Seems like there's two things going on in this commit?

yep, I should split it. Second one is just visual unification.

Commit b19f1760 fixes this pull-request

@ktdreyer @tkopecek How do we handle Debian Source Control artifacts (*.dsc + *.tar.gz + *.diff.gz files) with this?

And how would we handle "source debs" created by debbuild (*.sdeb)?

(To be clear, "Debian Source Control" is the source artifacts for building packages using the traditional debian/ folder based process, whereas the "source deb" is the source artifacts for building packages using the debbuild tool).

We don't (except tar.gz which is generic tar type). Do you need these? I can add it in additional PR.

Note, now that we have the addArchiveType RPC, it's possible for anyone to add their own archive types to their own instances. You can do it with "koji call addArchiveType" or the koji_archivetype Ansible module. https://github.com/ktdreyer/koji-ansible#koji_archivetype

When we add entries to schema.sql, it slows down integration tests that frequently reset the database. For example, the integration tests I run for koji-ansible frequently flush and reinitialize the DB, so I'm interested in keeping that fast.

My opinion is that we should not add sdeb entries at this point until there is some other content generator that can make use of them so we do not bloat koji with unimplemented features.

Considering all of these I brought up are generated outputs of Debian package builds, yes they do need to be added to Koji.

I agree with Ken that anybody can add the btype now via API so we don't need to have prefilled everything.

Then why did you add this?

Then why did you add this?

You've got me - it shouldn't have been merged in first case. I was too quick to have it in upstream when I was adding it to my instance.

Then let's revert it unless you plan to add all the necessary outputs to be collected as Debian package artifacts.

Agreed - reverted via https://pagure.io/koji/c/3f71076

When we want Koji to be able to handle building Debian packages, we'll need to add it to the schema, but until then, it's not terribly useful there.