But there are situations where you might need to access the `Request` object directly.

## Details about the `Request` object { #details-about-the-request-object }

As **FastAPI** is actually **Starlette** underneath, with a layer of several tools on top, you can use Starlette's [`Request`](https://www.starlette.dev/requests/) object directly when you need to.

 *   Added `jit_compile` as a settable property to `tf.keras.Model`.
 *   Added `synchronized` optional parameter to `layers.BatchNormalization`.

    Works with **any HTTP method** (`GET`, `POST`, etc.), which makes it compatible
    with protocols like MCP that stream SSE over `POST`.

    The actual encoding logic lives in the FastAPI routing layer. This class
    serves mainly as a marker and sets the correct `Content-Type`.
    """

    media_type = "text/event-stream"


def _check_id_no_null(v: str | None) -> str | None:
    if v is not None and "\0" in v:

A single HTTP call may require follow-up requests to be made to handle authentication challenges, redirects, and HTTP-layer timeouts. In such cases multiple connections, requests, and responses may be attempted. Follow-ups are another reason a single call may trigger multiple events of the same type.

other tools **build** these container images **incrementally**, adding **one layer on top of the other**, starting from the top of the `Dockerfile` and adding any files created by each of the instructions of the `Dockerfile`.

Docker and similar tools also use an **internal cache** when building the image, if a file hasn't changed since the last time building the container image, then it will **reuse the same layer** created the last time, instead of copying the file again and creating a new...

* Automatic data model documentation with [**JSON Schema**](https://json-schema.org/) (as OpenAPI itself is based on JSON Schema).
* Designed around these standards, after a meticulous study. Instead of an afterthought layer on top.
* This also allows using automatic **client code generation** in many languages.

### Automatic docs { #automatic-docs }

* Certificates have a **lifetime**.
    * They **expire**.
    * And then they need to be **renewed**, **acquired again** from the third party.
* The encryption of the connection happens at the **TCP level**.
    * That's one layer **below HTTP**.
    * So, the **certificate and encryption** handling is done **before HTTP**.
* **TCP doesn't know about "domains"**. Only about IP addresses.

    handler.windowUpdate(streamId, increment)
  }

  @Throws(IOException::class)
  override fun close() {
    source.close()
  }

  /**
   * Decompression of the header block occurs above the framing layer. This class lazily reads
   * continuation frames as they are needed by [Hpack.Reader.readHeaders].
   */
  internal class ContinuationSource(
    private val source: BufferedSource,
  ) : Source {

And it's also used internally when mounting sub-applications.

Having a proxy with a stripped path prefix, in this case, means that you could declare a path at `/app` in your code, but then, you add a layer on top (the proxy) that would put your **FastAPI** application under a path like `/api/v1`.

In this case, the original path `/app` would actually be served at `/api/v1/app`.

    * One Uvicorn **process manager** would listen on the **IP** and **port**, and it would start **multiple Uvicorn worker processes**.
* **Kubernetes** and other distributed **container systems**
    * Something in the **Kubernetes** layer would listen on the **IP** and **port**. The replication would be by having **multiple containers**, each with **one Uvicorn process** running.
* **Cloud services** that handle this for you