tdf157829-rtl.fodt test failure

[Caolán McNamara]

On Sat, 2024-12-28 at 09:41 +0000, Mattia Verga wrote:
> WHile running tests for 25.2.0.1 I see a new test failure:
> 
> file:///builddir/build/BUILD/libreoffice-25.2.0.1-build/libreoffice-
> 25.2.0.1//sw/qa/extras/layout/data//tdf157829-ltr.fodt:
> (anonymous namespace)::TestTdf157829LTR::TestBody finished in: 42ms
> [_RUN_____] (anonymous namespace)::TestTdf157829RTL::TestBody
> file:///builddir/build/BUILD/libreoffice-25.2.0.1-build/libreoffice-
> 25.2.0.1//sw/qa/extras/layout/data//tdf157829-rtl.fodt:
> terminating test due to missing font: Noto Sans Hebrew
> 
> The font is installed (as you can see tdf157829-ltr.fodt is passing),
> yet tdf157829-rtl.fodt complains it isn't.

It's possible that in the passing document, that while Noto Sans Hebrew
is listed in the document, that it might not actually be used, while it
is used in the other documents. That said, assuming this is the .spec
file:
https://src.fedoraproject.org/rpms/libreoffice/blob/rawhide/f/libreoffice.spec#_297
then "BuildRequires: google-noto-sans-hebrew-fonts" does suggest that
the font is installed.

So it *might* be that the font is installed, but missing some glyphs
that it needs to have to render the offending text. Hard to tell. You
might have to attach a debugger to where that abort is, or some
additionally dumping of the contents of "rMissingCodes" to see if
there's any merit in that theory. If there are some missing codes, you
can check with fontforge to see if the "Noto Sans Hebrew" font is
really missing the glyphs that LibreOffice might believe are missing.