ant: use upstream launcher script

[tomodachi94]

Warning

Superseded by #364094.

Let's use Ant's upstream launcher script. It handles far more edge-cases than ours.

Instead of writing our own, let's simply hard-code values into the launcher that are known in advance.

One small step towards #353512. Closes #4105, because a path to the JRE is one of the things we now explicitly set. Something similar first attempted by @linsui in #266313.

Things done

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandboxing enabled in nix.conf? (See Nix manual)
  • sandbox = relaxed
  • sandbox = true
• Tested, as applicable:
  • NixOS test(s) (look inside nixos/tests)
  • and/or package tests
  • or, for functions and "core" functionality, tests in lib/tests or pkgs/test
  • made sure NixOS tests are linked to the relevant packages
• Tested compilation of all packages that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage
• Tested basic functionality of all binary files (usually in ./result/bin/)
• 25.05 Release Notes (or backporting 24.11 and 25.05 Release notes)
  • (Package updates) Added a release notes entry if the change is major or breaking
  • (Module updates) Added a release notes entry if the change is significant
  • (Module addition) Added a release notes entry if adding a new NixOS module
• Fits CONTRIBUTING.md.

---

Add a 👍 reaction to pull requests you find important.

[TomaSajt]

Do we actually need jre in buildInputs, if we just substitute it like this?
(haven't checked)

[tomodachi94]

I don't think so. I did it because that's what the convention is, as far as I understand it; happy to remove it if that isn't the case.

[TomaSajt]

Also, will this change make it so that we don't actually need to add both jdk and ant into nativeBuildInputs when using ant?

Also also, jdk vs jre? (I'd say jdk)

[tomodachi94]

From my testing, if we use jdk, that is the case, but setting it to jdk makes it difficult to use a different version of Java without overriding ant.

[tomodachi94]

Is there a reason that we shouldn't have consumers explicitly specify a JDK?

[TomaSajt]

I'm not sure if using substituteInPlace is a good idea.
(${VAR} and $VAR mean the same thing, but only one of them is replaced).

I would probably either use wrapProgram $out/lib/ant/bin --set ANT_HOME "..." or just put export ANT_HOME="..." after the shebang in the script.

Here's the unsubstituted launcher script for easier viewing: https://gist.github.com/TomaSajt/485fc88337e768e2d065cc61eb0c4560

[tomodachi94]

I'm unsure about whether we should consider including perl and python in nativeBuildInputs so that these scripts have their shebangs patched; it potentially bloats the closure, but users might expect to be able to execute these out-of-the-box.

[FliegendeWurst]

They would go in buildInputs. But at least perl ends up in the default system PATH anyhow.

[FliegendeWurst]

LGTM.
I assume the Java version used for the wrappers is not of significance for ant users, right?

[tomodachi94]

No, it shouldn't be.

[SuperSandro2000]

The rebuild amount doesn't justify staging.

[tomodachi94]

The rebuild amount doesn't justify staging.

OfBorg's rebuild count isn't accurate. It's approximately 800 cumulative rebuilds (a few weeks ago; not much has changed since then, just a few leaf packages being dropped, unless I missed a larger drop of a non-leaf package): #357162 (comment)

[tomodachi94]

CI failures look unrelated.

[tomodachi94]

Rebased onto common ancestor of master and staging, with the hope that CI will succeed.