Hello again!
As promised, today will be a shorter update. I’m still getting plugged into the vast CI machinery surrounding GHC, and progress is about as fast as to be expected on the second week of the job.
My main area of effort continues to be improving GHC CI, and I’m still working on adding some observability. The specific ticket is ghc#21585, where the task is to add residency profiling when GHC compiles head.hackage. I have a merge request for generating and archiving the eventlogs during CI, but there are still bugs to work out.
While waiting for CI (heh), I also started looking into generating a baseline for the new data based on older versions of GHC. But that turned out to be more work than is reasonable, so we’ll go with Plan B: start immediately parsing data out of the CI archive into a database of metrics, and create the baseline later after enough successful runs have gone by.
Next week I’ll continue on this!
-Bryan