On Mon, 28 Oct 2024 13:59:12 GMT, Aleksey Shipilev <sh...@openjdk.org> wrote:
>> .github/actions/get-jtreg/action.yml line 59: >> >>> 57: - name: 'Build JTReg' >>> 58: run: | >>> 59: # If runner architecture is x64 set JAVA_HOME_17_X64 otherwise >>> set to JAVA_HOME_17_arm64 >> >> This logic seems missing in your new script? Is it not needed? The new >> script seems to have X64 hardcoded; it looks wrong compared to this. > > Intentional simplification: since we no longer build JTReg on actual various > platforms, we can rely on current `ubuntu-22.04` runner that executes > `prepare` job to be Linux x86_64. This also dodges the transient headache > with building JTreg on different (broken) MSYS envs on Windows, if you > remember that mess. Right. *doh* I don't know what kind of brain meltdown I had there... >> .github/workflows/main.yml line 80: >> >>> 78: with: >>> 79: sparse-checkout: | >>> 80: .github >> >> I'm guessing this does not add much extra time? > > Yup! This checks out _only these files_, not the whole JDK workspace. Seems > to run in < 1s. This would also eventually help to resolve a TODO to move the > rest of `prepare` job to scripts: it was apparently inlined _because_ we did > not want to check out the entire JDK repo. Yes, that was my doing. I did not know about, or think about, the ability to check out just part of the tree. This might open for some cleanup and better structuring of the code... ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/21692#discussion_r1819124696 PR Review Comment: https://git.openjdk.org/jdk/pull/21692#discussion_r1819128594
