Chat Logs

  1. dreamrealmorning
  2. CandleWith a pull-request based workflow (and a protected `main` branch), how do people handle `mvn release:prepare` via a branch & pull request?
  3. dreamrealwouldn't that be the CI/CD's job in that case?
  4. CandlePartly, let's restrict that even more, and say that CI/CD does not have write access to the repo.
  5. dreamrealdon't know then
  6. dreamrealgpt
  7. dreamreal~gpt With a pull-request based workflow (and a protected `main` branch), how do people handle `mvn release:prepare` via a branch & pull request?
  8. javabotIn a pull-request based workflow with a protected `main` branch, you'd typically handle `mvn release:prepare` via a branch & pull request by creating a separate release branch. Here's a simple process: 1. Checkout to a new branch (e.g., `release/v1.0.0`). 2. Run `mvn release:prepare` in that branch. This will update the project versions, commit, and tag the new version. 3. Push the branch and create a pull request for it. 4. Once the pull request is re
  9. javabotbe merged into the `main` branch. This way, `mvn release:prepare` is executed in a controlled environment and the `main` branch is updated in a clean, traceable manner through the pull request.
  10. * dreamreal shrugs.
  11. CandleIt is feels like a fast-forward merge - to get the two sequential commits into main, but then the tag still references a commit on the branch.
  12. deebowhat does release:prepare do?
  13. dreamrealI'd think you'd tag the release preparation, not the incoming
  14. Candle1. commits a non SNAPSHOT version (e.g. changes 1.0.5-SNAPSHOT to 1.0.5). 2. creates a tag at that commit. 3. commits the next SNAPSHOT version (e.g. 1.0.6-SNAPSHOT).
  15. deebodo you really need like a pom.xml with <verion>5.0.0</version> in the repo? with gradle we just rewrite the the version for the build when releasing
  16. dreamrealmore than one road to rome
  17. CandleHaving the release version in VCS That gives you something you can tag in the repo.
  18. deebotrue, it just seems weird (to me), main/master should produce a snapshot every time, a separate release pipeline builds a specific commit as a release
  19. Candle(this is a long-standing problem that stricter repository permissions are ... encouraging.. us to actually try to resolve!)
  20. * dreamreal shrugs. Everyone has a lot of local practices, and somehow the world manages to not end as a result
  21. deebowe just ./gradlew build && ./gradlew -Pversion=2026.q2 publish && git tag release/2026.q2
  22. deebosnapshots are satans tool against humanity, along with nodejs and xml
  23. dreamrealthat's just about my process, too, although I don't use gradle
  24. dreamrealand I have no problem with either nodejs nor XML
  25. dreamrealmost people who resent XML resent its verbosity and don't know it
  26. dreamrealnot all, but most
  27. Candledeebo: Depending on a SNAPSHOT in production, that is certainly satanic!
  28. deebowell hopefully noone does that, and no doubt the situation has improved but i just hated trying to get an actual update to a dependency as a snapshot to work
  29. CandleAlso, switching to use Gradle is a pretty much a non-starter due to the size of the team and quantity of code.
  30. deebodon't repositories require unique snapshots these days anyhow? they append like a timestamp, :1.0-SNAPSHOT ~ :1.0-SNAPSHOT-2026041309101213.999 or something
  31. deeboremember dealing with that when managing a nexus repo and older maven versions that tried to overwrite snapshots or something
  32. CandleOnly if you push that snapshot to a repo.
  33. Parabtw, never do fast-forwarding merges
  34. ParaThey break actual history by removing merge points, and merge points are semantically important
  35. dreamrealI just don't use VCS, makes management a lot easier
  36. Paramanage, damage, same difference
  37. cheeserdeebo: they've always had timestamps on snapshots.
  38. dreamrealhttps://bytecode.news/posts/2026/04/bring-back-idiomatic-design
  39. GreenResponsenice piece, I think there’re too many people using computer technology, some only to “fit in” because of FOMO, they absorb solutions given them by AI because they are incapable of make them themselves
  40. Swayzeyeah but then you need to have teh right person for the rigth job
  41. Swayzeif its a personal project it doesnt matter what you're doing but in a professional setting there are ways to ensure you have the appropriate person in charge of architecting solutions and making decisions
  42. GreenResponseI was thinking more about end-users than developers tho
  43. jreicherPara FF merges remove merge points???
  44. kcomhnall~jep 527
  45. javabot'JEP 527: Post-Quantum Hybrid Key Exchange for TLS 1.3' can be found at http://openjdk.java.net/jeps/527
  46. nevetJEP 527: Post-Quantum Hybrid Key Exchange for TLS 1.3