RFR: 8341861: GHA: Use only retention mechanism to remove bundles

[Aleksey Shipilev]

On Thu, 31 Oct 2024 12:45:30 GMT, Doug Simon <dnsimon at openjdk.org> wrote:

> The initial discussion I'm referring to is a brief internal discussion with no pertinent details apart from what I've stated here.
> 
> Anyway, this is a trivial PR and so it seems appropriate to discuss the impacts as part of the review on GitHub so let me know of any further concerns you might have about this change.

Hold on, I am not complaining about the code change, I am complaining about the process here. Internal discussions are happening all the time, sure. Every PR comes with some sort of idea how things should be done, fine. It is not, however, a good move to claim the consensus is achieved just because you had a brief internal discussion, to which open project participants are not privy to. In other words, the bug/PR should clearly state things to try and make sense for everyone else, not only to PR authors. _Then_ we reach consensus.

For this particular thing, for years we operated under the policy that we do not ship any binaries in OpenJDK infra anywhere, or have an appearance of doing so. This is also captured by the code comment this PR deletes. AFAIU, this was the whole reason to have this block originally: https://github.com/openjdk/jdk/pull/9063#issuecomment-1153539168, and it was a long-standing behavior in GHA workflows. If that changes, you have to mention it, and you have to argue why it is not applicable / relevant anymore. FTR, I agree with the "24 hours retention makes no practical difference to 2 hours retention" argument, but that argument _has to be made_.

-------------

PR Comment: https://git.openjdk.org/jdk/pull/21801#issuecomment-2449863767