Closed andreasabel closed 9 months ago
I looks a lot like this commit pushed the checking time for the library up from around 35min to 50min. Is there a reason for that?
Might be just noise in the github-builttimes...
Looking at recent runs, I see build times between 3 and 110 minutes: https://github.com/agda/cubical/actions?query=branch%3Amaster+event%3Apush What seems to matter most here is how much of the cache can be reused.
A priori, I do not see a reason why bumping the GHC version would cause a regression in Agda type-checking times, but of course there could be a GHC regression... One would need some reproducer...
Thanks for the input - I guess I should do some profiling locally, when I have some time...
Looks reasonable to me -> merging.