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

[Erik Joelsson]

On Thu, 31 Oct 2024 11:09:13 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

> > > Reliability was not the problem the `remove-bundles` block solves. Note the comment: "# Remove bundles so they are not misconstrued as binary distributions from the JDK project".
> > 
> > Based on discussions about the reasoning behind that comment, it appears to have been overly defensive.
> 
> Which discussion you are referring to?
> 
> > The consensus now is that deleting the bundles after 24 hours vs 1 or 2 hours should not change how people may construe these artifacts.
> 
> Who are the parties to that consensus?
> 
> I am not trying to be hard here, but this PR really surprised me. The long-standing way of doing policy changes in OpenJDK is to discuss things first (in the open!), and then do PRs. I would have expected to read up on the discussion about the relevance of that comment somewhere public, e.g. in bug report.

You are correct Alexey, policy changes for OpenJDK should be discussed in the open. This particular policy was implemented as a (probably misguided) attempt to adhere to Oracle policy, and that has now been deemed unnecessary, at least from Oracle's point of view. My assumption has always been that this accelerated deletion of binaries was something only we (Oracle) cared about (similar to other build system quirks, like autoconf build-aux files etc), but I may be wrong, and if so, we are of course ready to hear your arguments.

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

PR Comment: https://git.openjdk.org/jdk/pull/21801#issuecomment-2449832788