authors:
  - family-names: Abadi
    given-names: Martín
  - family-names: Agarwal
    given-names: Ashish
  - family-names: Barham
    given-names: Paul
  - family-names: Brevdo
    given-names: Eugene
  - family-names: Chen
    given-names: Zhifeng
  - family-names: Citro
    given-names: Craig
  - family-names: Corrado
    given-names: Greg S.
  - family-names: Davis
    given-names: Andy
  - family-names: Dean

    *   XLA reduction emitter is deterministic when the environment variable
        `TF_DETERMINISTIC_OPS` is set to "true" or "1". This extends
        deterministic `tf.nn.bias_add` back-prop functionality (and therefore
        also deterministic back-prop of bias-addition in Keras layers) to
        include when XLA JIT compilation is enabled.
    *   Fix problem, when running on a CUDA GPU and when either environment

Also allowing to have multiple alternative security schemes that are then checked in a single dependency instead of each one verifying and returning the error to the client automatically when not satisfied. PR [#134](https://github.com/tiangolo/fastapi/pull/134).

* Update [SQL Tutorial](https://fastapi.tiangolo.com/tutorial/sql-databases/#create-a-middleware-to-handle-sessions) to close database sessions even when there are exceptions. PR [#89](https://github.com/tiangolo/fastapi/pull/89)...

* Fix panic in ControllerManager on GCE when it has a problem with creating external loadbalancer healthcheck ([#52646](https://github.com/kubernetes/kubernetes/pull/52646), [@gmarek](https://github.com/gmarek))

Just keep in mind that when you read "server" in general, it could refer to one of those two things.

And if you update that local FastAPI source code when you run that Python file again, it will use the fresh version of FastAPI you just edited.

That way, you don't have to "install" your local version to be able to test every change.

!!! note "Technical Details"
    This only happens when you install using this included `requirements.txt` instead of running `pip install fastapi` directly.

        }
    }

    inner class TypeBuilder {
        fun buildKtType(coneType: ConeKotlinType): KtType {
            return when (coneType) {
                is ConeClassLikeTypeImpl -> {
                    when {
                        coneType.lookupTag.toSymbol(rootSession) == null -> {

The `root_path` is used to handle these specific cases.

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

## Proxy with a stripped path prefix

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`.

For consistency, the same profiles are used across each chart, even if they do not impact a given chart.

Explicitly set values have highest priority, then profile settings, then chart defaults.

As an implementation detail of profiles, the default values for the chart are all nested under `defaults`.
When configuring the chart, you should not include this.
That is, `--set some.field=true` should be passed, not `--set defaults.some.field=true`.

# `Request` class

You can declare a parameter in a *path operation function* or dependency to be of type `Request` and then you can access the raw request object directly, without any validation, etc.

You can import it directly from `fastapi`:

```python
from fastapi import Request
```

!!! tip