leana and I tried cloning ghc just now and it just fails on the submodule,
furthermore cloning from the github mirror doesn’t work either because all submodules point to gitlab ![]()
leana and I tried cloning ghc just now and it just fails on the submodule,
furthermore cloning from the github mirror doesn’t work either because all submodules point to gitlab ![]()
Hoogle is also down: Haskell Infra Status
The fact that the GHC’s mirror on GitHub doesn’t also mirror the submodules pretty much renders the mirror useless in this kind of incident, and Haskell GitLab becomes the single point of failure when people don’t have a local copy of GHC’s source tree.
I think it would be nice to also have submodules mirrored on GitHub or Codeberg to have redundancy, what do you think?
That’s a great idea! I will make sure this is given a consideration.
Setting up a mirror on codeberg should be easy. ![]()
In my opinion we should get rid of sub modules because they go against the spirit of git.
Imposing upon themselves, sub modules break other commands like cloning and remote management all the sudden has to be tracked, making mirroring harder than it needs to be.
we spend half an hour trying to figure out how to use a worktree with the module system, to revert back to main from our branching point. only to find out that submodules do shallow clones, so reverting without remote was impossible ![]()
There is also @simonmic’s:
I have added downloads to the list, because (for me) https://downloads.haskell.org/ in the browser yields:
Error 503 Backend is unhealthy
Backend is unhealthy
Error 54113
Details: cache-lcy-eglc8600070-LCY 1764415973 4253875345
Varnish cache server
However, Stack still appears to be able to fetch GHC binary distributions from URLs starting with that subdomain.
We’re doing that in stable haskell: Remove submodules by andreabedini · Pull Request #108 · stable-haskell/ghc · GitHub