Aggregate output from Annotation Processors of different buck modules

Question

I have a project composed of several BUCK modules. Each module uses the same annotation processor to generate a single file for each annotated class. For the sake of simplicity, for each annotated class:

package com.myproject.module1;

@Marker
public class SomeFoo {
  // ... code ...
}

a single class will be generated in the same package:

package com.myproject.module1;

@Generated
public class SomeFoo$$Marker {
  // ... generated code ...
}

Let's say I have two such modules: module1 and module2

Lastly, I have a module3 which is dependent on both above-mentioned modules.

This module contains another annotation:

package com.myproject.module3;

@MarkerCollector
public class Bar {
}

I'm trying to use the same annotation processor (could be a separate one as well), to generate a class that references all previously generated files (due to classes marked with @Marker .

The problem is that the annotation processor runs for each module separately , when run in module3, the annotation processor doesn't 'see' the files generated in other modules .

I've tried using the annotation processor's Filer to create a resource file and write the class names of all generated files, but this resource file is too, created per module .

What's the best way to aggregate/access generated code from different modules?

Answer 1

If you are working with gradle you can create a JavaCompile task and configure it as I did here . My project was targeting android so you will see variant just make sure you put your classPath there instead 🤓