Cs guava upgrade

[sahusanket]

No description provided.

[gemini-code-assist]

Code Review

This pull request performs a major cleanup of deprecated Guava and Twill APIs across the codebase, notably replacing synchronous service lifecycle methods (such as startAndWait and stopAndWait) with their asynchronous counterparts (startAsync and stopAsync), replacing Throwables.propagate with standard exceptions, and migrating to MoreObjects and modern CharMatcher and Stopwatch APIs. Additionally, the custom GuavaClassRewriter is removed, and stub suppliers are introduced to maintain compatibility. The review feedback highlights critical issues and improvement opportunities: first, in Services.java, retrieving methods via reflection using service.getClass() instead of Service.class will fail with an IllegalAccessException if the concrete service class is non-public; second, in AppMetadataStore.java, removing IllegalStateException from the catch block during JSON parsing exposes the scan operation to unhandled runtime exceptions from malformed metadata; and third, several files, including LocalizationUtils.java and MapReduceRuntimeService.java, can be simplified by using standard Java NIO Files.copy or native Location stream methods directly instead of deprecated custom supplier wrappers.

[gemini-code-assist]

Using service.getClass().getMethod(...) to retrieve methods via reflection will throw an IllegalAccessException if the concrete class of the Service is non-public (such as package-private or anonymous classes), even if the methods themselves are public. Since Service is a public interface, you should retrieve the methods from Service.class instead of service.getClass().

Suggested change

	service.getClass().getMethod("startAsync").invoke(service);
	service.getClass().getMethod("awaitRunning").invoke(service);
	} catch (NoSuchMethodException e) {
	// Guava 13
	Object future = service.getClass().getMethod("start").invoke(service);
	Service.class.getMethod("startAsync").invoke(service);
	Service.class.getMethod("awaitRunning").invoke(service);
	} catch (NoSuchMethodException e) {
	// Guava 13
	Object future = Service.class.getMethod("start").invoke(service);

[gemini-code-assist]

Using service.getClass().getMethod(...) to retrieve methods via reflection will throw an IllegalAccessException if the concrete class of the Service is non-public (such as package-private or anonymous classes), even if the methods themselves are public. Since Service is a public interface, you should retrieve the methods from Service.class instead of service.getClass().

Suggested change

	service.getClass().getMethod("stopAsync").invoke(service);
	service.getClass().getMethod("awaitTerminated").invoke(service);
	} catch (NoSuchMethodException e) {
	// Guava 13
	Object future = service.getClass().getMethod("stop").invoke(service);
	Service.class.getMethod("stopAsync").invoke(service);
	Service.class.getMethod("awaitTerminated").invoke(service);
	} catch (NoSuchMethodException e) {
	// Guava 13
	Object future = Service.class.getMethod("stop").invoke(service);

[gemini-code-assist]

The JsonReader operations (such as beginObject, peek, nextName, and skipValue) as well as GSON.fromJson can throw IllegalStateException or JsonSyntaxException (which are RuntimeExceptions) if the JSON metadata is malformed or has an unexpected structure (e.g., if spec is a string instead of an object). Since you removed IllegalStateException from the catch block, any such malformed metadata in the database will cause the entire scan operation to fail with an unhandled exception. Please catch Exception or IOException | RuntimeException to safely handle and log these parsing errors.

Suggested change

	} catch (IOException e) {
	LOG.warn("Failed to extract artifact id from raw application metadata", e);
	}
	} catch (Exception e) {
	LOG.warn("Failed to extract artifact id from raw application metadata", e);
	}

[gemini-code-assist]

Instead of manually opening a FileOutputStream and using ByteStreams.copy, you can use the standard Java NIO Files.copy method to copy the InputStream directly to the target path. This is cleaner, more concise, and utilizes optimized platform-specific copy operations.

    try (InputStream in = Resources.asByteSource(url).openStream()) {
      Files.copy(in, tempFile.toPath(), StandardCopyOption.REPLACE_EXISTING);
    }

[gemini-code-assist]

Since Location provides a direct getOutputStream() method, you can use it directly instead of wrapping it with Locations.newOutputSupplier. This simplifies the code and avoids using the custom/deprecated supplier wrapper.

Suggested change

	try (InputStream in = new FileInputStream(file);
	OutputStream out = Locations.newOutputSupplier(targetLocation).getOutput()) {
	ByteStreams.copy(in, out);
	}
	try (InputStream in = new FileInputStream(file);
	OutputStream out = targetLocation.getOutputStream()) {
	ByteStreams.copy(in, out);
	}

[gemini-code-assist]

Since both programJarLocation and programJarCopy are Location objects, you can use their native getInputStream() and getOutputStream() methods directly instead of wrapping them with Locations.newInputSupplier and Locations.newOutputSupplier. This simplifies the code and avoids using the custom/deprecated supplier wrappers.

Suggested change

	try (InputStream in = Locations.newInputSupplier(programJarLocation).getInput();
	OutputStream out = Locations.newOutputSupplier(programJarCopy).getOutput()) {
	ByteStreams.copy(in, out);
	}
	try (InputStream in = programJarLocation.getInputStream();
	OutputStream out = programJarCopy.getOutputStream()) {
	ByteStreams.copy(in, out);
	}