Eu tenho criado APIs com requisitos complexos por vários anos (Aprendizado de Máquina, sistemas distribuídos, tarefas assíncronas, banco de dados NoSQL etc.), liderando vários times de desenvolvedores.

Como parte disso, eu precisava investigar, testar e usar muitas alternativas.

A história do **FastAPI** é, em grande parte, a história de seus predecessores.

Como dito na seção [Alternativas](alternatives.md):

<blockquote markdown="1">

///

## OpenID Connect { #openid-connect }

OpenID Connect est une autre spécification, basée sur **OAuth2**.

Elle étend simplement OAuth2 en précisant certains points relativement ambigus dans OAuth2, afin d'essayer de la rendre plus interopérable.

Par exemple, la connexion Google utilise OpenID Connect (qui, en arrière-plan, utilise OAuth2).

* Esto significa que funcionará bien para archivos grandes como imágenes, videos, binarios grandes, etc. sin consumir toda la memoria.
* Puedes obtener metadatos del archivo subido.
* Tiene una interfaz `async` [parecida a un archivo](https://docs.python.org/3/glossary.html#term-file-like-object).

The number of existing Gradle properties is already in the double digits. More properties will be added in the future.
Although properties offer the benefit of precise user-side control and support gradual evolution practices, the high number of properties also presents challenges.

The main challenge on the user side is comprehensibility.

{* ../../docs_src/cors/tutorial001_py310.py hl[2,6:11,13:19] *}


Os parâmetros padrão usados ​​pela implementação `CORSMiddleware` são restritivos por padrão, então você precisará habilitar explicitamente as origens, métodos ou cabeçalhos específicos para que os navegadores tenham permissão para usá-los em um contexto cross domain.

Os seguintes argumentos são suportados:

   * is unnecessary; for example, {@code copyOf(copyOf(anArrayList))} will copy the data only once.
   * This reduces the expense of habitually making defensive copies at API boundaries. However, the
   * precise conditions for skipping the copy operation are undefined.
   *
   * @throws NullPointerException if any of {@code elements} is null
   * @since 7.0 (source-compatible since 2.0)
   */

Mais pour le chemin d'accès de connexion, nous devons utiliser ces noms pour être compatibles avec la spécification (et pouvoir, par exemple, utiliser le système de documentation API intégré).

La spécification précise également que `username` et `password` doivent être envoyés en données de formulaire (donc pas de JSON ici).

### `scope` { #scope }

   * is set to {@code null}.
   *
   * <p>Like the {@link #toArray()} method, this method acts as bridge between array-based and
   * collection-based APIs. Further, this method allows precise control over the runtime type of the
   * output array, and may, under certain circumstances, be used to save allocation costs.
   *

 *       unnecessary; for example, {@code copyOf(copyOf(anArrayList))} should copy the data only
 *       once. This reduces the expense of habitually making defensive copies at API boundaries.
 *       However, the precise conditions for skipping the copy operation are undefined.
 *   <li><b>Warning:</b> a view collection such as {@link ImmutableMap#keySet} or {@link

     * manage that 4 byte header specifically and therefore the initial
     * bufferIndex, and thus headerStart, would be 4).(NOTE: If one where
     * looking for a way to improve perfomance this is precisly what you
     * would want to do as the jcifs.smb1.netbios.SocketXxxputStream classes
     * arraycopy all data read or written into a new buffer shifted over 4!)
     */

    byte command, flags;