feat: use local cache to avoid generating logics twice #52
Conversation
| } | ||
| } | ||
|
|
||
| private getFileHash(filePath: string): string { |
There was a problem hiding this comment.
using a hash of the file contents rather than modified time so it's stable when run on different machines (e.g. in CI)
| } | ||
| } | ||
|
|
||
| public getChangedFiles(filePaths: string[]): string[] { |
There was a problem hiding this comment.
We're basing this off the original logic file. If a user modifies their generated type file, this will ignore it, and they'd need to run without the cache flag to generate it again
There was a problem hiding this comment.
we could also check hashes for the type files here, but if we run kea watch it'll basically negate all the benefit locally, as that'll keep writing over the logic type files (or at least that is what happens to me when running things locally)
|
@mariusandra I am rethinking this now and thinking that this implementation is very naive - we really need to check all imports of a logic file and their hashes to determine if the logic has changed and needs type generation. This will make caching much less productive for some changes e.g. if I make a change to something like the |
Adds a
--cacheflag that uses a local cache stored innode_modules/kea-typegen/cache.jsonto avoid generating logics twice if a file is unchanged.This speeds up running
kea-typegen writea lot on subsequent calls since we were previously generating all of them. In most cases, only one logic has been changed.We can also save this cache together with type files we've previously generated between CI runs to speed up type generation.