zathura 2026

[dkwo]

• I tested the changes in this PR: YES
• I built this PR locally for my native architecture, (aarch64-glibc)

cc @lemmi

[dogknowsnx]

Regarding girara: Upstream moved Gtk+-specific parts to zathura

Regarding musl: #57776 (comment)

[fanyx]

can be rebased on master, meson fix is merged.

[dkwo]

@dogknowsnx I believe girara soname was already taken care, and disabled seccomp on musl for now (issue upstream), since we already use landlock that should be fine. What do you think?
@fanyx Thanks, rebased.

[dogknowsnx]

girara builds fine w/o gtk+3-devel last time I tried, should probably be moved to zathura's makedepends?

EDIT (post-thumbs-up):

disabled seccomp on musl for now

Some thoughts: Not entirely sure, but personally I'd still prefer (temporarily) disabling tests (waiting for fix by upstream) over switching off a built-in security measure, given that the zathura-sandbox binary still runs fine on a native musl system (can't test myself) despite the one failing weston-based test - again, we currently don't even run tests for weston itself.

[dkwo]

@dogknowsnx Sure, I can do that. Note that the musl fix has been merged upstream, so it will be in the next release, and also that the landlock snadbox (imho better) is always enabled.

[dogknowsnx]

There don't seem to be any advantages of updating right now, imho. Personally I'm happy to keep using zathura-0.5.13_1 as long as the issues mentioned above aren't sorted out upstream.

[dkwo]

All issues are fixed for me with v2026.03.27

[lemmi]

Alright. Thanks for keeping up. I'll try and have a look this week.

[Andrew-Elia]

I would love to see zathura updated soon to help with the issues of PDFs having different page sizes. pulls: pwmt/zathura#851 and pwmt/zathura#849 (some issues fixed: pwmt/zathura#724, and pwmt/zathura#447 Felipe-9's comment, the latter is the issue I am currently having.)

I really appreciate your work, Thanks a lot!

[dkwo]

This happens on ci only, after update to gtk4:

6/12 xvfb_session     ERROR             0.60s   (exit status 134 or signal 6 SIGABRT)
>>> MESON_TEST_ITERATION=1 MALLOC_PERTURB_=130 GTK_A11Y=none UBSAN_OPTIONS=halt_on_error=1:abort_on_error=1:print_summary=1:print_stacktrace=1 MSAN_OPTIONS=halt_on_error=1:abort_on_error=1:print_summary=1:print_stacktrace=1 G_TEST_SRCDIR=/builddir/zathura-2026.07.02/tests G_TEST_BUILDDIR=/builddir/zathura-2026.07.02/build/tests ASAN_OPTIONS=halt_on_error=1:abort_on_error=1:print_summary=1 EGL_LOG_LEVEL=fatal LIBGL_DEBUG=quiet MESA_LOG=null NO_AT_BRIDGE=1 /usr/bin/xvfb-run -s '-screen 0 1400x900x24 -ac +extension GLX +render -noreset' -a tests/test_session

[dkwo]

@sebastinas Thanks for taking a look! Is there a way in meson to skip a single test, which should not be run in ci?

[sebastinas]

If it's the tests running with xvfb that should be skipped, don't have xvfb installed. The tests requiring xvfb are skipped in this case.

[dkwo]

It is 6/12 xvfb_session that should be skipepd. If I omit xvfb-run rom checkdepends, then I get

Program xvfb-run found: NO
tests/meson.build:46:7: ERROR: Program 'xvfb-run' not found or not executable

Did I miss something?

[sebastinas]

The meson options are forcing the tests to be enabled. Leave it on auto to make the detection-based skip work.

EDIT: In the next release there will be an additional option to control running tests under Wayland / X11.