[15:48] <rbasak> Who can push to a git repo? The team owner, or the "repository reviewer", or both?
[15:48] <rbasak> Or something else?
[15:59] <cjwatson> rbasak: Just the owner of the repository, or anyone in that team if the repository owner is a team.  (Also currently Launchpad admins, but that's a bug.)
[15:59] <rbasak> That's useful. Works well for us. Thanks!
[16:00] <cjwatson> rbasak: Planned future change: the default repository for a source package in a distribution (i.e. lp:ubuntu/+source/PACKAGE) should be pushable-to by anyone who can upload to the corresponding source package; but that's not done yet.
[16:01] <rbasak> We don't want that - unless we can further filter that by ref wildcard.
[16:01] <rbasak> IOW, we want an uploader of a package to be able to push refs/tags/upload/*, but nothing else.
[16:01] <rbasak> And only the importer should be able to push refs/*
[16:02] <rbasak> Otherwise an uploader could cause the git repository to mismatch the archive's single source of truth
[16:02] <cjwatson> Ref-wildcard ACLs have long been planned though are also not yet implemented.
[16:03] <rbasak> Do you have a plan for us to be able to combine those two things though?
[16:03] <cjwatson> In general it sounds like we need a bit of an uplift in policy capabilities to match those requirements.
[16:03] <cjwatson> They aren't planned in a lot of detail.
[16:03] <rbasak> Understood, thanks
[16:04] <cjwatson> (If you need it any time soon you probably need to go through my management chain so I get some time ...)
[16:36] <nacc> cjwatson: did you have any further thoughts on where I could debug the JSONDecodeError I was hitting yesterday? I'm getting it pretty consistently now (and if i ignore it, the review does get requested correctly)
[16:42] <cjwatson> nacc: if it's returning an empty byte-string then that sounds like an LP bug; it might be worth seeing if you can construct a matching situation in the LP test suite
[16:42] <cjwatson> I don't see anything obviously wrong with that method
[16:43] <nacc> cjwatson: will do, thanks!
[16:44] <cjwatson> (sorry, I realise that's quite a big next step, but ...)
[16:44] <nacc> cjwatson: not a problem! I know you've got a lot on your plate -- I think for now I can handle that exception in our script and i'll note it down as a bug to follow-up on my end