[08:53] <fwereade> hazmat, is it reasonable for a charm to have two relations with the same name?
[08:54] <fwereade> hazmat, IMO it is clearly crack -- you can't specify endpoints sanely, and you can end up running the wrong hooks
[08:55] <fwereade> hazmat, the fact  that charms currently happen to do it, and we've been lucky enough that the bugs have not yet been hit, is neither here nor there
[08:55] <fwereade> hazmat, (and the juju-* relation names are (almost?) all juju-info, which is *definitely* crack)
[08:56] <fwereade> hazmat, certainly the inconvenience of fixing any other juju-* names is minimal compared to the inconvenience of fixing the juju-info ones
[08:57] <fwereade> hazmat, is there some clever way around the problem that I'm missing?
[12:42] <hazmat> fwereade, the alternative is recognizing that the implicit provide is special and that hooks are not allowed for it, ie. don't attempt to execute them since those hooks are for the require side, and thus not breaking a bunch of charms.
[19:08] <fwereade> hazmat, I don't think that holds water -- sure, we could paper over the hooks bug, but I don't see any way in which it's actually sane to allow relation name collisions
[19:08] <fwereade> hazmat, and AFAICT that's what you're advocating