1. Forum
  2. Usenet
  3. LINUX.DEBIAN.BUGS.DIST

  • Bug#1100712: diffoscope: pytest autopkgtest on s390x fails due to new s

    From Luca Boccassi@21:1/5 to All on Mon Mar 17 19:00:02 2025
    Source: diffoscope
    Version: 289

    Dear Maintainer(s),

    Since the upload of file/1:5.46-2 there is a test in the pytest suite
    that consistently fails on s390x:

    https://qa.debian.org/excuses.php?package=file

    204s tests/test_quines.py:59:
    204s _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
    204s
    204s difference = <Difference filetype from file(1) -- filetype from file(1) []>
    204s filename = 'quine_expected_diff'
    204s
    204s def assert_diff(difference, filename):
    204s # Assign seen and expected values to local variables to improve contextual
    204s # information in failed tests.
    204s seen = difference.unified_diff
    204s expected = get_data(filename)
    204s > assert seen == expected
    204s E AssertionError
    204s
    204s difference = <Difference filetype from file(1) -- filetype from file(1) []>
    204s expected = ('@@ -1 +1 @@\n'
    204s '-gzip compressed data, was "recursive", from FAT filesystem (MS-DOS, OS/2, '
    204s 'NT)\n'
    204s '+Zip archive data, made by v2.0, extract using at least v2.0, last modified, '
    204s 'last modified Sun, Mar 04 2010 00:24:16, uncompressed size 440, '
    204s 'method=deflate\n')
    204s filename = 'quine_expected_diff'
    204s seen = ('@@ -1 +1 @@\n'
    204s '-gzip compressed data, was "recursive", from FAT filesystem (MS-DOS, OS/2, '
    204s 'NT)\n'
    204s '+Zip archive data, made by v2.0, extract using at least v2.0, last modified, '
    204s 'last modified Sun, Jan 28 2030 01:00:06, uncompressed size 440, '
    204s 'method=deflate\n')
    204s
    204s tests/utils/data.py:64: AssertionError


    I've re-run it a few times but no changes. Not sure if it's a genuine regression or just a test case that needs updating.

    This is affecting other packages such as libzstd:

    https://qa.debian.org/excuses.php?package=libzstd

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Luca Boccassi@21:1/5 to All on Tue Mar 18 12:10:01 2025
    On Mon, 17 Mar 2025 17:55:32 +0000 Luca Boccassi <bluca@debian.org>
    wrote:
    Source: diffoscope
    Version: 289

    Dear Maintainer(s),

    Since the upload of file/1:5.46-2 there is a test in the pytest suite
    that consistently fails on s390x:

    https://qa.debian.org/excuses.php?package=file

    204s tests/test_quines.py:59:
    204s _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
    _ _ _ _ _ _ _ _
    204s
    204s difference = <Difference filetype from file(1) -- filetype from
    file(1) []>
    204s filename = 'quine_expected_diff'
    204s
    204s     def assert_diff(difference, filename):
    204s         # Assign seen and expected values to local variables to
    improve contextual
    204s         # information in failed tests.
    204s         seen = difference.unified_diff
    204s         expected = get_data(filename)
    204s >       assert seen == expected
    204s E       AssertionError
    204s
    204s difference = <Difference filetype from file(1) -- filetype from
    file(1) []>
    204s expected   = ('@@ -1 +1 @@\n'
    204s  '-gzip compressed data, was "recursive", from FAT filesystem
    (MS-DOS, OS/2, '
    204s  'NT)\n'
    204s  '+Zip archive data, made by v2.0, extract using at least v2.0,
    last modified, '
    204s  'last modified Sun, Mar 04 2010 00:24:16, uncompressed size
    440, '
    204s  'method=deflate\n')
    204s filename   = 'quine_expected_diff'
    204s seen       = ('@@ -1 +1 @@\n'
    204s  '-gzip compressed data, was "recursive", from FAT filesystem
    (MS-DOS, OS/2, '
    204s  'NT)\n'
    204s  '+Zip archive data, made by v2.0, extract using at least v2.0,
    last modified, '
    204s  'last modified Sun, Jan 28 2030 01:00:06, uncompressed size
    440, '
    204s  'method=deflate\n')
    204s
    204s tests/utils/data.py:64: AssertionError


    I've re-run it a few times but no changes. Not sure if it's a genuine regression or just a test case that needs updating.

    This is affecting other packages such as libzstd:

    https://qa.debian.org/excuses.php?package=libzstd

    The very first time this failed:

    https://ci.debian.net/packages/d/diffoscope/testing/s390x/58769019/

    This was triggered by the src:file upload, and still had libzstd 1.5.6,
    proving that the issue is unrelated to src:libzstd, and most likely due
    to src:file.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Chris Lamb@21:1/5 to Luca Boccassi on Wed Mar 19 11:40:01 2025
    Luca Boccassi wrote:

    This was triggered by the src:file upload, and still had libzstd 1.5.6, proving that the issue is unrelated to src:libzstd, and most likely due
    to src:file.

    Thanks for looking at this. Updates to src:file have indeed caused
    regressions in the past as diffoscope relies heavily on its output.

    Christoph, is there a chance that this could be a s390x-specific issue
    with file? (Or, perhaps more likely, a Big Endian-specific issue? A
    quick glance at [0] suggests that s390x is Debian's only BE
    architecture at the moment.)

    [0] https://wiki.debian.org/ArchitectureSpecificsMemo


    Regards,

    --
    ,''`.
    : :' : Chris Lamb
    `. `'` lamby@debian.org 🍥 chris-lamb.co.uk
    `-

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Christoph Biedl@21:1/5 to All on Wed Mar 19 14:00:01 2025
    Control: reassign 1100712 file

    Christoph, is there a chance that this could be a s390x-specific issue
    with file? (Or, perhaps more likely, a Big Endian-specific issue? A
    quick glance at [0] suggests that s390x is Debian's only BE
    architecture at the moment.)

