And if the server is restarted (for example after updates, or migrations from the cloud provider) you probably **won't notice it**. And because of that, you won't even know that you have to restart the process manually. So, your API will just stay dead. 😱

### Run Automatically on Startup { #run-automatically-on-startup }

    KOqkqm57TH2H3eDJAkSnh6/DNFu0Qg==
    -----END CERTIFICATE-----
    """.trimIndent().decodeCertificatePem()

  private val client: OkHttpClient

  init {
    // This implementation just embeds the PEM files in Java strings; most applications will
    // instead read this from a resource file that gets bundled with the application.
    val certificates =
      HandshakeCertificates
        .Builder()

{* ../../docs_src/async_tests/app_a_py310/test_main.py hl[7] *}

/// tip

Note that the test function is now `async def` instead of just `def` as before when using the `TestClient`.

///

Then we can create an `AsyncClient` with the app, and send async requests to it, using `await`.

{* ../../docs_src/async_tests/app_a_py310/test_main.py hl[9:12] *}

	tx = tx.Preload("Orders", "state != ?", "cancelled")
	args, ok := tx.Statement.Preloads["Orders"]
	if !ok || len(args) != 2 {
		t.Fatalf("Preload expected with args, got %v", tx.Statement.Preloads)
	}

	// Scopes: just ensure calling Scopes doesn't panic and returns a DB
	tx = tx.Scopes(func(d *gorm.DB) *gorm.DB { return d.Where("status = ?", "ok") })
	if tx == nil {
		t.Fatalf("Scopes returned nil")
	}

	// Unscoped

		// binary arrays) - perhaps we need to check the annotation to
		// ensure it's UTF8 encoded.
		value = string(val)
	case [12]byte:
		// TODO: This is returned for the parquet INT96 type. We just
		// treat it same as []byte (but AWS S3 treats it as a large int)
		// - fix this later.
		value = string(val[:])
	case int32:
		value = int64(val)
		if logicalType := se.GetLogicalType(); logicalType != nil {

   * types. CharBuffer has poor performance when being written into or read out of so round tripping
   * all the bytes through the buffer takes a long time. With these specialized types we can just
   * use a char array.
   *
   * @param from the object to read from
   * @param to the object to write to
   * @return the number of characters copied
   * @throws IOException if an I/O error occurs
   */

        if (shaFiles.isEmpty() == false) {
            throw new GradleException("Unused sha files found: \n" + joinFilenames(shaFiles));
        }

    }

    // This is just a marker output folder to allow this task being up-to-date.
    // The check logic is exception driven so a failed tasks will not be defined
    // by this output but when successful we can safely mark the task as up-to-date.

     */
    void setDegreeOfConcurrency(int degreeOfConcurrency);

    /**
     * @return the degree of concurrency for the build.
     */
    int getDegreeOfConcurrency();

    // Recursive (really to just process the top-level POM)
    MavenExecutionRequest setRecursive(boolean recursive);

    boolean isRecursive();

    MavenExecutionRequest setPom(File pom);

    File getPom();

    // Errors

        healthyChannel.setState(ChannelState.ESTABLISHED);

        // Use reflection or package-private method to add channel for testing
        // For now, just test the empty case
        assertEquals(0, channelManager.getHealthyChannels().size());
    }

    @Test
    void testChannelSelection() {
        try {
            channelManager.selectChannel(mockMessage);

{* ../../docs_src/server_sent_events/tutorial004_py310.py hl[25,27,31] *}

## SSE with POST { #sse-with-post }

SSE works with **any HTTP method**, not just `GET`.

This is useful for protocols like [MCP](https://modelcontextprotocol.io) that stream SSE over `POST`:

{* ../../docs_src/server_sent_events/tutorial005_py310.py hl[14] *}