I can create a new Project in secure.phabricator.com, to help with task triage.

The Releeph application can start off being disabled by default. If we're happy there won't be any other disaster inducing architectural changes, I could document it more publicly as well.

Are there any other logistics issues I should be thinking about?

I'd like to review it before we land it, which might or might not be easier in one big commit. Maybe throw it up as a one diff and we can decide from there? As long as it isn't like 50K lines we can probably do it in one go.

As long as the app is default-disabled I don't think it can cause any logistical problems. Do you have hooks/events into other apps or markup rules or top-level /Dnn style routes? The only other stuff I can think of is stuff like duplicate resource names, but that should already be getting caught.

Basically, fire when ready.

Just /RQnn, which takes you to the ReleephRequest of that number. I think that's the only top level route: everything else is in PhabricatorApplicationReleeph.

The first diff here was 10 kilolines, and with a bit of churn that might be a bit bigger but I don't think it will be too long. Releeph embraces quite a lot of concepts from Differential and Maniphest, so it should all be pretty familiar looking :D

I'm going to move these dependencies to T2714 and reorient this around the arc part.