You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Poor filesystem scanning performance on very large repositories #176
At JetBrains formatting our monorepo takes about 31.8s which is quite slow, maybe we're not configuring it correctly?
We tried some git ls-files approach which takes it down to 5.6s, maybe there are other more hermetic ways (because in our PoC git is resolved from PATH) to reach a similar level of performance 🤔.
yea i matched the original upstream file discovery just to make sure it was 1:1, I think that upstream solution is probably ok but I'm sure faster versions could be done like fd based ones. I wonder if we should provide that as an option (or just do it by default). but probably easiest if we first get excludes into the tool so we don't have to worry about matching that syntax
Activity
ArchangelX360 commentedon Oct 28, 2026
Could be mitigated by bazel-contrib/buildtools#1486 actually
keith commentedon Oct 30, 2026
yea i matched the original upstream file discovery just to make sure it was 1:1, I think that upstream solution is probably ok but I'm sure faster versions could be done like
fdbased ones. I wonder if we should provide that as an option (or just do it by default). but probably easiest if we first get excludes into the tool so we don't have to worry about matching that syntax