for (const Node* n : fn_body->graph.nodes()) {
    stack_traces[n->name()] = n->GetStackTrace();
  }

  TF_Function* tf_function = new TF_Function();
  tf_function->record = new tensorflow::FunctionRecord(
      std::move(fdef), std::move(stack_traces), false);

  return tf_function;
}

TF_Function* TF_GraphToFunction(const TF_Graph* fn_body, const char* fn_name,

    checkArgument(!list.isEmpty(), "number of hash functions (%s) must be > 0", list.size());
    return new ConcatenatedHashFunction(list.toArray(new HashFunction[0]));
  }

  private static final class ConcatenatedHashFunction extends AbstractCompositeHashFunction {

    private ConcatenatedHashFunction(HashFunction... functions) {
      super(functions);
      for (HashFunction function : functions) {
        checkArgument(

func panic(v any)

// The recover built-in function allows a program to manage behavior of a
// panicking goroutine. Executing a call to recover inside a deferred
// function (but not any function called by it) stops the panicking sequence
// by restoring normal execution and retrieves the error value passed to the
// call of panic. If recover is called outside the deferred function it will

  private static final List<String> SOME_STRING_LIST = asList("1", "2", "3", "4");

  private static final Function<Number, String> SOME_FUNCTION = new SomeFunction();

  private static class SomeFunction implements Function<Number, String>, Serializable {
    @Override
    public String apply(Number n) {
      return String.valueOf(n);
    }

  // (`output_tensors`).
  //
  // If provided, a non-null `forward_function` will be used instead of the
  // backward function (`backward_function_getter` /
  // `backward_function_deleter`) to compute jvps for this operation. If
  // `forward_function` is null, a GradientTape is used on the backward function
  // to compute the jvp, which will waste computation when executing eagerly.
  //

!!! tip
    You'll see what other "things", apart from functions, can be used as dependencies in the next chapter.

Whenever a new request arrives, **FastAPI** will take care of:

* Calling your dependency ("dependable") function with the correct parameters.
* Get the result from your function.
* Assign that result to the parameter in your *path operation function*.

```mermaid
graph TB

      Function<? super F, T> function, Supplier<F> supplier) {
    return new SupplierComposition<>(function, supplier);
  }

  private static class SupplierComposition<F extends @Nullable Object, T extends @Nullable Object>
      implements Supplier<T>, Serializable {
    final Function<? super F, T> function;
    final Supplier<F> supplier;

    SupplierComposition(Function<? super F, T> function, Supplier<F> supplier) {

        execute ->> code: return the result
    end

    rect rgba(0, 255, 255, .1)
        code ->> function: say_hi(name="Camila")
        function ->> code: return stored result
    end

    rect rgba(0, 255, 0, .1)
        code ->> function: say_hi(name="Rick")
        function ->> execute: execute function code
        execute ->> code: return the result
    end

   * {@code function} to the argument, then evaluating using {@code this}. That is, for any pair of
   * non-null objects {@code x} and {@code y}, {@code equivalence.onResultOf(function).equivalent(a,
   * b)} is true if and only if {@code equivalence.equivalent(function.apply(a), function.apply(b))}
   * is true.
   *
   * <p>For example:
   *
   * <pre>{@code

   * {@code function} to the argument, then evaluating using {@code this}. That is, for any pair of
   * non-null objects {@code x} and {@code y}, {@code equivalence.onResultOf(function).equivalent(a,
   * b)} is true if and only if {@code equivalence.equivalent(function.apply(a), function.apply(b))}
   * is true.
   *
   * <p>For example:
   *
   * <pre>{@code