Open ProofOfKeags opened 5 years ago
When was the last time you tried h-i-e
? The performance has got a lot better in recent releases (but is still not perfect). I can use it on GHC's code base for example, which is over 500 modules.
Still falls over a bunch for me with extensions in .cabal
files and quasi-quoters.
I haven't determined the cause but I just compiled 0.12.0.0 and it is eating all the available memory on my machine.
From the above discussion it seems like this isn't a goal? Would it not be more performant? Maybe there's some caveats I don't know about?
It is a goal, e.g. in ghcide
(I think it is already implemented there) which hls
is built on top of, and hls
will be the successor to hie
.
https://gitlab.haskell.org/ghc/ghc/wikis/hie-files#making-use-of-hie-files-as-a-tooling-developer
As a partially informed observer, this seems to be a major helper for things like the IDE engine. Currently the performance of HIE makes it unusable for all but the smallest of projects. However, I expect a lot of that can be alleviated by taking advantage of GHC emitting these files. How much of the core architecture needs to be changed in order to take advantage of this?