It looks like there is a lot of great discussion here, but there are some things that are looping again :-). One important discussion in particular is the (existing!) problem we have with too few reviewers:
This thread is devolving into unprovable theories about what happens when we move to Github PR’s, e.g. (not intending to pick on anyone, just giving examples), arguing that move to GH will help:
Arguing that move to GH will hurt:
Unfortunately, it is impossible to do science with an A/B comparison of both theories here. At the same time I think we can all agree that we need to invest in building the base of code reviewers in general regardless of what happens with PRs.
I’m personally also concerned about a subthread underlying this discussion. We have a few folks who feel like they are personally essential to the operation of the project through reviews in their specific areas. I believe them, but if so, that isn’t a good thing for the health of the LLVM project. This becomes a problem if that person “gets hit by a bus”, gets a different job, loses interest, etc. They also become short-term bottlenecks if/when they have a bad day, are busy, on vacation, or if contributors to the area they oversee grow. Trust me: “been there, done that” numerous times. I think this is a critical problem that we should look to scale by bringing in new reviewers and scaling our processes in general. (again, regardless of the discussion of PRs)
In an effort to help make the discussion productive, I think it would make sense to split this sub-discussion into its own thread. It is super important to the long term health of the project - I think it would make sense as a BOF or other talk at the devmtg as well.
-Chris