ref: Optional[str] = Field(default=None, alias="$ref")
    dynamicRef: Optional[str] = Field(default=None, alias="$dynamicRef")
    defs: Optional[dict[str, "SchemaOrBool"]] = Field(default=None, alias="$defs")
    comment: Optional[str] = Field(default=None, alias="$comment")
    # Ref: JSON Schema 2020-12: https://json-schema.org/draft/2020-12/json-schema-core.html#name-a-vocabulary-for-applying-s

    linkshared = 1,
    linkstatic = 1,
    deps = tf_custom_op_library_additional_deps_impl(),
)

# Get a DEF file generated by parsing all object files
# of tf_custom_op_library_additional_deps.so
filegroup(
    name = "tensorflow_def_file",
    srcs = [":tf_custom_op_library_additional_deps.dll"],
    output_group = "def_file",
)

# Filter the DEF file to reduce the number of symbols to 64K or less.

	if tok != scanner.Ident {
		in.expectText("expected identifier after '#'")
	}
	if !in.enabled() {
		// Can only start including again if we are at #else or #endif but also
		// need to keep track of nested #if[n]defs.
		// We let #line through because it might affect errors.
		switch in.Stack.Text() {
		case "else", "endif", "ifdef", "ifndef", "line":
			// Press on.
		default:
			return false
		}
	}

La même chose s'applique aux dépendances. Si une dépendance est définie avec `def` plutôt que `async def`, elle est exécutée dans la threadpool externe.

### Sous-dépendances

@app.websocket("/", dependencies=[create_dependency("index")])
async def index(websocket: WebSocket, deps: DepList):
    await websocket.accept()
    await websocket.send_text(json.dumps(deps))
    await websocket.close()


@router.websocket("/router", dependencies=[create_dependency("routerindex")])
async def routerindex(websocket: WebSocket, deps: DepList):
    await websocket.accept()

Stellen wir uns vor, dass das Laden des Modells **eine ganze Weile dauern** kann, da viele **Daten von der Festplatte** gelesen werden müssen. Sie möchten das also nicht für jeden Request tun.

        def projectDir = 'C:\\some\\agent\\workspace'

        expect:
        (line =~ KillLeakingJavaProcesses.generateLeakingProcessKillPattern(projectDir)).find()
    }

    def "matches daemon process started by performance test on Windows"() {

from fastapi import Depends, FastAPI, Header, status
from fastapi.testclient import TestClient

app = FastAPI()


def get_header(*, someheader: str = Header()):
    return someheader


def get_something_else(*, someheader: str = Depends(get_header)):
    return f"{someheader}123"


@app.get("/")
def get_deps(dep1: str = Depends(get_header), dep2: str = Depends(get_something_else)):
    return {"dep1": dep1, "dep2": dep2}

Elle peut être une fonction asynchrone (`async def`) ou une fonction normale (`def`), **FastAPI** saura la gérer correctement.

Dans cet exemple, la fonction de tâche écrira dans un fichier (afin de simuler un envoi d'email).

L'opération d'écriture n'utilisant ni `async` ni `await`, on définit la fonction avec un `def` normal.

```Python hl_lines="1"
async def get_burgers(number: int):
    # Mache hier etwas Asynchrones, um die Burger zu erstellen
    return burgers
```

... statt mit `def`:

```Python hl_lines="2"
# Dies ist nicht asynchron
def get_sequential_burgers(number: int):