* cost as fresh ones (1/QPS is the cost for each). We use this trick later.
   *
   * If we pick a function that goes /below/ that horizontal line, it means that we reduce the area
   * of the function, thus time. Thus, the RateLimiter becomes /faster/ after a period of
   * underutilization. If, on the other hand, we pick a function that goes /above/ that horizontal

We verify that we get a `username`, and extract the scopes.

And then we validate that data with the Pydantic model (catching the `ValidationError` exception), and if we get an error reading the JWT token or validating the data with Pydantic, we raise the `HTTPException` we created before.

For that, we update the Pydantic model `TokenData` with a new property `scopes`.

Now that we have **multiple models**, we can update the parts of the app that use them.

We receive in the request a `HeroCreate` *data model*, and from it, we create a `Hero` *table model*.

This new *table model* `Hero` will have the fields sent by the client, and will also have an `id` generated by the database.

the server. We typically separate features from the server because they require
permissions that we don't believe *all* of Elasticsearch should have or because
they depend on libraries that we don't believe *all* of Elasticsearch should
depend on.

For example, reindex requires the `connect` permission so it can perform
reindex-from-remote but we don't believe that the *all* of Elasticsearch should

  // * We want to behave reasonably for timeouts of 0
  // * We are more responsive to completion than timeouts. This is because parkNanos depends on
  //   system scheduling and as such we could either miss our deadline, or unpark() could be delayed
  //   so that it looks like we timed out even though we didn't. For comparison FutureTask respects
  //   completion preferably and AQS is non-deterministic (depends on where in the queue the waiter

///

We are going to include this `APIRouter` in the main `FastAPI` app, but first, let's check the dependencies and another `APIRouter`.

## Dependencies { #dependencies }

We see that we are going to need some dependencies used in several places of the application.

So we put them in their own `dependencies` module (`app/dependencies.py`).

- we also seem to have information like the plugin lifecycle model that's buried inside the maven execution model
- we also seem to have artifact information tangled inside the plugin model
- we have to deal with scripting implementations (groovy, beanshell, ruby)
- we need to deal with a shared context for plugins, like the guarded mojos

         */
        return 0;
    }

    /*
     * We overload this method from ServerMessageBlock because
     * we want writeAndXWireFormat to write the parameterWords
     * and bytes. This is so we can write batched smbs because
     * all but the first smb of the chaain do not have a header
     * and therefore we do not want to writeHeaderWireFormat. We
     * just recursivly call writeAndXWireFormat.
     */

    could use new flow control capacity before acknowledging it, causing strict HTTP/2 servers to
    fail the call.


## Version 4.1.1

_2019-09-05_

 *  Fix: Don't drop repeated headers when validating cached responses. In our Kotlin upgrade we
    introduced a regression where we iterated the number of unique header names rather than then

            of those modules are ones that it uses but forgets to list in its own <inherits>, we'd
            like to get an error. Currently we do, but if we add the extra <inherits> lines, we
            won't.

            I have one idea for a better approach, but it's painful, and I haven't tested it: We
            could postprocess Collect.gwt.xml to add <skip> lines for all the files that should be