We use *Specification* and *Discovery* documents stored in Google Drive, but they present some downsides:

* They are rarely updated after creation and initial review, and then become hard to follow, especially after important decisions are made

import java.lang.annotation.Retention;
import java.lang.annotation.RetentionPolicy;
import java.lang.reflect.Method;
import java.util.Locale;
import junit.framework.TestCase;

/**
 * Since annotations have some reusability issues that force copy and paste all over the place, it's
 * worth having a test to ensure that all our Feature enums have their annotations correctly set up.
 *
 * @author George van den Driessche
 */

 * second invocation, so implementors have made various choices, including:
 *
 * <ul>
 *   <li>returning the same iterator again
 *   <li>throwing an exception of some kind
 *   <li>or the usual, <i>robust</i> behavior, which all known {@link Collection} implementations
 *       have, of returning a new, independent iterator
 * </ul>
 *

    return indexOf(target) >= 0;
  }

  @GwtIncompatible // super.subListUnchecked does not exist; inherited subList is valid if slow
  /*
   * TODO(cpovirk): if we start to override indexOf/lastIndexOf under GWT, we'll want some way to
   * override subList to return an ImmutableSortedAsList for better performance. Right now, I'm not
   * sure there's any performance hit from our failure to override subListUnchecked under GWT
   */
  @Override

            println("CACHE_MISS in task $taskPath")
            cacheMiss.set(true)
        }
    }
}


/**
 *  We monitor some tasks in non-seed builds. If a task executed in a non-seed build, we think it as "CACHE_MISS".
 */
fun isCacheMissMonitoredTask(task: Task) = task.isCompileCacheMissMonitoredTask() || task.project.isAsciidoctorCacheMissTask()

		t.Run(tt.name, func(t *testing.T) {
			cw := &ConfigWriter{}
			cd, _ := os.ReadFile(tt.inputFile)
			err := cw.Prime(cd)
			if cw.ztunnelDump == nil {
				if tt.wantConfigs != 0 {
					t.Errorf("wanted some configs loaded but config dump was nil")
				}
			} else if len(cw.ztunnelDump.Workloads) != tt.wantConfigs {
				t.Errorf("wanted %v configs loaded in got %v", tt.wantConfigs, len(cw.ztunnelDump.Workloads))
			}

 * JdkFutureAdapters}. However, when possible, it is more efficient and reliable to create a {@code
 * ListenableFuture} directly.
 *
 * @author Sven Mawson
 * @author Nishant Thakkar
 * @since 1.0
 */
/*
 * Some of the annotations below were added after we released our separate
 * com.google.guava:listenablefuture:1.0 artifact. (For more on that artifact, see

 * JdkFutureAdapters}. However, when possible, it is more efficient and reliable to create a {@code
 * ListenableFuture} directly.
 *
 * @author Sven Mawson
 * @author Nishant Thakkar
 * @since 1.0
 */
/*
 * Some of the annotations below were added after we released our separate
 * com.google.guava:listenablefuture:1.0 artifact. (For more on that artifact, see

    // Check that this thread has been marked as interrupted again now that the thread has been
    // returned by SequentialExecutor. Clear the bit while checking so that the test doesn't hose
    // JUnit or some other test case.
    assertThat(Thread.interrupted()).isTrue();
  }

  public void testInterrupt_doesNotInterruptSubsequentTask() throws Exception {
    // Run a task that interrupts its thread (i.e. this thread).

For discovery, the JWT token will be read directly from a file and sent as is. For CA, this logic is a bit more complex,
as the support for external CAs is more mature than external discovery servers. This supports some additional
configuration, a `CredentialFetcher` which allows fetching a token from places other than a file (for example, a local