On Tue, 22 Aug 2023 02:14:54 GMT, Andrew John Hughes <and...@openjdk.org> wrote:
> GHA regularly breaks because we specify a very explicit GCC version, even > down to the release versioning of the Ubuntu package. In just this last > week, it has caused issues with the testing of PRs on 11u > (https://github.com/openjdk/jdk11u-dev/pull/2084) and 17u > (https://github.com/openjdk/jdk17u-dev/pull/1672) > > Rather than bumping this yet again like > [JDK-8313428](https://bugs.openjdk.org/browse/JDK-8313428), this PR suggests > dropping the specific version as we did some time ago in 8u and have now done > in 11u (https://github.com/openjdk/jdk11u-dev/pull/2087). The requirement > still specifies a specific major version of GCC. It just means the dependency > isn't broken every time Ubuntu bumps to a new minor release or even just > makes a minor change to the package alone. > > Note that the current setup does not guarantee sticking with an exact version > of GCC anyway, because - as seen by recent GHA breakage - older versions get > removed from the package repository. All we get from this exact version > requirement is sporadic breakage of testing and developer time wasted fixing > & reviewing (and, in the case of backport trees, approving). If we truly want > a static version of GCC, we need to provide our own - or maybe even a full > devkit - as we do with the JDK. This looks okay. That said, it is rather unnatural to forward-port the patches. Next time, I would like to see a clean mainline RFE, which would then be cleanly backported to each of the JDK updates trees. ------------- Marked as reviewed by shade (Reviewer). PR Review: https://git.openjdk.org/jdk/pull/15374#pullrequestreview-1588655871