I've been exploring buck2 for work a lot recently. We're considering switching to either it or Bazel.
It'll be an incredibly risky decision to go with buck2 today, but it's so darn zippy in comparison to Bazel it's hard to not at least give it a real try.
My impression so far has been it's pretty robust in it's core, but anything that FB doesn't use is in a mediocre state. It's yet to be seen how willing they'll be to take PRs that are explicitly for community usage.
> It's yet to be seen how willing they'll be to take PRs that are explicitly for community usage.
Their stated position is that they're interested in receiving them, and watching it play out in practice, so far, they've been very responsive. Both with taking PRs and for simply fixing issues by writing their own.
Neil from the Buck2 team here. I think your points are pretty much spot on - everything we use internally is battle tested, everything we don't is less tested. We've tried to use that to our advantage by making as much as we can available externally, so its really just talking to services that only exist internally that are custom written for open source.
We're very willing to take PRs explicitly for community usage, with the standard caveats that it needs to be maintainable, testable, not regress other behaviours etc. But those shouldn't be insurmountable. Best way to tell for real is go through our PR backlog.
It'll be an incredibly risky decision to go with buck2 today, but it's so darn zippy in comparison to Bazel it's hard to not at least give it a real try.
My impression so far has been it's pretty robust in it's core, but anything that FB doesn't use is in a mediocre state. It's yet to be seen how willing they'll be to take PRs that are explicitly for community usage.