return max(momentAvailable - nowMicros, 0);
  }

  /**
   * Returns the earliest time that permits are available (with one caveat).
   *
   * @return the time that permits are available, or, if permits are available immediately, an
   *     arbitrary past or present time
   */
  abstract long queryEarliestAvailable(long nowMicros);

  /**

    * So, by using FastAPI you are saving development time, bugs, lines of code, and you would probably get the same performance (or better) you would if you didn't use it (as you would have to implement it all in your code).

# Note that we must use nightly here as these are used in nightly jobs
# For release jobs, we will pin these on the release branch
keras-nightly ~= 3.0.0.dev
tb-nightly ~= 2.18.0.a
# Test dependencies
grpcio >= 1.24.3, < 2.0
portpicker == 1.6.0
scipy == 1.11.3
requests >= 2.31.0
packaging==23.2
setuptools==70.0.0
jax==0.4.7
# The dependencies below are needed for TF wheel testing.
tensorflow-io-gcs-filesystem==0.37.1

		// we should figure out if dataDirs are also missing > dataBlocks.
		dataBlocks := (len(metaArr) + 1) / 2
		if notFoundPartsErrs > dataBlocks {
			// Not using parity to ensure that we do not delete
			// any valid content, if any is recoverable. But if
			// notFoundDataDirs are already greater than the data
			// blocks all bets are off and it is safe to purge.
			//

   * <p>For the case of {@link RoundingMode#HALF_EVEN}, this implementation uses the IEEE 754
   * default rounding mode: if the two nearest representable values are equally near, the one with
   * the least significant bit zero is chosen. (In such cases, both of the nearest representable
   * values are even integers; this method returns the one that is a multiple of a greater power of
   * two.)
   *

       safe, it needs to re-run.

       - This is unfortunate, because actually it would be safe to declare
       the task as up-to-date, because these two files, which are based on
       the generated index.xml, are outputs, not inputs. We can be sure of
       this because they are deleted in the @BeforeEach method of the
       OsgiTest test class.

       - To enable the benefit of incremental builds, we can ask Gradle

 */
@ElementTypesAreNonnullByDefault
abstract class AbstractDirectedNetworkConnections<N, E> implements NetworkConnections<N, E> {
  /** Keys are edges incoming to the origin node, values are the source node. */
  final Map<E, N> inEdgeMap;

  /** Keys are edges outgoing from the origin node, values are the target node. */
  final Map<E, N> outEdgeMap;

  private int selfLoopCount;

### Server-Side Encryption with client-provided Keys

 * maintained with a doubly linked list through the entries. All optional operations (put and
 * remove) are supported. Null keys and values are supported.
 *
 * <p>{@code containsKey(k)}, {@code put(k, v)} and {@code remove(k)} are all (expected and
 * amortized) constant time operations. Expected in the hashtable sense (depends on the hash

But you don't have to worry about them either, incoming dicts are converted automatically and your output is converted automatically to JSON too.

## Bodies of arbitrary `dict`s

You can also declare a body as a `dict` with keys of some type and values of some other type.

This way, you don't have to know beforehand what the valid field/attribute names are (as would be the case with Pydantic models).