Closed tianon closed 6 months ago
Hmmm, as I look at this, 3.1 is also imminently EOL, so I think my opinion on this is that we should only revert the change for 3.0 and let 3.1 stay as-is (esp. so we can remove the conditional after 3.0 is removed and any downstream breakage from this should be pretty minimal).
Updated.
@yosifkit is AFK at the moment, but he's ACKed the approach/choices :heart:
Sorry for being a bit harsh earlier on, I was just surprised how the pull of the RUBY_MAJOR
env var did not satisfy a changelog and refering to the PR description it might indeed have been unintended. I was a bit confused about the process. I know that it might be just wrong to rely on env variable being set in a base image to be still available even with a minor version change. You proved me wrong, thanks. We should have had tests ourself, I get it!
(since it's effectively EOL now except the possibility of a severe regression: https://www.ruby-lang.org/en/news/2024/04/23/ruby-3-0-7-released/)
See https://github.com/docker-library/ruby/pull/435