operand is a s http://bugs.python.org/issue29139 opened by xiang.zhang #29140: time_hash() reads the wrong bytes to get microseconds http://bugs.python.org/issue29140 opened by haypo #29141: error in 2to3 http://bugs.python.org/issue29141 opened by Sandeep Srinivasa #29142: urllib: no_proxy variable values with leading dot not properly http://bugs.python.org/issue29142 opened by tloetzer #29143: Logger should ignore propagate property for disabled handlers. http://bugs.python.org/issue29143 opened by Oleg Serov #29144: Implicit namespace packages in Python 3.6 http://bugs.python.org/issue29144 opened by frenzy #29145: failing overflow checks in replace_* http://bugs.python.org/issue29145 opened by matejcik #29147: registry value to be cleared when python is uninstalled http://bugs.python.org/issue29147 opened by jha.amit6666 #29151: test_asyncore.TestAPI_UseIPv4Select / test_asyncore.TestAPI_Us http://bugs.python.org/issue29151 opened by patrila #29152: unittest subTest does not call addFailure http://bugs.python.org/issue29152 opened by kristall #29153: Test test.test_asynchat.TestAsynchat / test.test_asynchat.Test http://bugs.python.org/issue29153 opened by patrila #29154: 5 failures in test_doctest: ModuleNotFoundError: No module nam http://bugs.python.org/issue29154 opened by Gerrit.Holl #29155: test.test_spwd.TestSpwdNonRoot failure with FileNotFoundError http://bugs.python.org/issue29155 opened by Gerrit.Holl #29157: random.c: Prefer getrandom() over getentropy(), handle ENOSYS http://bugs.python.org/issue29157 opened by haypo #29158: Possible glitch in the interaction of a thread and a multiproc http://bugs.python.org/issue29158 opened by luke_16 #29164: make test always fail at 218/405 ( AssertionError: ',' not fou http://bugs.python.org/issue29164 opened by OO O #29165: Use forward compatible macro in example code for creating new http://bugs.python.org/issue29165 opened by inada.naoki #29167: Race condition in enum.py:_decompose() http://bugs.python.org/issue29167 opened by simon.percivall #29169: update zlib to 1.2.10 http://bugs.python.org/issue29169 opened by doko #29171: blake2: Remove unused function to avoid undefined references http://bugs.python.org/issue29171 opened by ericvw #29172: blake2: Use lowest-common denominator signature of #pragma pac http://bugs.python.org/issue29172 opened by ericvw #29174: 'NoneType' object is not callable in subprocess.py http://bugs.python.org/issue29174 opened by ita1024 #29175: Tutorial links to file object methods are broken. http://bugs.python.org/issue29175 opened by jonroach(a)icloud.com #29176: /tmp does not exist on Android and is used by curses.window.pu http://bugs.python.org/issue29176 opened by xdegaye #29177: skip tests using socketserver.UnixStreamServer when bind() rai http://bugs.python.org/issue29177 opened by xdegaye #29178: Adding bytes.frombuffer(byteslike) constructor http://bugs.python.org/issue29178 opened by inada.naoki #29179: Py_UNUSED is not documented http://bugs.python.org/issue29179 opened by encukou #29180: skip tests that raise PermissionError in test_os (non-root use http://bugs.python.org/issue29180 opened by xdegaye #29181: skip tests that raise PermissionError in test_tarfile (non-roo http://bugs.python.org/issue29181 opened by xdegaye #29182: Remove the warning in urllib docs that it doesn't do certifica http://bugs.python.org/issue29182 opened by orsenthil Most recent 15 issues with no replies (15) ========================================== #29182: Remove the warning in urllib docs that it doesn't do certifica http://bugs.python.org/issue29182 #29181: skip tests that raise PermissionError in test_tarfile (non-roo http://bugs.python.org/issue29181 #29180: skip tests that raise PermissionError in test_os (non-root use http://bugs.python.org/issue29180 #29179: Py_UNUSED is not documented http://bugs.python.org/issue29179 #29171: blake2: Remove unused function to avoid undefined references http://bugs.python.org/issue29171 #29165: Use forward compatible macro in example code for creating new http://bugs.python.org/issue29165 #29153: Test test.test_asynchat.TestAsynchat / test.test_asynchat.Test