WIP: chore: Add back LTS nightlies #24710
Conversation
| uses: crazy-max/ghaction-import-gpg@v6 | ||
| with: | ||
| gpg_private_key: ${{ secrets.SCALA_PGP_KEY }} | ||
| passphrase : ${{ secrets.SCALA_PGP_PASSPHRASE }} | ||
| fingerprint : ${{ vars.SCALA_PGP_FINGERPRINT }} |
There was a problem hiding this comment.
I'm not really a fan of using random actions from internet and feeding them with our secrets. Potentially we can use here commit SHA instead of mutable tag version (even though it's less visible) - same should be done eventually for next nightlies.
For 3.8.0 releases I'm currently using the old workflow:
- I've restored the PGP setup in sbt 64a1c4d
- seperated the org.scalajs and org.scala-lang aritfacts (otherwise it would fail Maven validation, cannot publish multiple namespaces in same deployement) 1987102
I'm planning to forward-port these to main when we'll be closer to the stable release.
There's even more issues that arised recently in the new build which are fine for nightly releases, but fail in Sonatype:
- clean recentlly stopped pruning
target/sona-staging- 2968bbf
There was a problem hiding this comment.
I just copied over the other nightly releasing, for sure I can adjust to your changes yeah. I guess we also need scala#659 ?
Btw. wouldn't the command I use here only release the compiler? I copied over the command from the original nightly releasing.
There was a problem hiding this comment.
Ok, I reverted using that action, crazy-max does not indeed sound great confidence wise. Should I add back all the steps from the original nightly workflow? I think the current changes should work as we didn't change almost anything in the build in scala3-lts repo (which I now realized)
| set -e | ||
|
|
||
| # Usage: | ||
| # SONATYPE_USER=<sonatype user> SONATYPE_PW=<sonatype pw> PGP_PW=<pgp pw> PGP_SECRET=<pgp secret> ./sbtPublish <publish cmd> |
There was a problem hiding this comment.
Still uses PGP_PW and PGP_SECRET not used in this script.
I think it would better to continue using the old script if possible, should be possible if we'd restore changes from release-3.8.0 mentioned below
There was a problem hiding this comment.
I think we might not need the script altogether. I will just use sbt
There was a problem hiding this comment.
I got back to the state in 3.7.4 tag
tgodzik
force-pushed
the
publish-lts-nightly
branch
from
59678ea to
f719ce7
Compare
2 participants
I still need to test it out, let me know if anyone spots something obviously wrong