-
So we're adding a new feature to Streamplace via a
place.stream.example.record, which then obviously needs to be in theplace.stream.authFullpermission set. Am I correct in understanding that the best practice is: release the change to the permission-set, wait 30 minutes, release the feature? And then everything should just work? It's kind of a bummer to add a 30-minute timer to our release process; currently the latest version of the schemas go out automatically with new versions of the software. -
How the heck do I hack on a permission set in development? If I want to add, say, permissions to create/update/delete
place.stream.example.recordto the permission set, I can't validate the permission set works until it gets rolled out to production. The only option I can think of is allow the Streamplace dev environment to operate in a fully parallel universe where every record created and indexed is e.g.com.elis-dev-environment.chat.message; but that's a pretty massive change.
(Note: allowing our permission set to do create/update/delete for
place.stream.* would solve 1 and 2 at the same time really well.
I mean, I still couldn't dev on it, but we could be pretty confident
that it wouldn't break.)
- Filed this separately as
bluesky-social/atproto#4424 but that one's rough
because our motivation to move off of
transition:genericis that users are asking us why we're asking for permissions to access their Bluesky account and we aren't.