deps: alpine 3.19.1

[codefromthecrypt]

I also updated the description, but re-tagging the image will only update the alpine layer as latest JRE is the same build as before.

[reta]

Seems like 21.0.2_p13-r2?

[codefromthecrypt]

see comment above the line? :D

[reta]

Or sorry, my bad

[codefromthecrypt]

@reta hmm odd.. I wonder if there's an non-determinism issue in the package selection, as you can see 21.0.1 in the breaks part despite the ultimate expression used being ~openjdk21-jdk-21.0.2_p13 and ~openjdk21-jmods-21.0.2_p13 (where the tilde says get this prefix, which we need because it reduces maintenance when there's a package re-do)

 0.943 ERROR: unable to select packages:
0.945   openjdk21-jmods-21.0.2_p13-r2:
0.945     breaks: world[openjdk21-jmods~21.0.1_p12]
0.945     satisfies: openjdk21-jdk-21.0.2_p13-r2[openjdk21-jmods]
0.945   openjdk21-jdk-21.0.2_p13-r2:
0.945     breaks: world[openjdk21-jdk~21.0.1_p12]

[codefromthecrypt]

I think we don't need to separately add the jmods things, because it is already a dependency of the jdk
https://pkgs.alpinelinux.org/package/edge/community/x86_64/openjdk21-jdk Maybe this will reduce the issue

[codefromthecrypt]

can someone with an x86 machine, run from this branch ./build-bin/test 21.0.2_p13? from my macbook zero issues, but obviously something is happening here on x86, and not sure if there's some change in the runner or if something else is at play

[reta]

@reta hmm odd.. I wonder if there's an non-determinism issue in the package selection, as you can see 21.0.1 in the breaks part despite the ultimate expression used being ~openjdk21-jdk-21.0.2_p13 and ~openjdk21-jmods-21.0.2_p13 (where the tilde says get this prefix, which we need because it reduces maintenance when there's a package re-do)

Yeah, it confused (and derailed) me as well, not sure what is happening

[codefromthecrypt]

I think it is repository order, since openjdk is now not only in edge/community.. will do a quick change to verify, then maybe update the base layer

[reta]

can someone with an x86 machine, run from this branch ./build-bin/test 21.0.2_p13? from my macbook zero issues, but obviously something is happening here on x86, and not sure if there's some change in the runner or if something else is at play

No issues on Linux with ./build-bin/test 21.0.2_p13 for me

[codefromthecrypt]

lol force-broken-world just doesn't do anything

[codefromthecrypt]

so this isn't about 3.19.1 as same error reverting to 3.19.0... there's something funny and I don't think it is the mirror as I already switched that. Maybe something changed in the runner 🤷

[codefromthecrypt]

ok it had to do with the github workflow run. basically it ran install twice, once with a slightly older patch and then again with the current one. I guess the change in build-args isn't tainting the layers?

[codefromthecrypt]

ok good to go. sorry about the thrash, I forgot build-arg cache invalidation behavior