* elements directly and null for empty positions, we store indices of the keys in the hash table,
   * and ABSENT for empty positions.  We then look up the keys in alternatingKeysAndValues.
   *
   * (The index actually stored is the index of the key in alternatingKeysAndValues, which is
   * double the index of the entry in entrySet.asList.)
   *

   * the spliterator implementation.
   */
  // @IgnoreJRERequirement is not necessary because this compiles down to a constant.
  // (which is fortunate because Animal Sniffer doesn't look for @IgnoreJRERequirement on fields)
  static final int SPLITERATOR_CHARACTERISTICS =
      Spliterator.IMMUTABLE | Spliterator.NONNULL | Spliterator.ORDERED;

  ImmutableCollection() {}

#### Preview the result

As already mentioned above, you can use the `./scripts/docs.py` with the `live` command to preview the results (or `mkdocs serve`).

Once you are done, you can also test it all as it would look online, including all the other languages.

To do that, first build all the docs:

<div class="termy">

```console
// Use the command "build-all", this will take a bit
$ python ./scripts/docs.py build-all

            eventTimings.put(type, new AtomicLong(0));
        }

        // Start async logging processor
        startAsyncProcessor();

        // Register shutdown hook
        Runtime.getRuntime().addShutdownHook(new Thread(this::shutdown));
    }

    private void startAsyncProcessor() {
        logExecutor.submit(() -> {
            while (!shutdown) {
                try {

        });
    }

    /**
     * Retrieves a parameter value from the data store parameters map.
     *
     * @param paramMap the parameter map to search
     * @param key the parameter key to look up
     * @param defaultValue the default value to return if key is not found
     * @return the parameter value as a string, or the default value if not found
     */

	// Fast path for ASCII: look for the first ASCII non-space byte.
	for lo, c := range s {
		if c >= utf8.RuneSelf {
			// If we run into a non-ASCII byte, fall back to the
			// slower unicode-aware method on the remaining bytes.
			return TrimFunc(s[lo:], unicode.IsSpace)
		}
		if asciiSpace[c] != 0 {
			continue
		}
		s = s[lo:]
		// Now look for the first ASCII non-space byte from the end.

				c.healBuckets(ctx, objAPI)   // heal buckets subsequently

				took := time.Since(refreshStart).Seconds()
				if took > maxRefreshDurationSecondsForLog {
					// Log if we took a lot of time.
					logger.Info("Site replication healing refresh took %.2fs", took)
				}

				// wait for 200 millisecond, if we are experience lot of I/O

   *
   * (To "go out of its way," the method tracks a `set` bit so that it can distinguish "the final
   * split has a last element of null, so throw NPE" from "the final split was empty, so look for an
   * element in the prior one.")
   */
  public static <T> java.util.Optional<T> findLast(Stream<T> stream) {
    final class OptionalState {
      boolean set = false;
      @Nullable T value = null;

The way you use it is very simple. For example, to do a `GET` request, you would write:

```Python
response = requests.get("http://example.com/some/url")
```

The FastAPI counterpart API *path operation* could look like:

```Python hl_lines="1"
@app.get("/some/url")
def read_url():
    return {"message": "Hello World"}
```

See the similarities in `requests.get(...)` and `@app.get(...)`.

     *
     * All that said, don't forget that everyone should call converter.convert() instead of
     * converter.apply(), anyway. If clients use only converter.convert(), then their nullness
     * checkers are unlikely to ever look at the annotations on this declaration.
     *
     * Historical note: At one point, we'd declared this method as accepting and returning nullable
     * values. For details on that, see earlier revisions of this file.
     */