func newExitCodeFilter(w io.Writer) (*exitCodeFilter, string) {
	const exitStr = "exitcode="

	// Build a regexp that matches any prefix of the exit string at the end of
	// the input. We do it this way to avoid assuming anything about the
	// subcommand output (e.g., it might not be \n-terminated).
	var exitReStr strings.Builder
	for i := 1; i <= len(exitStr); i++ {
		fmt.Fprintf(&exitReStr, "%s$|", exitStr[:i])
	}

<img src="/img/tutorial/custom-response/image01.png">

## Available responses { #available-responses }

Here are some of the available responses.

Keep in mind that you can use `Response` to return anything else, or even create a custom sub-class.

/// note | Technical Details

You could also use `from starlette.responses import HTMLResponse`.

    for (TwoArg.Action first : TwoArg.Action.values()) {
      for (TwoArg.Action second : TwoArg.Action.values()) {
        TwoArg bar = new TwoArg(first, second);
        verifyBarPass(method, bar); // All args nullable:  anything goes!
      }
    }
  }

  /*
   * This next part consists of several sample classes that provide
   * demonstrations of conditions that cause NullPointerTester
   * to succeed/fail.
   */

       */
      @SuppressWarnings("unchecked")
      ImmutableSet<E> result = (ImmutableSet<E>) untypedResult;
      return result;
    }

    if (elements.isEmpty()) {
      // We avoid allocating anything.
      return of();
    }
    // Collection<E>.toArray() is required to contain only E instances, and all we do is read them.
    // TODO(cpovirk): Consider using Object[] anyway.
    @SuppressWarnings("unchecked")

    assertFalse("Heap does not contain 1", mmHeap.contains(1));
    assertTrue("Heap contains 2", mmHeap.remove(2));
    assertEquals(0, mmHeap.size());
    assertFalse("Heap does not contain anything", mmHeap.contains(1));
    assertFalse("Heap does not contain anything", mmHeap.remove(2));
  }

  public void testIteratorPastEndException() {
    MinMaxPriorityQueue<Integer> mmHeap = MinMaxPriorityQueue.create();

   * completely unused, and an expensive acquire(100) request comes. It would be nonsensical to just
   * wait for 100 seconds, and /then/ start the actual task. Why wait without doing anything? A much
   * better approach is to /allow/ the request right away (as if it was an acquire(1) request
   * instead), and postpone /subsequent/ requests as needed. In this version, we allow starting the

            throw new ChecksumFailedException(message, cause);
        } else if (!ArtifactRepositoryPolicy.CHECKSUM_POLICY_IGNORE.equals(checksumPolicy)) {
            // warn if it is set to anything other than ignore
            logger.warn("*** CHECKSUM FAILED - " + message + " - IGNORING");
        }
        // otherwise it is ignored
    }

    private void verifyChecksum(

///

As SQLModel is based on SQLAlchemy, you can easily use **any database supported** by SQLAlchemy (which makes them also supported by SQLModel), like:

* PostgreSQL
* MySQL
* SQLite
* Oracle

### About `root_path` { #about-root-path }

Keep in mind that the server (Uvicorn) won't use that `root_path` for anything else than passing it to the app.

But if you go with your browser to <a href="http://127.0.0.1:8000/app" class="external-link" target="_blank">http://127.0.0.1:8000/app</a> you will see the normal response:

```JSON
{

      // Fine: If it doesn't exist, then there's no chance that we're going to be asked to test it.
    }

    /*
     * These methods "should" call checkNotNull. However, I'm wary of accidentally introducing
     * anything that might slow down execution on such a hot path. Given that the methods are only
     * package-private, I feel OK with just not testing them for NPE.
     *