STINNER Victor <vstin...@redhat.com> added the comment:

Ronald Oussoren gave more info on my previous PR 10780 ("platform.platform() 
uses mac_ver() on macOS"):

https://github.com/python/cpython/pull/10780#issuecomment-444529371

"""
The information does not include data about fat binaries, resulting amongst 
others in the following inconsistency:

ronald@Menegroth[0]$ arch -i386 python3.6 -m platform
Darwin-18.2.0-x86_64-i386-64bit

ronald@Menegroth[0]$ arch -i386 python3.6 -c 'import sys; print(sys.maxsize)'
2147483647

This platform output includes "64bit" because the binary for python3.6 includes 
support for both i386 and x86_64, and doesn't show that the command is using 
i386 instructions.
"""


I made some tests:

$ file /usr/local/bin/python3
/usr/local/bin/python3: Mach-O universal binary with 2 architectures: 
[i386:Mach-O executable i386] [x86_64:Mach-O 64-bit executable x86_64]
/usr/local/bin/python3 (for architecture i386): Mach-O executable i386
/usr/local/bin/python3 (for architecture x86_64):       Mach-O 64-bit 
executable x86_64

$ /usr/local/bin/python3 -c 'import struct, sys, platform; 
print(platform.architecture(), struct.calcsize("P"), sys.maxsize)'
('64bit', '') 8 9223372036854775807

$ arch -x86_64 /usr/local/bin/python3 -c 'import struct, sys, platform; 
print(platform.architecture(), struct.calcsize("P"), sys.maxsize)'
('64bit', '') 8 9223372036854775807

$ arch -i386 /usr/local/bin/python3 -c 'import struct, sys, platform; 
print(platform.architecture(), struct.calcsize("P"), sys.maxsize)'
('64bit', '') 4 2147483647


IMHO platform.architecture() should return 32bit when running "arch -i386 
/usr/local/bin/python3" to be consistent with struct.calcsize("P") == 4 and 
sys.maxsize == 2147483647. Otherwise, how would you notice that you are using 
the 32-bit flavor of Python?

My PR 11186 implements this fix.


Ronald Oussoren:
> Using sizeof(void*) or sys.maxsize suffers from the a simular problem: this 
> will only detect the pointer-size of the current proces and not that the 
> binary is capable of running with a different pointer-size as well.

Right, but I don't think that it's possible to report that Python executable is 
FAT binary in platform.architecture() result. If you want to provide such 
information, IMHO you should write a new function or at least add a new 
parameter to platform.architecture().

IMHO it's more consistent to report "32bit" for "arch -i386 python3" and 
"64bit" for "arch -x86_64 python3".

----------

_______________________________________
Python tracker <rep...@bugs.python.org>
<https://bugs.python.org/issue35348>
_______________________________________
_______________________________________________
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

Reply via email to