http://bugs.python.org/issue29153 #29151: test_asyncore.TestAPI_UseIPv4Select / test_asyncore.TestAPI_Us http://bugs.python.org/issue29151 #29144: Implicit namespace packages in Python 3.6 http://bugs.python.org/issue29144 #29127: Incorrect reference names in asyncio.subprocess documentation http://bugs.python.org/issue29127 #29126: Fix comments about _PyThreadState_Current http://bugs.python.org/issue29126 #29121: sqlite3 Controlling Transactions documentation not updated http://bugs.python.org/issue29121 #29117: dir(<bound method>) should include dunder attributes of the un http://bugs.python.org/issue29117 #29113: modulefinder no longer finds all required modules for Python i http://bugs.python.org/issue29113 #29098: document minimum sqlite version http://bugs.python.org/issue29098 Most recent 15 issues waiting for review (15) ============================================= #29182: Remove the warning in urllib docs that it doesn't do certifica http://bugs.python.org/issue29182 #29178: Adding bytes.frombuffer(byteslike) constructor http://bugs.python.org/issue29178 #29176: /tmp does not exist on Android and is used by curses.window.pu http://bugs.python.org/issue29176 #29172: blake2: Use lowest-common denominator signature of #pragma pac http://bugs.python.org/issue29172 #29171: blake2: Remove unused function to avoid undefined references http://bugs.python.org/issue29171 #29169: update zlib to 1.2.10 http://bugs.python.org/issue29169 #29157: random.c: Prefer getrandom() over getentropy(), handle ENOSYS http://bugs.python.org/issue29157 #29153: Test test.test_asynchat.TestAsynchat / test.test_asynchat.Test http://bugs.python.org/issue29153 #29145: failing overflow checks in replace_* http://bugs.python.org/issue29145 #29142: urllib: no_proxy variable values with leading dot not properly http://bugs.python.org/issue29142 #29140: time_hash() reads the wrong bytes to get microseconds http://bugs.python.org/issue29140 #29133: Minor inaccuracy in shlex.shlex punctuation_chars example http://bugs.python.org/issue29133 #29132: shlex.shlex with punctuation_chars and posix doesn't handle pu http://bugs.python.org/issue29132 #29126: Fix comments about _PyThreadState_Current http://bugs.python.org/issue29126 #29125: Shell injection via TIX_LIBRARY when using tkinter.tix http://bugs.python.org/issue29125 Top 10 most discussed issues (10) ================================= #28180: sys.getfilesystemencoding() should default to utf-8 http://bugs.python.org/issue28180 23 msgs #29145: failing overflow checks in replace_* http://bugs.python.org/issue29145 12 msgs #29057: Compiler failure on Mac OS X - sys/random.h http://bugs.python.org/issue29057 11 msgs #29102: Add an id field to PyInterpreterState. http://bugs.python.org/issue29102 9 msgs #29157: random.c: Prefer getrandom() over getentropy(), handle ENOSYS http://bugs.python.org/issue29157 9 msgs #25750: tp_descr_get(self, obj, type) is called without owning a refer http://bugs.python.org/issue25750 6 msgs #29120: Move hash randomisation initialisation out of Python/random.c http://bugs.python.org/issue29120 6 msgs #27200: make doctest in CPython has failures http://bugs.python.org/issue27200 5 msgs #28518: execute("begin immediate") throwing OperationalError http://bugs.python.org/issue28518 5 msgs #28961: unittest.mock._Call ignores `name` parameter http://bugs.python.org/issue28961 5 msgs Issues closed (45) ================== #15812: inspect.getframeinfo() cannot show first line http://bugs.python.org/issue15812 closed by berker.peksag #24932: Use proper command line parsing in _testembed http://bugs.python.org/issue24932 closed by steve.dower #26267: UUID docs should say how to get "standard form" http://bugs.python.org/issue26267 closed by berker.peksag #26851: android compilation and link flags http://bugs.python.org/issue26851 closed by xdegaye #28443: Logger methods never use kwargs http://bugs.python.org/issue28443 closed by vinay.sajip #28524: Set default argument of logging.disable() to logging.CRITICAL http://bugs.python.org/issue28524 closed by python-dev #28592: Installation freezes on C Runtime Install http://bugs.python.org/issue28592 closed