Hello, Maxim Cournoyer <ma...@guixotic.coop> writes:
> SW is supposed to have support to ingest tarballs as far as I remember, > so there shouldn't be an advantage to one over the other in this regard. More precisely, once the package is in Guix proper (and only then), its tarball metadata is computed with Disarchive¹: https://ci.guix.gnu.org/jobset/disarchive Then there’s a program that runs periodically and updates the Disarchive database published at https://disarchive.guix.gnu.org: https://codeberg.org/guix/maintenance/src/branch/master/hydra/sync-disarchive-db.scm At that point, ‘guix lint -c mupdf’ would normally find the tarball metadata in the Disarchive database, find the source tree it refers to in the SWH archive, and report success (well, report nothing). You can see that in action with the “Source” field of the package browser at <https://hpc.guix.info/browse>, as in this example (the ‘openmpi’ package uses tarballs as its source): https://hpc.guix.info/package/openmpi In the case of mupdf though, it seems that the tarball didn’t make it in the Disarchive database: https://hpc.guix.info/package/mupdf HTH! Ludo’. ¹ You can learn more about Disarchive + SWH in this wonderful paper: <https://hal.science/hal-04586520v1>. :-)
