NixOS JVM | 119 Members | |
| 26 Servers |
| Sender | Message | Time |
|---|---|---|
| 16 Jul 2025 | ||
| This is totally uncharted territory for me. | 06:39:38 | |
| Wondering if I should create an issue and/or a draft PR. I think the issue might be a good place to discuss requirements and features in a way that can get people involved. | 06:41:55 | |
| I was thinking it might make sense to define some kind of JDK specific metadata that could live in the root directory of each JDK package-set. | 06:42:47 | |
For example, I currently have definitions of which versions are supposed to be supported by each JDK and whether their is a default (e.g. corretto doesn't exist, but zulu defaults to zulu21. | 06:45:41 | |
* For example, I currently have definitions of which versions are supposed to be supported by each JDK and whether there is a default (e.g. corretto doesn't exist, but zulu defaults to zulu21. | 06:46:00 | |
* For example, I currently have definitions of which versions are supposed to be supported by each JDK and whether there is a default (e.g. corretto doesn't exist, but zulu defaults to zulu21.) | 06:46:12 | |
| I'd rather have a list of JDK root paths and query them for metadata (perhaps with defaults for package-sets that haven't implemented the metadata yet) | 06:48:05 | |
I don't know what this means. | 06:48:31 | |
Currently jdk-dashboard.nix and jdk-dashboard.sh are in nixpkgs/maintainers/scripts but I'm not sure that is the right place. | 06:49:59 | |
In reply to @msgilligan:matrix.orgGenerally on sets of packages, there is a specific attr (I forget which exactly) that is set to true to flag that functions should look through the attribute set for derivations | 06:51:18 | |
| e.g.
and
| 06:51:34 | |
| I believe there should be more info attached to lib.recurseIntoAttrs | 06:51:37 | |
and I'm including - and _ as a prefix in some versions as there is currently no consistent mechanism for concatenating name to version. e.g. zulu21, but temurin-bin-21 and graalvmPackages.graalvm-oracle_17 | 06:54:17 | |
| We should probably look at what SDKMAN! does: https://sdkman.io/jdks | 07:18:07 | |
| Note that they are using Temurin as their default JDK. | 07:18:27 | |
| And of course, what I would like to be able to do is to view the versions of each Nixpkgs JDK and display it next to the latest version available for that JDK. We have updaters for some of the JDKs that will pull the latest versions from a server. | 07:42:33 | |
| It looks like there is an API that can give that info for many distributions: https://github.com/foojayio/discoapi | 07:43:14 | |
| I opened an issue for the "JDK dashboard Utility": https://github.com/NixOS/nixpkgs/issues/425860 | 18:10:28 | |
| I also set some of the metadata tags on this issue, let me know if I did that wrong. | 18:27:11 | |
| 17 Jul 2025 | ||
| they don't build their own JDKs, so this isn't really a comparable situation | 13:06:52 | |
| we already use Temurin as the "vanilla" binary build to bootstrap our OpenJDK source build, e.g. | 13:07:08 | |
| Temurin is "just" a vanilla-aiming OpenJDK build | 13:07:54 | |
| a distribution in itself | 13:08:01 | |
| it is standard for package collections, as distributors themselves, to compile their own OpenJDKs | 13:08:24 | |
| there is really not much magic in the build glue https://github.com/adoptium/temurin-build | 13:09:08 | |
| the value-add of Temurin is in providing the infra and testing and brand name, not the build logic | 13:09:37 | |
| which is why a separate Temurin source build makes no sense (either the Temurin build scripts are a good fit for us and we can use them in our OpenJDK source builds, or they would make the derivation less maintainable or less flexible and so we should keep driving the OpenJDK build on our own) | 13:10:35 | |
| the Temurin product is builds | 13:11:07 | |
| not sure if they even allow their trademark to be used for builds of OpenJDK that happen to use the same build scripts | 13:11:20 | |
| feel like the proliferation of near-identical OpenJDK distributions has led to a lot of confusion about what they actually are… | 13:12:19 | |