We need either option 1 or 3 from [[ https://phabricator.wikimedia.org/T49578#9915561 | this comment ]] on T49578 to be done to enable SVG output for the Score extension. Option 2 (backport Lilypond 2.24 to Buster) is no longer tenable since Buster is now in ELTS per [[ https://phabricator.wikimedia.org/T49578#10425725 | comment ]] from PuckNZ.Could serviceops please update the LilyPond binary accessible to MediaWiki?
I strongly recommend option 3 (use upstream release tarball) which can be done easily on an existing Buster shellbox, without having to build Lilypond or a new Bookworm shellbox (although, this option will also work on a rebuilt Bookworm or Trixie shellbox). Upstream Lilypond now do compiled tarball releasesTo unblock the parent task, and installation is described in the Score extension [[ https://gerrit.wikimedia.org/r/plugins/gitiles/mediawiki/extensions/Score/+/refs/heads/master#installing-lilypond-and-prerequisites | README ]].we need either a Debian package for LilyPond 2.24 patched to include Cairo support, This involves downloading and extracting the (presumably x86_64) tarball from the upstream [[ or a container with the [[https://gitlab.com/lilypond/lilypond/-/releases/v2.24.4 | 2.24.4 release ]] page into `/usr/local/lilypond` (or wherever makes sense) and symlinking Lilypond's binaries into `/usr/local/bin` and/or configuring the two Score extension variables `$wgScoreLilyPond` and `$wgScoreAbc2Ly` to point to the binaries as appropriate3/ | upstream tarball]].
I do not recommend option 1 (rebuild the Lilypond shellbox with Bookworm and use the Debian packages) because the Bookworm Debian packages of Lilypond 2.24 are //still// broken (missing libcairo build flag).LilyPond currently runs on the `shellbox` cluster, Using these packages with the Score extension configured for SVG output results in an error caused by Lilypond exiting with:it doesn't have its own special cluster.
`fatal error: compiled without CAIRO_BACKEND`
Fixed Lilypond 2.24 packages for Bookworm //might possibly// appear in backports one day, but I really don't think we should wait around for who knows how long, given the total lack of communication on the relevant [[ https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1067578 | Debian bug ]] from the package maintainer.