| 26 May 2026 |
 dish [Fox/It/She] | In reply to @britter:yatrix.org
@pyrox:pyrox.dev is this something you could merge? is there anything that needs to be done with the pending reviews | 14:13:15 |
 Benedikt Ritter (britter) | Yes! Actually @msgilligan:matrix.org raised a good point. So I think we're going back to the drawing board. You could voice your opinion on what he said. It's more of a Nix packaging question than a maven question. | 14:40:37 |
 msgilligan | I think we can keep the same approach and just add the JDK and any other parameters that need to be overridden.
I also think maven_4 should default to JDK 25.
| 17:17:23 |
| msgilligan | And Iām hoping we can eventually get from-source builds of Maven 3 and 4. | 18:01:16 |
| 27 May 2026 |
| Benedikt Ritter (britter) | I'll look into improving the maven 4 PR later today or tomorrow. | 16:42:24 |
| msgilligan | My suggestion is to make minimal changes to the PR to make the JDK overridable, rather than switch to a new approach. | 16:46:01 |
| Benedikt Ritter (britter) | I'd also like to discuss what we want to do with the maven 3 upgrade. I see two options:
- we update the hashes of all packages that break due to the hash issue.
- we implement @msgilligan:matrix.org idea of putting the default plugin versions into the FOD.
| 16:47:06 |
| dish [Fox/It/She] | In reply to @msgilligan:matrix.org
My suggestion is to make minimal changes to the PR to make the JDK overridable, rather than switch to a new approach. I'm pretty sure that it is overridable because of function chaining, but I'm double checking that now | 16:47:19 |
| dish [Fox/It/She] | since maven4 is just an .overrideAttrs of maven3, they should be able to just be chained again for overrides | 16:47:50 |
| dish [Fox/It/She] | *additional overrides | 16:48:16 |
| dish [Fox/It/She] | hmm, maybe not. nix is weird š
| 16:51:16 |
| Benedikt Ritter (britter) | @msgilligan:matrix.org do you have time to explore your idea? If not I can look into it at the beginning of next week. | 17:00:31 |
| msgilligan | Not before next week, so maybe we can work on it together starting then? | 17:01:27 |
| Benedikt Ritter (britter) | Sounds great! | 17:18:54 |
| 28 May 2026 |
| msgilligan |
Cannot deploy artifacts when Maven is in offline mode
Why not, Maven? I'm deploying to a local directory.
| 21:22:58 |
 Infinidoge š³ļøāā§ļø | At what point do we reach "Reimplement Maven's protocol as a dummy local directory that pretends to be online" territory | 21:23:59 |
| msgilligan | This point! I feel I'm kinda at that point... | 21:24:33 |
| msgilligan | But I think we might be able to get the bare minimum without going quite so far. | 21:25:21 |
| msgilligan | I've been experimenting with using mvn deploy (using a local directory as the destination) in derivations using buildMavenPackage. In order to do this I need two new parameters to buildMavenPackage: mvnGoal and mvnOffline (which default to package and true.)
These two parameters (and sometimes the existing mvnParameters) are needed to use ``buildMavenPackagewith thedeploy` goal.
I made a PR that adds the two parameters: https://github.com/NixOS/nixpkgs/pull/525382
| 23:25:59 |
| msgilligan | The idea is to generate output in Maven repo layout for publishing to Maven Central or for using in other derivations. | 23:33:33 |
| 29 May 2026 |
| Benedikt Ritter (britter) | Why does running deploy to a local dir require online mode? š¤ | 05:46:43 |
| Infinidoge š³ļøāā§ļø | Maven is a bit silly | 05:47:56 |
| msgilligan | I feel like we should give some input and possibly even a few PRs upstream to Maven (and/or Ant!) | 15:10:37 |
| msgilligan | Can someone please review or merge my simple bug-fix for jextract on Darwin? | 16:03:22 |
| 31 May 2026 |
| msgilligan | Merged! Now we can remove the unmaintained/obsolete jextract-21: https://github.com/NixOS/nixpkgs/pull/519958 | 16:17:51 |
| dish [Fox/It/She] | cheers, merged | 18:07:38 |
| msgilligan | Thanks, but I did a GitHub search after my last comment and found a project (chisel) using jextract-21. So maybe we shouldn't have removed it and give a warning instead. (It looks like swapping the JDK 25 version would break them, so the warnAlias would not be good for them.) | 18:11:02 |
| msgilligan | * Thanks, but I did a GitHub search after my last comment and found a project (chisel) using jextract-21. So maybe we shouldn't have removed it and should give a warning instead. (It looks like swapping the JDK 25 version would break them, so the warnAlias would not be good for them.) | 18:14:43 |
| msgilligan | I can make a PR to revert the change and add a warning upon use. Should I do that? | 18:17:25 |
| msgilligan | People really shouldn't be using jextract-21 though. It's an early-access build supporting a preview feature (FFM/Panama) of the JDK that didn't finalize until JDK 22 and is only supported by OpenJDK in version 25 and later. So we probably did the right thing. | 18:36:16 |
