*
 * <p>Disadvantages compared to {@code long[]}:
 *
 * <ul>
 *   <li>Memory footprint has a fixed overhead (about 24 bytes per instance).
 *   <li><i>Some</i> construction use cases force the data to be copied (though several construction
 *       APIs are offered that don't).
 *   <li>Can't be passed directly to methods that expect {@code long[]} (though the most common
 *       utilities do have replacements here).

 *
 * <p>Disadvantages compared to {@code int[]}:
 *
 * <ul>
 *   <li>Memory footprint has a fixed overhead (about 24 bytes per instance).
 *   <li><i>Some</i> construction use cases force the data to be copied (though several construction
 *       APIs are offered that don't).
 *   <li>Can't be passed directly to methods that expect {@code int[]} (though the most common
 *       utilities do have replacements here).

  /**
   * Not checking the CA bit created a vulnerability in old OkHttp releases. It is exploited by
   * triggering different chains to be discovered by the TLS engine and our chain cleaner. In this
   * attack there's several different chains.
   *
   *
   * The victim's gets a non-CA certificate signed by a CA, and pins the CA root and/or
   * intermediate. This is business as usual.
   *
   * ```

 *   <li>To wrap nullable references for storage in a collection that does not support {@code null}
 *       (though there are <a
 *       href="https://github.com/google/guava/wiki/LivingWithNullHostileCollections">several other
 *       approaches to this</a> that should be considered first)
 * </ul>
 *
 * <p>A common alternative to using this class is to find or create a suitable <a

      suite.addTest(
          builder
              .withFeatures(
                  CollectionSize.ZERO,
                  CollectionSize.ONE,
                  CollectionSize.SEVERAL,
                  CollectionFeature.ALLOWS_NULL_QUERIES,
                  CollectionFeature.RESTRICTS_ELEMENTS,
                  CollectionFeature.KNOWN_ORDER,

      suite.addTest(
          builder
              .withFeatures(
                  CollectionSize.ZERO,
                  CollectionSize.ONE,
                  CollectionSize.SEVERAL,
                  CollectionFeature.ALLOWS_NULL_QUERIES,
                  CollectionFeature.RESTRICTS_ELEMENTS,
                  CollectionFeature.KNOWN_ORDER,

    url: HttpUrl,
  ) {
  }

  /**
   * Invoked after proxy selection.
   *
   * Note that the list of proxies is never null, but it may be a list containing
   * only [Proxy.NO_PROXY]. This comes up in several situations:
   *
   * * If neither a proxy nor proxy selector is configured.
   * * If the proxy is configured explicitly as [Proxy.NO_PROXY].
   * * If the proxy selector returns only [Proxy.NO_PROXY].

           * modified concurrently), or it might have been only an estimate to begin with (for a
           * collection like ConcurrentHashMap, which sums up several counters that may not be in
           * sync with one another). We accommodate that by resizing as necessary.
           */
          toJoin = Arrays.copyOf(toJoin, expandedCapacity(toJoin.length, toJoin.length + 1));

    typically will live in
    [tensorflow/addons](https://github.com/tensorflow/addons) to get some
    airtime before a decision is made regarding whether they are to be migrated
    to the core.
*   As every PR requires several CPU/GPU hours of CI testing, we discourage
    submitting PRs to fix one typo, one warning,etc. We recommend fixing the
    same issue at the file level at least (e.g.: fix all typos in a file, fix

 *       typically implemented by opening a reader using one of the methods in the first category,
 *       doing something and finally closing the reader that was opened.
 * </ul>
 *
 * <p>Several methods in this class, such as {@link #readLines()}, break the contents of the source
 * into lines. Like {@link BufferedReader}, these methods break lines on any of {@code \n}, {@code