| 1 Jun 2026 |
 msgilligan | This will also help (especially on Darwin) users migrate away from jextract-21 | 02:49:28 |
| msgilligan | * For jextract, I opened PRs to backport the darwin fix and 0-unstable-2025-11-12 update.
To release-25.11:
- https://github.com/NixOS/nixpkgs/pull/526579 - merged ✅
- https://github.com/NixOS/nixpkgs/pull/526580
And to release-26.05:
- https://github.com/NixOS/nixpkgs/pull/526581
- https://github.com/NixOS/nixpkgs/pull/526587
| 03:06:10 |
| msgilligan | * For jextract, I opened PRs to backport the darwin fix and 0-unstable-2025-11-12 update.
To release-25.11:
- https://github.com/NixOS/nixpkgs/pull/526579 - merged ✅
- https://github.com/NixOS/nixpkgs/pull/526580
And to release-26.05:
- https://github.com/NixOS/nixpkgs/pull/526581 - merged ✅
- https://github.com/NixOS/nixpkgs/pull/526587
| 03:23:42 |
 Benedikt Ritter (britter) | I have updated the maven_4 PR. It's now a proper derivation and therefore duplicates parts of maven. I think this is better because people can more easily overrideAttrs.
@msgilligan:matrix.org I haven't set jdk25 as the default. Why should we do that if maven doesnt require it? People can still customize if they want. | 10:10:37 |
| Benedikt Ritter (britter) | I also have the change for default maven plugin implemented that @msgilligan:matrix.org suggested. Will push thst when I'm back at my computer. | 13:08:03 |
| msgilligan | * For jextract, I opened PRs to backport the darwin fix and 0-unstable-2025-11-12 update.
To release-25.11:
- https://github.com/NixOS/nixpkgs/pull/526579 - merged ✅
- https://github.com/NixOS/nixpkgs/pull/526580
And to release-26.05:
- https://github.com/NixOS/nixpkgs/pull/526581 - merged ✅
- https://github.com/NixOS/nixpkgs/pull/526587 - merged ✅
| 20:44:03 |
| msgilligan |
I haven't set jdk25 as the default. Why should we do that if maven doesn't require it?
- maven_4 is a new package so nothing is depending on it yet. It seems to me it should start with the latest JDK to encourage dependent packages to use JDK 25
- JDK 25 can build projects for all earlier JDKs plus JDK 25. This means users won't have to use toolchains or env variables or other mechanisms to use JDK 25
- JDK 25 is the fastest LTS JDK
Downsides:
- Some packages may have build issues using JDK 25. (If so they can workaround with JDK override and at least they become aware of the issue)
- It's possible a system, home manager, or dev shell configuration that is using default (typically JDK 21, presently) dependencies for other Java packages might end up pulling in an additional JDK for Maven. (But on the other hand as we begin migrating packages to JDK 25, eventually more packages will depend on JDK 25 than JDK 21)
| 21:41:35 |
 raboof | discovering which packages don't build with jdk25 can also be an advantage, so you can start the conversation with upstream to fix that 😊 | 21:44:17 |
| 2 Jun 2026 |
| Benedikt Ritter (britter) | Fair enough. I would typically go the other way around and set the minimal JDK that is required by maven 4. But I can update the package to use JDK 25. | 05:32:57 |
| Benedikt Ritter (britter) | Here's the fix for changibg mvnHashes following @msgilligan:matrix.org idea: https://github.com/NixOS/nixpkgs/pull/527061
Lets get this merged and then rebase the maven version bump on top of it. | 09:13:56 |
|  jlle changed their display name from jlle to Jose Luis. | 09:39:49 |
| Benedikt Ritter (britter) | I've also updated the maven_4 PR to use jdk25_headless. | 09:54:26 |
| 3 Jun 2026 |
| Benedikt Ritter (britter) | @pyrox:pyrox.dev is this something you could review? | 11:44:54 |
 dish [Fox/It/She] | reviewed | 16:09:15 |
| 4 Jun 2026 |
|  Echo changed their profile picture. | 04:23:43 |
| msgilligan | I took a first look. Overall it looks awesome. I have a few minor requests and I really should do a more thorough second review. I'm currently trying to run nixpkgs-review on it. | 06:51:04 |
| msgilligan | * I took a first look. Overall it looks awesome. I have a few minor requests and I really should do a more thorough second review. I'm currently trying to run ran nixpkgs-review on it. (Update ❌ 2 failures ✅ 74 packages built) | 07:18:35 |
| Benedikt Ritter (britter) | I'll have a look at the failures later today. I also want to get the community builder connectes to my machine so I can more easily run these massive rebuilds. | 09:05:50 |
| Benedikt Ritter (britter) | Didn't get around to doing this... sorry! | 20:02:44 |
| Benedikt Ritter (britter) | Just a heads up, I'm traveling the next two weeks and am not sure whether I will take my computer. So my maven PRs will have to wait until I'm back. | 20:03:29 |
| 7 Jun 2026 |
| Benedikt Ritter (britter) | Can somebody merge this for me? It's just another rec -> finalAttrs PR: https://github.com/NixOS/nixpkgs/pull/527149 | 07:05:35 |
| msgilligan | Trying @NixOS/nixpkgs-merge-bot merge... | 07:15:22 |
 Infinidoge 🏳️⚧️ | Hey look fancy button | 07:21:50 |
| Infinidoge 🏳️⚧️ | Should auto merge when checks pass | 07:22:03 |
| msgilligan | Sometimes it works (and when it doesn’t it gives a checklist explaining why) | 16:11:14 |
| 18 Jun 2026 |
| Benedikt Ritter (britter) | I'm back from vacation and will pick up the maven PR again that I was working on. maven_4 has landed in the meantime. | 05:37:41 |
|  whispers [& it/fae] changed their display name from whispers [& it/fae] to meow meow. | 18:46:32 |
| whispers [& it/fae] changed their display name from meow meow to whispers [& it/fae]. | 19:12:08 |
| Benedikt Ritter (britter) | @pyrox:pyrox.dev I've updated the PR that introduces the default plugins repo. I've addressed review feedback and added two commits that handle unrelated package build failures. Not sure if that should be separate PRs. I wanted a state where nixpkgs-review passes with out having to wait for other PRs to be merged first.
I've also rebased onto main and ran into an issue with the new maven_4 package that was merged recently: unfortunately they changed completely how default plugin versions are handled inside maven. Previously it was some xml files, which made it easy to extract this data. Now it' constants in Java code which are constructed by concatination of other constants. So there's no easy grep in the java sources solution for maven_4.
I didnt want to blow the scope of the PR, so I decided to add some logic that make maven_4 behave like maven_3 did before: essentially package dependency FOD hashes need to be updated when maven_4 package is updated. I want to address this in a follow up PR.
Link: https://github.com/NixOS/nixpkgs/pull/527061
| 20:23:00 |
| 19 Jun 2026 |
| Benedikt Ritter (britter) | Why does changing maven still trigger a rebuild of authentik which is a python application? | 09:50:13 |
