You could then update `test_main.py` with the extended tests:

{* ../../docs_src/app_testing/app_b/test_main.py *}


Whenever you need the client to pass information in the request and you don't know how to, you can search (Google) how to do it in `httpx`, or even how to do it with `requests`, as HTTPX's design is based on Requests' design.

Then you just do the same in your tests.

E.g.:

                if (edge == null) { // no edges - don't need this vertex anymore
                    if (entry.equals(v)) { // unless it's an entry point.
                        // currently processing the entry point - it should not have any entry incident edges
                        res.getEntry().getMd().setWhy("This is a graph entry point. No links.");
                    }
                } else {

		if val := globalServiceFreeze.Swap(_ch); val != nil {
			if ch, ok := val.(chan struct{}); ok && ch != nil {
				// Close previous non-nil channel.
				xioutil.SafeClose(ch)
			}
		}
		globalServiceFreezeCnt = 0 // Don't risk going negative.
	}
	globalServiceFreezeMu.Unlock()

Specifically, to see Uvicorn, Starlette and FastAPI compared together (among many other tools).

The simpler the problem solved by the tool, the better performance it will get. And most of the benchmarks don't test the additional features provided by the tool.

The hierarchy is like:

* **Uvicorn**: an ASGI server
    * **Starlette**: (uses Uvicorn) a web microframework

                    try {
                        if (node.filterTrail(filter)) {
                            // If it was optional and not a direct dependency,
                            // we don't add it or its children, just allow the update of the version and artifactScope
                            if (node.isChildOfRootNode() || !artifact.isOptional()) {

//tensorflow/core/tfrt
//tensorflow/python/kernel_tests/signal
//tensorflow/examples
//tensorflow/tools/android
//tensorflow/python/eager/benchmarks
EOF

  # grep patterns for files and targets which don't need to be in the pip
  # package, ever.
  cat > $BATS_TEST_TMPDIR/ignore_these_deps <<EOF
benchmark
_test$
_test.py$
_test_cpu$
_test_cpu.py$
_test_gpu$
_test_gpu.py$
_test_lib$

		size = 32
	}
	a := make([]byte, size)
	b := make([]byte, size)
	b_init := make([]byte, size)
	// randomish but deterministic data
	for i := 0; i < size; i++ {
		a[i] = byte(17 * i)
		b_init[i] = byte(23*i + 100)
	}

	for len := 0; len <= size; len++ {
		for x := 0; x <= size-len; x++ {
			for y := 0; y <= size-len; y++ {
				copy(b, b_init)
				copy(b[y:y+len], a[x:x+len])

   * For example, the tags in this schema are 0 and 1:
   *
   * ```
   * Point ::= SEQUENCE {
   *   x [0] INTEGER OPTIONAL,
   *   y [1] INTEGER OPTIONAL
   * }
   * ```
   *
   * You may also specify a tag class like [DerHeader.TAG_CLASS_APPLICATION]. The default tag class
   * is [DerHeader.TAG_CLASS_CONTEXT_SPECIFIC].
   *
   * ```
   * Point ::= SEQUENCE {
   *   x [APPLICATION 0] INTEGER OPTIONAL,

  fun addPlan(): FakePlan =
    FakePlan(nextPlanId++).also {
      plans += it
    }

  override fun isCanceled() = canceled

  override fun plan(): FakePlan {
    // Return deferred plans preferentially. These don't require addPlan().
    if (deferredPlans.isNotEmpty()) return deferredPlans.removeFirst() as FakePlan

    if (nextPlanIndex >= plans.size && autoGeneratePlans) addPlan()

    require(nextPlanIndex < plans.size) {

           *
           * ...when we build with JDK 22 and run under JDK 8.
           */
          || info.getName().contains("MultimapsTest")
      /*
       * Luckily, we don't care about analyzing tests at all. We'd skip them all if we could do so
       * trivially, but it's enough to skip these ones.
       */
      ) {
        continue;
      }
      Class<?> clazz = info.load();