[tool.fastapi]
entrypoint = "app.main:app"
```

that is equivalent to importing like:

```python
from app.main import app
```

That way the `fastapi` command will know where to find your app.

/// Note

You could also pass the path to the command, like:

```console
$ fastapi dev app/main.py
```

But you would have to remember to pass the correct path every time you call the `fastapi` command.

The same models are shared among requests, so, it's not one model per request, or one per user or something similar.

Let's imagine that loading the model can **take quite some time**, because it has to read a lot of **data from disk**. So you don't want to do it for every request.

 * from uniform), and amortized since some operations can trigger a hash table resize.
 *
 * <p>As compared with {@link java.util.LinkedHashMap}, this structure places significantly reduced
 * load on the garbage collector by only using a constant number of internal objects.
 *
 * <p>This class should not be assumed to be universally superior to {@code

# Extra Models { #extra-models }

Continuing with the previous example, it will be common to have more than one related model.

This is especially the case for user models, because:

* The **input model** needs to be able to have a password.
* The **output model** should not have a password.
* The **database model** would probably need to have a hashed password.

/// danger

        listen       9000;
        listen  [::]:9000;
        server_name  localhost;

        # To allow special characters in headers
        ignore_invalid_headers off;
        # Allow any size file to be uploaded.
        # Set to a value such as 1000m; to restrict file size to a specific value
        client_max_body_size 0;
        # To disable buffering
        proxy_buffering off;
        proxy_request_buffering off;

    }

    /**
     * Scrolls through all documents in the specified index and applies the callback function to each hit.
     *
     * @param index the name of the index to scroll through
     * @param callback the function to apply to each search hit, returning true to continue or false to stop
     * @return the number of documents processed
     */

   * eliminate duplicate keys (for buildKeepingLast()) then we have to ensure that a later call to
   * buildOrThrow() will still throw as if the duplicates had not been eliminated. And the exception
   * message must mention two values that were associated with the duplicate key in two different
   * calls to Builder.put (though we don't really care *which* two values if there were more than

 *
 * <p>To use a reference of this type (say, named {@code guavaPredicate}) in a context where {@code
 * java.util.function.Predicate} is expected, use the method reference {@code
 * guavaPredicate::apply}. For the other direction, use {@code javaUtilPredicate::test}. A future
 * version of this interface will be made to <i>extend</i> {@code java.util.function.Predicate}, so

        return true;
      }
    }
  }

  @Override
  public int size() {
    return elements.length;
  }

  // We're careful to put only E instances into the array in the mainline.
  // (In the backport, we don't need this suppression, but we keep it to minimize diffs.)
  @SuppressWarnings("unchecked")
  @Override
  public UnmodifiableIterator<E> iterator() {

)

// Fdatasync - fdatasync() is similar to fsync(), but does not flush modified metadata
// unless that metadata is needed in order to allow a subsequent data retrieval
// to  be  correctly  handled.   For example, changes to st_atime or st_mtime
// (respectively, time of last access and time of last modification; see inode(7))
// do not require flushing because they are not necessary for a subsequent data