Rust tests

Not thaaat experimental

Running tests has been supported by crate2nix for a while now but is still marked as experimental. Don’t let that scare you too much: we intend to support tests in the future but the interface might slightly change.

Running rust tests

There is some experimental support for running tests of your rust crates. All of the crates in the workspace will have their tests executed. When enabling test execution (runTests = true;), failing tests will make the whole build fail unless you explicitly disable this via test hooks: see the section below.

      let cargo_nix = callPackage ./Cargo.nix {};
          crate2nix = cargo_nix.rootCrate.build.override {
     runTests = true;
     testInputs = [ pkgs.cowsay ];
   };
      in ...

testInputs is optional and allows passing inputs to the test execution that should be in scope. Defaults to an empty list and is ignored when runTests equals false.

Custom pre/post test hooks

Want to customize your test execution? Use the testPreRun and testPostRun crate attributes(next to runTests in the example above). crate2nix executes the bash snippets in testPreRun and testPostRun directly before and after the actual test command, and in the same shell. Some example use-cases include:

• Setting some environment variable that is needed for the test.

• Setting (and then unsetting) the bash set +e option to not fail the derivation build even if a test fails. This is quite useful if your tests are not flaky and you want to cache failures.