feat: allow to stop the invoker

[loicmathieu]

Allow to stop the invoker so it can be easily used inside an integration test like this.

Fixes #18

        // start the invoker without joining to avoid blocking the thread
        Invoker invoker = new Invoker(
                8081,
                "org.example.HelloWorldHttpFunction",
                "http",
                Thread.currentThread().getContextClassLoader()
        );
        invoker.startServer(false);

        // test the function using RestAssured
        when()
                .get("http://localhost:8081")
                .then()
                .statusCode(200)
                .body(is("Hello World!"));

        // stop the invoker
        invoker.stopServer();

[eamonnmcmanus]

Thanks for doing this! I think it will be very useful for people testing their functions. I did have a few requests.

[eamonnmcmanus]

That looks great, thanks! Do you know why the indentation of some lines has changed? The changes don't look like improvements.

[loicmathieu]

@eamonnmcmanus sorry, something may happens when I try to fix the format issue using the Google formatter tool.

I revert the identation changes.

[eamonnmcmanus]

The google-java-format check is failing because of these diffs:

8:25:13.000000000 -0800
+++ -	2022-02-21 08:27:05.000000000 -0800
@@ -228,8 +228,8 @@
   }

   /**
-   * This will start the server and wait (join) for function calls.
-   * To start the server inside a unit or integration test, use {@link #startTestServer()} instead.
+   * This will start the server and wait (join) for function calls. To start the server inside a
+   * unit or integration test, use {@link #startTestServer()} instead.
    *
    * @see #stopServer()
    * @throws Exception
@@ -238,31 +238,21 @@
     startServer(true);
   }

-
   /**
    * This will start the server and return.
    *
-   * This method is designed to be used for unit or integration testing only.
-   * For other use cases use {@link #startServer()}.
+   * <p>This method is designed to be used for unit or integration testing only. For other use cases
+   * use {@link #startServer()}.
    *
-   * Inside a test a typical usage will be:
-   * {@code
-   *         // Create an invoker
-   *         Invoker invoker = new Invoker(
-   *                 8081,
-   *                 "org.example.MyHttpFunction",
-   *                 "http",
-   *                 Thread.currentThread().getContextClassLoader()
-   *         );
+   * <p>Inside a test a typical usage will be: {@code // Create an invoker Invoker invoker = new
+   * Invoker( 8081, "org.example.MyHttpFunction", "http",
+   * Thread.currentThread().getContextClassLoader() );
    *
-   *         // Start the test server
-   *         invoker.startTestServer();
+   * <p>// Start the test server invoker.startTestServer();
    *
-   *         // Test the function
+   * <p>// Test the function
    *
-   *         // Stop the test server
-   *         invoker.stopServer();
-   * }
+   * <p>// Stop the test server invoker.stopServer(); }
    *
    * @see #stopServer()
    * @throws Exception
@@ -317,7 +307,6 @@
    *
    * @see #startServer()
    * @see #startTestServer()
-   *
    * @throws Exception
    */
   public void stopServer() throws Exception {

I am not sure why you are seeing the indentation change when you run g-j-f. Perhaps --aosp is set? Anyway the main thing is that you need <pre>...</pre> around the {@code...} block.

[eamonnmcmanus]

Could you put something like this first?

    if (server != null) {
      throw new IllegalStateException("Server already started");
    }

[loicmathieu]

Would it be better to do it inside the startServer(boolean) method instead of here ?

[eamonnmcmanus]

Yes, you are right of course.

[eamonnmcmanus]

And here:

    if (server == null) {
     throw new IllegalStateException("Server not started");
   }

(which is a bit nicer than NullPointerException).

[loicmathieu]

Maybe setting the server to null after stopping it so the same invoker can be stoped and stated again without needed to create it again ?
This can be convenient if someone want to stop it / start it inside a @Before and create it inside an @BeforeAll ?

[eamonnmcmanus]

Yes, thanks for catching that, it should indeed be set to null after the stop() call.

[loicmathieu]

Changes done

[eamonnmcmanus]

Thanks again! I'll fix the formatting problems myself, since the formatter works fine for me.

[loicmathieu]

Thanks @eamonnmcmanus to have run the formater for me.