connection.setFixedLengthStreamingMode(1024 * 1024 * 1024) // 1 GB
    connection.connect()
    val out = connection.outputStream
    val data = ByteArray(1024 * 1024)
    var i = 0
    while (i < 1024) {
      try {
        out.write(data)
        out.flush()
        if (i == 513) {
          // pause slightly after halfway to make result more predictable
          Thread.sleep(100)
        }

    connection.setFixedLengthStreamingMode(1024 * 1024 * 1024) // 1 GB
    connection.connect()
    val out = connection.outputStream
    val data = ByteArray(1024 * 1024)
    var i = 0
    while (i < 1024) {
      try {
        out!!.write(data)
        out.flush()
        if (i == 513) {
          // pause slightly after half way to make result more predictable
          Thread.sleep(100)
        }

  @SuppressWarnings("unchecked")
  public void ignore_testSerializationNoDuplication_regularImmutableMap() throws Exception {
    // Tests that serializing a map, its keySet, and values only writes the underlying data once.

    Entry<Integer, Integer>[] entries = (Entry<Integer, Integer>[]) new Entry<?, ?>[1000];
    for (int i = 0; i < 1000; i++) {
      entries[i] = ImmutableMap.entryOf(i, i);
    }

package kms

import (
	"bytes"
	"sort"
	"unicode/utf8"
)

// Context is a set of key-value pairs that
// are associated with a generate data encryption
// key (DEK).
//
// A KMS implementation may bind the context to the
// generated DEK such that the same context must be
// provided when decrypting an encrypted DEK.
type Context map[string]string

    void injectMirror(List<ArtifactRepository> repositories, List<Mirror> mirrors);

    /**
     * Injects the proxy information into the specified repositories. For each repository that is matched by a proxy,
     * its proxy data will be set accordingly. Repositories without a matching proxy will have their proxy cleared.
     * Note: This method must be called after {@link #injectMirror(List, List)} or the repositories will end up

  tsl::Set_TF_Status_from_Status(
      s, unwrap(op)->Execute(
             absl::MakeSpan(reinterpret_cast<AbstractTensorHandle**>(
                                unwrap(o)->outputs.data()),
                            unwrap(o)->outputs.size()),
             &num_outputs));
}

void TF_DeleteAbstractFunction(TF_AbstractFunction* func) {
  unwrap(func)->Unref();
}

 * multipartReader.use {
 *   while (true) {
 *     val part = multipartReader.nextPart() ?: break
 *     process(part.headers, part.body)
 *   }
 * }
 * ```
 *
 * Note that [nextPart] will skip any unprocessed data from the preceding part. If the preceding
 * part is particularly large or if the underlying source is particularly slow, the [nextPart] call
 * may be slow!
 *

 * array contains information about special ASCII characters such as {@code \\t} and {@code \\n}
 * while {@link #escapeUnsafe} is overridden to handle general escaping of the form {@code \\uxxxx}.
 *
 * <p>The size of the data structure used by {@link ArrayBasedCharEscaper} is proportional to the
 * highest valued character that requires escaping. For example a replacement map containing the

        Fingerprint2011.murmurHash64WithSeed(bytes, 0, bytes.length, 1));
  }

  public void testPutNonChars() {
    Hasher hasher = HASH_FN.newHasher();
    // Expected data is 0x0100010100000000
    hasher
        .putBoolean(true)
        .putBoolean(true)
        .putBoolean(false)
        .putBoolean(true)
        .putBoolean(false)
        .putBoolean(false)

Ça, ajouté au fait que Python soit le langage le plus populaire pour la **Data Science**, le **Machine Learning** et surtout le **Deep Learning**, font de **FastAPI** un très bon choix pour les APIs et applications de **Data Science** / **Machine Learning**.