cachepot
cachepot copied to clipboard
paritytech
Merge all of our improvements upstream
Since we forked, there has been some improvements made here but also we missed out on some of the PRs made against sccache (after the project got more action since the fork). This will be beneficial by getting new features/bug fixes but also will facilitate upstreaming some of our improvements as well.
We will be winding down cachepot as a separate project and instead plan to upstream all of our fixes to the sccache project, now that we have maintainer access and help maintain the crate and publish it. This issue tracks the required work.
As proposed at the meeting:
MVP for sccache in CI/CD
- [ ] finer cache hits tracking and non-hits
- [x] overlay and docker support
- #161
- [x] cache invalidation issue "artifacts fullpath" is worked around
- [ ] cache invalidation issue https://github.com/paritytech/ci_cd/issues/504
Outside of MVP
- distributed compilation for CI
- encrypted/signed shared artifacts
- undep OpenSSL https://github.com/paritytech/cachepot/issues/87
- metrics support
Stuff not related to the CI use-case, but nonetheless important
- distributed compilation for devs
cc @drahnr @kirushik
@Xanewok will be chipping away on that in the long run, I'll be focused on merging and reviewing upstream PRs. So there is no timeframe right now when which fix will be upstreamed.
We will be winding down cachepot as a separate project and instead plan to upstream all of our fixes to the sccache project, now that we have maintainer access and help maintain the crate and publish it. This issue tracks the required work.
Does this mean cachepot is basically deprecated and we should use sccache instead? I just want to confirm since it looks like this project is still fixing bugs and updating dependencies.
Yes, since sccache increased its dev bandwidth and we can actually upstream the work, I think it's fair to say you should use sccache now, as we merge our stuff there.