by steve.dower #28768: Warning: implicit declaration of function '_setmode' http://bugs.python.org/issue28768 closed by haypo #28781: On Installation of 3.5 Python get error message http://bugs.python.org/issue28781 closed by steve.dower #28848: Add CopyingMock to mock.py http://bugs.python.org/issue28848 closed by berker.peksag #28985: sqlite3 authorizer codes constants not up to date http://bugs.python.org/issue28985 closed by berker.peksag #29012: __bases__ is a tuple (possibly empty or a singleton) http://bugs.python.org/issue29012 closed by berker.peksag #29014: Python 3.5.2 installer doesn't register IDLE .py extensions on http://bugs.python.org/issue29014 closed by steve.dower #29024: Add Kivy entry to Graphic User Interface FAQ http://bugs.python.org/issue29024 closed by berker.peksag #29035: regrtest: simplify regex to match test names for the --fromfil http://bugs.python.org/issue29035 closed by haypo #29059: Windows: Python not using ANSI compatible console http://bugs.python.org/issue29059 closed by steve.dower #29090: PySerial crash on Windows XP and Python 3.4 http://bugs.python.org/issue29090 closed by haypo #29093: Windows launcher breaks history in Git Bash http://bugs.python.org/issue29093 closed by evan_ #29094: Regression in zipfile writing in 2.7.13 http://bugs.python.org/issue29094 closed by serhiy.storchaka #29105: code or doc improvement for logging.handlers.RotatingFileHandl http://bugs.python.org/issue29105 closed by python-dev #29118: Randit http://bugs.python.org/issue29118 closed by terry.reedy #29119: Unintentional hard reference assignment in Python version of O http://bugs.python.org/issue29119 closed by rhettinger #29122: set() bug http://bugs.python.org/issue29122 closed by rhettinger #29123: CheckSqlTimestamp is fragile http://bugs.python.org/issue29123 closed by berker.peksag #29128: No way to instsall win32com on python 3.6 http://bugs.python.org/issue29128 closed by zach.ware #29129: Copy/paste error in "8.13.14.1.1. Using auto" http://bugs.python.org/issue29129 closed by berker.peksag #29130: Exit code 120 returned from Python unit test testing SystemExi http://bugs.python.org/issue29130 closed by r.david.murray #29131: Calling printf from the cdll does not print the full string http://bugs.python.org/issue29131 closed by r.david.murray #29134: contextlib doc bug http://bugs.python.org/issue29134 closed by orsenthil #29135: run_proces logs the command without escaping parmaeters such t http://bugs.python.org/issue29135 closed by wgianopoulos #29138: No __hash__() inheritance warning with -Werror http://bugs.python.org/issue29138 closed by martin.panter #29146: Confusing "invalid token" exception when integers have leading http://bugs.python.org/issue29146 closed by berker.peksag #29148: Inheritance behaviour ambiguos http://bugs.python.org/issue29148 closed by xiang.zhang #29149: SSL.py recursion limit crash http://bugs.python.org/issue29149 closed by r.david.murray #29150: Bad cast@ _mysql_ResultObject_Initialize() result in code exec http://bugs.python.org/issue29150 closed by christian.heimes #29156: remove superfluous pow test for (nonexistent) long http://bugs.python.org/issue29156 closed by serhiy.storchaka #29159: Regression in bytes constructor http://bugs.python.org/issue29159 closed by inada.naoki #29160: pow with three int arguments works like it had two arguments http://bugs.python.org/issue29160 closed by mark.dickinson #29161: random.SystemRandom(seed) documentation issue http://bugs.python.org/issue29161 closed by rhettinger #29162: pyshell.py: name 'sys' is not defined http://bugs.python.org/issue29162 closed by OO O #29163: shlex error in posix mode and empty quotes http://bugs.python.org/issue29163 closed by berker.peksag #29166: Spam http://bugs.python.org/issue29166 closed by xiang.zhang #29168: multiprocessing pickle error http://bugs.python.org/issue29168 closed by vinay.sajip #29170: Curses Module should check for is_keypad and not window flags http://bugs.python.org/issue29170 closed by Roy Marples #29173: Python 3.6 on Windows wastes a lot of CPU cycles in a while lo http://bugs.python.org/issue29173 closed by prahladyeri

