!!! check
    Notice that the value your function received (and returned) is `3`, as a Python `int`, not a string `"3"`.

    So, with that type declaration, **FastAPI** gives you automatic request <abbr title="converting the string that comes from an HTTP request into Python data">"parsing"</abbr>.

## Data validation

The main thing you need to run a **FastAPI** application (or any other ASGI application) in a remote server machine is an ASGI server program like **Uvicorn**, this is the one that comes by default in the `fastapi` command.

There are several alternatives, including:

* <a href="https://www.uvicorn.org/" class="external-link" target="_blank">Uvicorn</a>: a high performance ASGI server.

  virtual Status NumElements(int64_t* num_elements) const = 0;
  // Returns size of specified dimension
  //
  // -1 indicates an unknown axis length; this is unreachable for most standard
  // ImmediateExecutionTensorHandles, but comes up for example when computing
  // the shape of a parallel tensor with component shapes differing across
  // devices.
  virtual Status Dim(int dim_index, int64_t* dim) const = 0;

		if err != nil {
			return nil, err
		}

		// Pick output column names
		if expr.As != "" {
			output, err = output.Set(expr.As, v)
		} else if comp, ok := getLastKeypathComponent(expr.Expression); ok {
			output, err = output.Set(comp, v)
		} else {
			output, err = output.Set(fmt.Sprintf("_%d", i+1), v)
		}
		if err != nil {
			return nil, err
		}
	}

	// Update count of records output.

              sneakyThrow(new SomeCheckedException());
              throw new AssertionError(); // unreachable
            }
          }
        };

    // The first time, the sneakily-thrown exception comes out
    try {
      iter.hasNext();
      fail("No exception thrown");
    } catch (Exception e) {
      if (!(e instanceof SomeCheckedException)) {
        throw e;
      }
    }

---
name: TensorFlow Lite Op Request
about: Use this template for reporting Lite ops you are using or missing
labels: 'comp:lite'

---

**System information**
- OS Platform and Distribution (e.g., Linux Ubuntu 16.04):
- TensorFlow installed from (source or binary):
- TensorFlow version (or github SHA if from source):


**Provide the text output from tflite_convert**

```
# Copy and paste here
```

  }

  /**
   * Either writes this request to [sink] or measures its content length. We have one method
   * do double-duty to make sure the counting and content are consistent, particularly when it comes
   * to awkward operations like measuring the encoded length of header strings, or the
   * length-in-digits of an encoded integer.
   */
  private fun writeOrCountBytes(
    sink: BufferedSink?,
    countBytes: Boolean,

 *       testlib), but why bother?
 *   <li>Stripping code entirely might help us keep under the method limit someday. Even if it never
 *       comes to that, it may at least help with build and startup times.
 * </ul>
 */
@Retention(CLASS)
@Target({ANNOTATION_TYPE, CONSTRUCTOR, FIELD, METHOD, TYPE})
@GwtCompatible

How to contribute
=================

Thank you so much for wanting to contribute to Guava! Here are a few important
things you should know about contributing:

1.  API changes require discussion, use cases, etc. Code comes later.
2.  Pull requests are great for small fixes for bugs, documentation, etc.
3.  Pull requests are not merged directly into the master branch.
4.  Code contributions require signing a Google CLA.

API changes
-----------

     * its reference to it.
     *
     * If the output future is *not* done, then the task field will be cleared after the task runs
     * or after the output future is done, whichever comes first.
     */
    if (reason == OUTPUT_FUTURE_DONE) {
      this.task = null;
    }
  }

  @Override
  protected void interruptTask() {
    CombinedFutureInterruptibleTask<?> localTask = task;