I'm still learning how the OCaml build tooling works. One of the recent issues about scaling of opam-repository got my attention. My take: I believe there needs to be a more sustainable approach to manually curating a central repository of packages rather than relying on volunteers to manually review and approve PRs. I appreciate all the hard work done by people in their free time for free! But that's a huge amount of busy work 😱🤯 I also still want to get an answer to the question: why do you need a central package repository at all? Can't you just be a thin wrapper around downloading packages from GitHub with the ability to pin by commits? You can put a thin proxy/caching layer between but that's about it. github.com/ocaml/opam-rep…

@ChShersh In terms of opam-repository the CI is there to validate the quality of packages added to opam. It’s a halfway solution between hackage and stackage from Haskell. This set of known mostly working pacakages is useful for testing changes to the compiler, adding new platforms etc

@ChShersh It does have the cost of compute, maintenance and people’s time. I think it is worth the effort compared to the chaos of an invalidated set of packages.

@lambda_foo @ChShersh It's worth it for now but the question is whether this effort and manpower is scalable as the repository grows in the future

@ChShersh The challenge as always for open source is funding and people’s time. If anyone wants an intro to how it works DM me, I was working on this full time until end of last year. Or get in touch with the maintainers or @tarides_ who does a lot of the maintenance on it.