1 0

SystemError: new style getargs format but argument is not a tuple
by Terry Reedy 05 Jan '17

05 Jan '17

There are several recent question on Stackoverflow about SystemError: new style getargs format but argument is not a tuple PIL/pillow, 2.7, Nov 28, 2016 https://stackoverflow.com/questions/40844212/systemerror-new-style-getargs-… ImageDraw.draw.line(), python, Sep 14, 2016 https://stackoverflow.com/questions/39497458/imagedraw-draw-line-systemerro… opencv, cv2.line, python, Sep 19, 2016 https://stackoverflow.com/questions/32673359/systemerror-new-style-getargs-… opencv, cv2.imread/show, python, Nov 5, 2016 https://stackoverflow.com/questions/13225525/system-error-new-style-getargs… (code mostly from doc) ??.imread?, python, Jan 1, 2017 https://stackoverflow.com/questions/41419688/systemerror-new-style-getargs-… (and maybe others) [Would that people would identify OS and exact version of both python and imported module ;-)] No one commenting has a clue. Is message from Victor's new calling code? Was it backported to 2.7? Could error be a result of running old 3rd party binary that needs to be recompiled? -- Terry Jan Reedy

2 3

IRC logs via BotBot.me
by INADA Naoki 05 Jan '17

05 Jan '17

Hi. IRC #python-dev channel is nice place to know what happens recently. But I can't log in always because I use only laptop PC. I found BotBot.me seems nice IRC log service and used by some major channels. https://botbot.me/ I wonder if #python-dev is logged by BotBot.me. I'm sorry if it had rejected already.

4 3

The "humor" section of the website is a bit outdated...
by Ryan Gonzalez 05 Jan '17

05 Jan '17

Ok, in advance, my apologies for probably sending this to the wrong mailing list... Anyway, the most holy page in the entire Python website, containing all sorts of deadly and wicked items, seems to have gathered a couple pounds of dust: https://www.python.org/doc/humor/ And by "pounds of dust", I mean that 2 of the links don't work anymore: http://quotations.amk.ca/python-quotes/ seems to have been down for at least 3 years; available from archive.org ( http://web.archive.org/web/20111111175522/http://quotations.amk.ca/python-q… ) The Python humor wiki page is also gone: https://wiki.python.org/PythonHumor Being that the primary purpose is for proposing additions to the humor list, I don't think archive.org would be of much help for this one. -- Ryan (ライアン) Yoko Shimomura > ryo (supercell/EGOIST) > Hiroyuki Sawano >> everyone else http://kirbyfan64.github.io/

2 1

[RELEASED] Python 3.4.6rc1 and Python 3.5.3rc1 are now available
by Larry Hastings 03 Jan '17

