Recently there was a
2 part change to rules_swift (and the other Apple related rules repos) to:
1. Stop tagging releases on the rules repos
2. Stop testing against and supporting bazel release versions
While I think #1 is fine and being on HEAD is the only way you can realistically work with bazel rules these days. I think it would be great if Google could reconsider #2 here.
This change has a lot of
repercussions that we're already seeing within days for rules_swift. It results in there being no bazel release version you can use with rules_swift HEAD (unless we want to build and distribute our own untested version of bazel at HEAD). This also means that as we submit fixes to rules_swift like
this one we have to hope that it patches cleanly on the actual version of rules_swift we can use at the time.
Overall this seems like a huge downside for the bazel community since I don't believe that encouraging users to build, test, and distribute their own versions of bazel is what we want. I also think it's a huge downside for users of rules_swift since it makes it less clear what rules versions (or shas) are compatible with what bazel versions, and what bazel releases they have to wait for in order to update. Finally, I think it’s a huge downside for contributors of rules_swift who now have less of an incentive to contribute since they can't update once their patches land.