Lennart Spitzner

I would assume they started simultaneously, but I'd have to check again. Will also look into those traces, but i gotta run for now.

Tested; i even observed a case where the failing module was first now (i.e. `GHC C.hs; GHC A.hs; GHC B.hs; [error in C]`. The discrepancy in timing compated to a...

i do see high (good 90%) cpu usage during the parallel A,B,C case. for me, when i was in any kind of "error loop", i never observed any noticeable delay...

Updated, made the one remaining feature gated behind commandline flag. > For making paths absolute, I don't think that's the right place, and many users will prefer to see relative...

> what if loadMessage was empty I don't understand. Why would `loadMessage` be empty? And if it was, you'd get empty output (for that `Load`) just like you did before....

so the relative paths are written just like `src/Ghcid.hs` (and not anything like `./src/Ghcid.hs`) which means that a simple suffix check should suffice, i.e. "is the thing being replaced a...

(I can understand if this workaround is becoming too complex for its own good and you'd rather not merge. I may point out that I have been using this flag...

doesn't that just mean that the profiled build is lacking the -fprof-auto (or related flags) in the right places (the right deps)?

Another example of a "common use" is the following: - Multiple package project: common, backend, frontend - in common, use `ghcid --run=":!touch ghcid.success"` - in backend and frontend, user `ghcid...

I don't want to discuss/argue "correct/better setup" here. I think there are sufficiently many cabal new-build users to warrant having corresponding documentation. (And as a general note, I think an...