03 Jan '17

On behalf of the Python development community and the Python 3.4 and Python 3.5 release teams, I'm pleased to announce the availability of Python 3.4.6rc1 and Python 3.5.6rc1. Python 3.4 is now in "security fixes only" mode. This is the final stage of support for Python 3.4. Python 3.4 now only receives security fixes, not bug fixes, and Python 3.4 releases are source code only--no more official binary installers will be produced. Python 3.5 is still in active "bug fix" mode. Python 3.5.3rc1 contains many incremental improvements over Python 3.5.2. Both these releases are "release candidates". They should not be considered the final releases, although the final releases should contain only minor differences. Python users are encouraged to test with these releases and report any problems they encounter. You can find Python 3.4.6rc1 here: https://www.python.org/downloads/release/python-346rc1/ And you can find Python 3.5.3rc1 here: https://www.python.org/downloads/release/python-353rc1/ Python 3.4.6 final and Python 3.5.3 final are both scheduled for release on January 16th, 2017. Happy New Year, //arry/

1 0

--with-fpectl changes the CPython ABI
by Nathaniel Smith 03 Jan '17

03 Jan '17

Hi all, Well, we finally got that ucs2/ucs4 stuff all sorted out (yay), but I just learned that there's another CPython build flag that also changes the ABI: --with-fpectl Specifically, it seems that if you build CPython with --with-fpectl, and then use that CPython to build an extension module, and that extension module uses PyFPE_{START,END}_PROTECT (like e.g. Cython modules do), and you then try to import that extension module on a CPython that *wasn't* built with --with-fpectl, then it will crash. This bug report has more gory details: https://github.com/numpy/numpy/issues/8415 The reverse is OK -- extensions built in a no-fpectl CPython can be imported by both no-fpectl and yes-fpectl CPythons. So one consequence is easy -- we need to make a note in the manylinux1 definition saying that you have to use a no-fpectl CPython build to make manylinux1 wheels, because that's the only way to guarantee compatibility. I just submitted a PR for this: https://github.com/python/peps/pull/166 (Fortunately the manylinux1 docker images are already set up this way, so in practice I think everyone is already doing this.) But... in general this is kind of an unfortunate issue, and it's not restricted to Linux. Should we do something? Some options: Add another ABI flag -- e.g. cp35dmf vs. cp35dm? Though AFAICT the offending macros are actually part of the allegedly stable ABI (!!), so I guess this isn't really a solution. (I'm not 100% confident about how to tell whether something is part of the stable ABI, but: Python.h unconditionally imports pyfpe.h, and pyfpe.h doesn't have any Py_LIMITED_API checks.) Drop support for fpectl entirely in 3.7 on the grounds that it's not worth the trouble? (The docs have said "don't use this" at the top forever[1], and after clicking through every hit on github code search for language = Python and string "turnon_sigfpe" [2], I found exactly 4 non-documentation usages [3], all of which are already broken in no-fpectl builds.) We obviously can't do this in a point release though, because there are lots of extant extension modules referencing these symbols. Or maybe make it so that even no-fpectl builds still export the necessary symbols so that yes-fpectl extensions don't crash on import? (This has the advantage that it can be done in a point release...) Thoughts? -n [1] https://docs.python.org/2/library/fpectl.html [2] https://github.com/search?l=Python&p=1&q=turnon_sigfpe&type=Code&utf8=%E2%9… [3] https://github.com/podhrmic/JSBSim/blob/36de9ac63c959cef5d7b2c56fb49c1a57fd… https://github.com/tmbdev/iuprlab/blob/239918b5ec0f8deecbc7c2ec1283a837d11a… https://github.com/wcs211/BEM-3D-Python/blob/874aaeffc3dac5f698f875478c3acc… https://github.com/neobonzi/SoundPlagiarism/blob/7cff7f0145217bffb3a3cebd59… -- Nathaniel J. Smith -- https://vorpus.org

3 5