    Hopefully I understood correctly this is about tests/data/quine.zip¹ in
    the diffoscope sources. This smells like Big Endian indeed as I'm seeing
    this on powerpc as well. A fix will be needed for trixie, I'll take care
    of take. There are some more stories about this to share, but better
    over a beer.

    Christoph, returning to work

    ¹ sha256: d152d368ca2da00de987cfe58c868534410ab270773d59d0d7956e671d3adafd


    -----BEGIN PGP SIGNATURE-----

    iQIzBAABCgAdFiEEWXMI+726A12MfJXdxCxY61kUkv0FAmfau5cACgkQxCxY61kU kv1bng//XjA75m2VXQ3cEB1xovmdADbswNilbDE+3yIY8r2ju0q1k0t9vRrFt6A5 rWa7BxzP/37HVCA0xm0NnYt3JabrMotgPhpYBzGzqRtwBtuMi1P0TIcmHt63fd3l uc0IwQACkthVlTrP0JNXBzdoAfUaq4oQQiUrHmw5S1LuwWo2VkJtBm2jkbmr5Zte NKHGQ9RNExQkXbOR++hkUdoP+SC26BmDlPehnkMfupt8/dxw94zkWEarFUPvLYeO UEP8M6S5KRJeKfGjvkU5AJ3Fnolc+o0+tZPqt7C4FIsjIfDvpZDUpb9km2RdgScY M7XAOH22lOCgi4Epfq8JzG4c3C4ZWpXDF6JGi3G5doVkJCZZZXWO6IBPGBL2ZyOX 4qCSSvwWLL6wQzKPyMHzHQ3j5+sX1Y5WyEDp4ZYXASZ4kmF6T+qquYQ3L8Z/F2iP 4gk0FOTH3UppSBlPL7JNe4AouIR1P44VmgLXDMgxSEFIIqKW6vG6W1fjYJrwBaue 1e8AD+sXv9Y1O/SPu9pgBk9RY6Jq3JySxZa+wDX6i0WeS79u+1DiC5iQCA25S8oe Nr9FW0G6bcm3uX7gpv5ecpmvDvQEO2X4tBA9pzhO43EZnzZIFoG/Vo3oZbOjQAOu fDiVCgUZtgL1abhYJLdVdSycTxgHkcaVpVIjsC1Y0ZdmQIuZ4AM=
    =kyGn
    -----END PGP SIGNATURE-----

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Chris Lamb@21:1/5 to Christoph Biedl on Wed Mar 19 14:30:01 2025
    Christoph Biedl wrote:

    Hopefully I understood correctly this is about tests/data/quine.zip¹ in
    the diffoscope sources.

    Yes — apologies for the lack of context so had you to hunt that down yourself.

    There are some more stories about this to share, but better over a
    beer.

    Oh, mysterious! Looking forward to that…


    Best wishes,

    --
    ,''`.
    : :' : Chris Lamb
    `. `'` lamby@debian.org 🍥 chris-lamb.co.uk
    `-

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Christoph Biedl@21:1/5 to All on Thu Mar 20 18:50:02 2025
    Control: forward 1100712 https://mailman.astron.com/pipermail/file/2025-March/001481.html

    Chris Lamb wrote...

    Christoph Biedl wrote:

    There are some more stories about this to share, but better over a
    beer.

    Oh, mysterious! Looking forward to that…

    Yeah, you can find that between the lines but I don't want to attract
    the wrong people.

    Nevertheless, I think I managed to fix the issue upstream and will
    upload a new file version as soon as upstream accepts the patch.

    Beware: I expect this to break various diffoscope tests. Seems nobody
    noticed "Mar 04 2010" was not a Sunday ...

    Christoph

    -----BEGIN PGP SIGNATURE-----

    iQIzBAABCgAdFiEEWXMI+726A12MfJXdxCxY61kUkv0FAmfcVEIACgkQxCxY61kU kv3G6BAAyAF89j3i6H8INVud583F8AXsafNIUxbbNOxxCHs0jTYpXytX9eVzaYpc BkC0JPTtQOmL62WhaY+qMdgKwPVY6jlahZnZ/+KIRv/o3eJxBv91R6giBqsMEeXb Mps68g+EIr0s8zzoeypBM1W08W1tgNeKitGWzNEX4SB0HzVckT6slwGvORVa5nuV A+0qeCPHPEvCcfDmhMW3m8fIqQbFZdfDmDoMOQuqArPltBJlwr25WKZ15Ax90M/K LZtXEU9rGXDW1cTB1jYgIyZog6VlR7hxHd5IpKuQsxPguf9NCAc7N8il5s5KMbJt 7gt8d1GhA2e1OUQtzd3Lh6DJWK/DZnKcZm36ZEwryHJj4OWxLmj9b5YfSwHfL0M1 gmdKwpH9gClooKHuUo3uU9XRPpHe5vUe+gMq8eL0YvRsVmHQH+tR92edQh0tv18y WElI2CQiKgFO/1NOKQwISWvHTGaiMDBbWWl8HHK5UCydmJ4hU6BNABdEItw8WUYx HrZ5AKMUB6YPjOoKo31vwftDHsFUGsSZ6C4ilTqC7HRc0gQhgRBjsWInVdcVvscq 1Qs4VJY19/QiiNi+DOi0oiOf4FmK2LeGj1fjEuE/H0EM0NpGd3LX01T9YTwKmbfS k+HAvmMoxaBXpLaj3cu9qqFwvLh4r6IxmPLyzd82NxsTzHMHwxU=
    =GlP+
    -----END PGP SIGNATURE-----

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)