I'm splitting this from T647, as HAL integration is not the only use case for SWORD-based deposit.
Hosting complete & corresponding source code tarballs for compliance reasons is another one.

LGTM, and nice catch

In T743#16529, @seirl wrote:

• shouldn't the format be a parameter instead of a part of the URL? i.e., ?format=foo instead of /foo at the end?

I see some reasons why not:

First of all thanks for this useful summary @seirl !

Thumbs up for the good work!
I like it.

This is now done on uffizi.
Potentially remaining sub-tasks before closing this:

Updated SQL to also delete objects from tables that references them, e.g., the indexer ones.

thanks for the review, I've updated the SQL query accordingly

And what would be the actual on-the-wire serialization format used to send the bare repo to users? Some sort of archive of the .git dir or what?

I'm not opposed to having explicit hash scheme names in the IDs—it is a good idea, only to be weighed against the cost in terms of length.
But we should also have schema version numbers, in case more radical changes will be needed in the future, e.g., renaming the object types in the graph.
If we retain both suggestions, that would give:

• swh:1:revision:sha1_git:<git sha1 of a revision>
• swh:1:content:blake2s256:<blake2s256 of a content>

(agreed)