# Recommended: use a local+remote cache.
#
#   Bazel will cache your builds in tensorflow/build_output/cache,
#   and will also try using public build cache results to speed up
#   your builds. This usually saves a lot of time, especially when
#   re-running tests. However, note that:
#
#    - New environments like new CUDA versions, changes to manylinux,
#      compilers, etc. can cause undefined behavior such as build failures

    val extraParameters: String = "",
) {
    PER_COMMIT(
        displayName = "Performance Regression Test",
        timeout = 420,
        channel = "commits",
    ),
    PER_DAY(
        displayName = "Slow Performance Regression Test",
        timeout = 420,
        channel = "commits",
    ),
    PER_WEEK(
        displayName = "Performance Experiment",
        timeout = 420,
        channel = "experiments",

Here are the general instructions for the tasks you can perform.

Thanks a lot for your help. 🙇

## Be Nice

First of all, be nice. 😊

You probably are super nice if you were added to the team, but it's worth mentioning it. 🤓

### When Things are Difficult

    try {
      constructor = type.getConstructor();
    } catch (NoSuchMethodException e) {
      return arbitraryConstantInstanceOrNull(type);
    }
    constructor.setAccessible(true); // accessibility check is too slow
    try {
      return constructor.newInstance();
    } catch (InstantiationException | IllegalAccessException impossible) {
      throw new AssertionError(impossible);
    } catch (InvocationTargetException e) {

    List<String> list = ImmutableList.of("a", "b");
    Iterator<String> iterator = skip(list, 2).iterator();
    assertThrows(UnsupportedOperationException.class, () -> iterator.remove());
  }

  @GwtIncompatible // slow (~35s)
  public void testSkip_iterator() {
    new IteratorTester<Integer>(
        5, MODIFIABLE, newArrayList(2, 3), IteratorTester.KnownOrder.KNOWN_ORDER) {
      @Override

Without a data validation system, you would have to do all the checks by hand, in code.

These features are what Marshmallow was built to provide. It is a great library, and I have used it a lot before.

But it was created before there existed Python type hints. So, to define every <dfn title="the definition of how data should be formed">schema</dfn> you need to use specific utils and classes provided by Marshmallow.

import okio.Socket
import okio.buffer
import okio.inMemorySocketPair
import org.junit.jupiter.api.AfterEach
import org.junit.jupiter.api.BeforeEach
import org.junit.jupiter.api.Tag
import org.junit.jupiter.api.Test

@Tag("Slow")
class RealWebSocketTest {
  // NOTE: Fields are named 'client' and 'server' for cognitive simplicity. This differentiation has
  // zero effect on the behavior of the WebSocket API which is why tests are only written once

    assertTrue(enormous instanceof RegularContiguousSet);
    // We can't use reserializeAndAssert because it calls hashCode, which is enormously slow.
    ContiguousSet<Integer> enormousReserialized = reserialize(enormous);
    assertEquals(enormous, enormousReserialized);
  }

  private static final DiscreteDomain<Integer> UNBOUNDED_THROWING_DOMAIN =

         * technique that NTLMv2 specifically thwarts. A real NTLM Filter would
         * need to do a NETLOGON RPC that JCIFS will likely never implement
         * because it requires a lot of extra crypto not used by CIFS.
         */
        p.setProperty("jcifs.smb.lmCompatibility", "0");
        p.setProperty("jcifs.smb.client.useExtendedSecurity", "false");

            ArtifactRepository selected,
            ArtifactRepository localRepository)
            throws RepositoryMetadataStoreException {
        // TODO this could be a lot nicer... should really be in the snapshot transformation?
        if (metadata.isSnapshot()) {
            Metadata prevMetadata = metadata.getMetadata();