Reopening since this was changed anonymously. Also why the hell is it so hard to prevent people from changing statuses anonymously in JIRA?

The link for adding a release to a collection currently looks like this:
http://musicbrainz.music.mcgill.ca/collection/296c0d92-4eed-4f45-a46c-d1e3c4c4118f/own_collection/add?release=428732

Are we happy using database ids to refer to releases or should we be replacing it with a uuid? My plan was to have a similar url structure for when collections are created:
http://musicbrainz.music.mcgill.ca/collection/create?release=uuid

For a one-time endpoint like this, I'm not sure it matters if we use rowids rather than MBIDs – the MBIDs are beneficial for storing more than for immediate things. I don't know how collections at present deal with merges, admittedly, since the schema uses rowids – I imagine it cascades them or updates those entries somehow.

Using an MBID does make links have a longer lifetime – so I guess the question is whether we think many people from external sites are going to be linking add-to-collection URLs (I'd guess not, in which case, rowid seems fine).

Right. From what I've found, merges are dealt with, because the actual links in the database tables use uuids, and there's a bunch of merge functionality. Good point about the difference between links and permanent entities. I'll stick with rowid for now

They don't use UUIDs, actually: https://github.com/metabrainz/musicbrainz-server/blob/master/admin/sql/CreateTables.sql#L890 but I'm guessing the way merges are done automatically updates those rows.

I would prefer we use an MBID. We should really be using MBIDs wherever we can, row IDs are an internal thing.