Java side gets AST as a string. When parsing the AST, the Java code uses constants in AstNodeTypes.java to figure out node types. These constants should match the numbers generated by Antlr on the C++ side (NDDL3.tokens). Often changes to the NDDL grammar cause these constants to shift, in which case AstNodeTypes.java needs to be updated. For now this update is manual, which is bad.

At the very least we need some unittests to verify that constants still match. Better yet, AstNodeTypes could take values directly from NDDL3.tokens, either during build process for JavaEuropaUI or dynamically (in which case NDDL3.tokens would need to be available to Java code).