<3> When submitting the unit of work, specify the unit of work implementation, in this case `GenerateMD5`, and configure its parameters.

At this point, you should be able to rerun your task:

[listing]
----
$ gradle clean md5

include::{snippetsPath}/workerApi/md5NoIsolation/tests/md5Task1.out[]
----

   * than this method, and this method in turn generally performs better than copying the list and
   * calling {@link Collections#sort(List)}.
   */
  // TODO(kevinb): rerun benchmarks including new options
  @SuppressWarnings("nullness") // TODO: b/316358623 - Remove after checker fix.
  public <E extends T> List<E> sortedCopy(Iterable<E> elements) {

[cols="~,~,~"]
|===
|Mechanism |Information |Example

|<<command_line_interface#command_line_interface,Command line interface>>
|Flags that configure build behavior and Gradle features
|`--rerun`

|<<sec:project_properties,Project properties>>
|Properties specific to your Gradle project
|`TestFilter::isFailOnNoMatchingTests=false`

|<<#sec:gradle_system_properties,System properties>>

The action deletes a target file if the corresponding input file has been removed.

If, for some reason, the task is executed non-incrementally (by running with `--rerun-tasks`, for example), all files are reported as `ADDED`, irrespective of the previous state.
In this case, Gradle automatically removes the previous outputs, so the incremental task must only process the given files.

 *       than MEDIUM. And so on. These constants are set to conservative values, but even so, if
 *       there is ever any doubt, they can all be increased in one spot to rerun tests on slower
 *       platforms.
 *   <li>All threads generated must be joined inside each test case method (or {@code fail} to do
 *       so) before returning from the method. The {@code joinPool} method can be used to do this

 *       than MEDIUM. And so on. These constants are set to conservative values, but even so, if
 *       there is ever any doubt, they can all be increased in one spot to rerun tests on slower
 *       platforms.
 *   <li>All threads generated must be joined inside each test case method (or {@code fail} to do
 *       so) before returning from the method. The {@code joinPool} method can be used to do this

        succeeds ':outgoingVariants'
        reportsCompleteAbsenceOfResolvableVariants()
    }

    def "if only legacy configuration present, and --all not specified, task produces empty report and prompts for rerun"() {
        given:
        buildFile << """
            configurations.create("legacy") {
                description = "My legacy configuration"
                assert canBeResolved
                assert canBeConsumed

	tg.grepStderr(`([\\/]link|gccgo).*t3\.test`, "did not relink t3_test")
	tg.grepStderr(`t3\.test.*-test.short`, "did not rerun t3_test")
	tg.grepStdoutNot(`ok  \tt/t3\t\(cached\)`, "reported cached t3_test result")

	// t4 imports p2, but p2 did not change, so t4 should be relinked, not recompiled,
	// and not rerun.
	tg.grepStderrNot(`([\\/]compile|gccgo).*t4_test.go`, "incorrectly recompiled t4")

	// much larger, so the cache-footprint-to-utility ratio
	// of binaries is much lower for binaries.
	// Not caching the link step also makes sure that repeated "go run" at least
	// always rerun the linker, so that they don't get too fast.
	// (We don't want people thinking go is a scripting language.)
	// Note also that if we start caching binaries, then we will
	// copy the binaries out of the cache to run them, and then

                    }
                }
            }
        """
    }

    def "incremental transform is rerun when output is #action between builds"() {
        given:
        buildFile << declareAttributes() << multiProjectWithJarSizeTransform(incremental: true) << withClassesSizeTransform()