Hello,

thanks for the report. I tend to disable these tests upon next upload and then 
follow the upstream issue, when the tests can be re-enabled.

Hilmar

08.11.2025 11:35:24 Niko Tyni <[email protected]>:

> Package: latexml
> Version: 0.8.8-3
> Tags: upstream
> Forwarded: https://github.com/brucemiller/LaTeXML/issues/2670
> Severity: important
> User: [email protected]
> Usertags: perl-5.42-transition
> 
> This package fails its test suite with Perl 5.42 (currently in
> experimental.)
> 
>   #   Failed test 't/fonts/abxtest'
>   #   at /tmp/autopkgtest.KoXZhU/build.SEA/src/blib/lib/LaTeXML/Util/Test.pm 
> line 120.
>   Wide character in print at /usr/share/perl/5.42/Test2/Formatter/TAP.pm line 
> 125.
>   # Difference at line 4810 for t/fonts/abxtest
>   #       got : '                    <XMTok font="italic" name="widecheck" 
> role="OVERACCENT">ˇ</XMTok>'
>   #  expected : '                    <XMTok name="widecheck" 
> role="OVERACCENT">ˇ</XMTok>'
>  
>   #   Failed test 't/fonts/mathaccents'
>   #   at /tmp/autopkgtest.KoXZhU/build.SEA/src/blib/lib/LaTeXML/Util/Test.pm 
> line 120.
>   Wide character in print at /usr/share/perl/5.42/Test2/Formatter/TAP.pm line 
> 125.
>   # Difference at line 124 for t/fonts/mathaccents
>   #       got : '                  <XMTok font="italic" name="check" 
> role="OVERACCENT" stretchy="false">ˇ</XMTok>'
>   #  expected : '                  <XMTok name="check" role="OVERACCENT" 
> stretchy="false">ˇ</XMTok>'
> [...]
>   Test Summary Report
>   -------------------
>   Failed 1/35 test programs. 2/452 subtests failed.
>   t/22_fonts.t            (Wstat: 512 (exited 2) Tests: 23 Failed: 2)
>     Failed tests:  2, 12
>   Non-zero exit status: 2
> 
> I'm attaching a full log from a local autopkgtest run with a Perl
> 5.42 test repository available (see below).
> 
> The test suite is not run at build time, but only with autopkgtest
> checks. Unfortunately the build-needed restriction doesn't really work
> with mixed repositories on ci.debian.net, so the test runs there are
> misleading and were really done with Perl 5.40. I only caught this
> regression by luck, otherwise it would have gone unnoticed until Perl
> 5.42 actually entered sid.
> 
> Please consider running the tests at build time too for better visibility
> in the future. That would have caught this much sooner.
> 
> As for the cause, it's a change in Unicode 16 boiling down to
> 
>   perl -e 'die if "\N{CARON}" =~ /^\p{Latin}$/'
> 
> which changed behaviour (started dying) with Perl 5.42 (bundling Unicode
> 16.) This was intentional, see
> 
>   https://corp.unicode.org/pipermail/unicode/2025-October/011530.html
> 
>   https://www.unicode.org/L2/L2023/23280-scx-additions.txt
> 
> The upstream issue at https://github.com/brucemiller/LaTeXML/issues/2670
> has the links and details, but there's no fix yet AFAICS.
> 
> If you want to test changes against Perl 5.42 in experimental and run
> into uninstallability problems, there is a test repository of rebuilt
> Debian sid packages for amd64 available at <https://perl.debian.net/>.
> 
> Thanks for your work on Debian,
> -- 
> Niko Tyni   [email protected]

Reply via email to