assertEquals(1, ((List<?>) pom.getValue("developers")).size());
        assertEquals("dev", pom.getValue("developers[1]/id"));
        assertEquals("project-developer", pom.getValue("developers[1]/name"));
        assertEquals("developer@", pom.getValue("developers[1]/email"));
        assertEquals("https://developer", pom.getValue("developers[1]/url"));
        assertEquals("developer", pom.getValue("developers[1]/organization"));

    }

    public void setDevelopers(List<Developer> developers) {
        getModel().setDevelopers(developers);
    }

    public List<Developer> getDevelopers() {
        return getModel().getDevelopers();
    }

    public void addDeveloper(Developer developer) {
        getModel().addDeveloper(developer);
    }

    public void setContributors(List<Contributor> contributors) {

build:cuda_clang --config=cuda
# Enable TensorRT optimizations https://developer.nvidia.com/tensorrt
build:cuda_clang --config=tensorrt
build:cuda_clang --action_env=TF_CUDA_CLANG="1"
build:cuda_clang --@local_config_cuda//:cuda_compiler=clang
# Select supported compute capabilities (supported graphics cards).
# This is the same as the official TensorFlow builds.
# See https://developer.nvidia.com/cuda-gpus#compute

	// Every platform checks the API on every GOOS/GOARCH/CGO_ENABLED combination anyway,
	// so we really only need to run this check once anywhere to get adequate coverage.
	// To help developers avoid trybot-only failures, we try to run on typical developer machines
	// which is darwin,linux,windows/amd64 and darwin/arm64.
	//
	// The same logic applies to the release notes that correspond to each api/next file.

    We don't expect this to be a problem in practice!

 *  **OkHttp has an experimental new API for tracking metrics.** The new
    `EventListener` API is designed to help developers monitor HTTP requests'
    size and duration. This feature is an unstable preview: the API is subject
    to change, and the implementation is incomplete. This is a big new API we
    are eager for feedback.

				})),
			}),
			expectCost: map[string]int64{
				// we do not expose a stable type for the self variable, even when it is an object that CEL
				// considers a named type. The only operation developers should be able to perform on the type is
				// equality checking.
				"type(self) == type(self)":                                     5,
				"type(self.nestedInMap['k1']) == type(self.nestedInMap['k2'])": 9,
			},

This will help you while iterating on the required changes and prevent future regressions.

Roll it out to your team::
Once you have your developer workflow working, for example running tests from the IDE, you can consider enabling it for your team.
A faster turnaround when changing code and running tests could be worth it.
You'll probably want to do this as an opt-in first.

    ask_cuda_compute_capabilities = (
        'Please specify a list of comma-separated CUDA compute capabilities '
        'you want to build with.\nYou can find the compute capability of your '
        'device at: https://developer.nvidia.com/cuda-gpus. Each capability '
        'can be specified as "x.y" or "compute_xy" to include both virtual and'
        ' binary GPU code, or as "sm_xy" to only include the binary '

[CAUTION]
====
You can not change the destination path of empty directories with this technique.
You can learn more in https://github.com/gradle/gradle/issues/2940[this issue].
====

If you're a Java developer wondering why there is no `jarTree()` method, that's because `zipTree()` works perfectly well for JARs, WARs, and EARs.

[[sec:creating_uber_jar_example]]
== Creating "uber" or "fat" JARs

     * {@code -rf :artifactId} will pick up the first module which matches, but when multiple modules in the reactor
     *   have the same artifactId, effective failed module might be later in build reactor.
     * This means that developer will either have to type groupId or wait for build execution of all modules which
     *   were fine, but they are still before one which reported errors.