Bug 439800 – sys-apps/pkgcore-0.8.5-r1: fails tests with Python 3 (encoding-related issues)

• Status: RESOLVED FIXED
• Product: Gentoo Linux

[Description Michał Górny 2012-10-26 21:14:57 UTC]

Created attachment 327496 [details]
The build log

sys-apps/pkgcore-0.8.5-r1 fails its own tests when Python 3 is used as the default Python implementation. I guess that's because portage sets some random locale (C?) for tests and it fails to handle UTF-8 files properly then.

[Comment 1 Mike Gilbert 2012-10-26 21:37:41 UTC]

Traceback (most recent call last):
  File "/usr/lib64/python3.2/site-packages/snakeoil/test/__init__.py", line 236, in run
    self.forced_extra_frame(testMethod)
  File "/usr/lib64/python3.2/site-packages/snakeoil/test/__init__.py", line 201, in forced_extra_frame
    test()
  File "/tmp/portage/sys-apps/pkgcore-0.8.5-r1/work/pkgcore-0.8.5/build/lib/pkgcore/test/fs/test_fs.py", line 109, in test_init
    raw_data = open("/etc/passwd").read()
  File "/usr/lib64/python3.2/encodings/ascii.py", line 26, in decode
    return codecs.ascii_decode(input, self.errors)[0]
UnicodeDecodeError: 'ascii' codec can't decode byte 0xc5 in position 795: ordinal not in range(128)

I wonder how it is supposed to know what encoding your password file (/etc/passwd) has. Is that standardized anywhere?

[Comment 2 Michał Górny 2012-10-26 21:44:02 UTC]

(In reply to comment #1)
> Traceback (most recent call last):
>   File "/usr/lib64/python3.2/site-packages/snakeoil/test/__init__.py", line
> 236, in run
>     self.forced_extra_frame(testMethod)
>   File "/usr/lib64/python3.2/site-packages/snakeoil/test/__init__.py", line
> 201, in forced_extra_frame
>     test()
>   File
> "/tmp/portage/sys-apps/pkgcore-0.8.5-r1/work/pkgcore-0.8.5/build/lib/pkgcore/
> test/fs/test_fs.py", line 109, in test_init
>     raw_data = open("/etc/passwd").read()
>   File "/usr/lib64/python3.2/encodings/ascii.py", line 26, in decode
>     return codecs.ascii_decode(input, self.errors)[0]
> UnicodeDecodeError: 'ascii' codec can't decode byte 0xc5 in position 795:
> ordinal not in range(128)
> 
> I wonder how it is supposed to know what encoding your password file
> (/etc/passwd) has. Is that standardized anywhere?

No. I think only the GECOS field is allowed to have non-ASCII characters but that field can basically hold anything and pkgcore need to make use of its contents.

[Comment 3 Brian Harring (RETIRED) 2012-10-29 08:41:54 UTC]

0.8.6 will resolve it.