Opam repository archival, Phase 1: unavailable packages

I think going on their Matrix channel and saying “hey, FYI there’s a new opam repository at $URL, could you please add it to the list of archived repository ?”

Since it is the case that all packages will first go to opam-repository, and only thereafter may appear in opam-repository-archive – does “software heritage” need to change a thing?

Indeed, it should not be necessary in case everything goes well, but it has been the case at least once that the archival mechanism was broken for a long time without anyone noticing. In such a case, they could miss a package. It will not cost more to archive this new repository because they are able to detect duplicated stuff.

EDIT: I can also imagine a case where a package was unavailable for a long time, but suddenly someone put the tarball back online and update the URL in this new repository. It would be nice to archive it.

Is there some description of how software heritage and opam-repository interact, so I can read up on that?

There’s a paper in french describing how 1. SWH archives some opam repositories (ocaml/opam-repo and the Coq one) 2. how opam is able to fetch a tarball from SWH when it is not available online anymore at the URL given in the opam package file. I can’t think of a place where this is described in english unfortunately.