I was looking a bit into the Stratego interpreter code as I noticed it’s slow because exceptions are used in control flow, and then I noticed this snippet near the end of CallT.java, in the evaluateWithArgs method:

class WithArgsHook extends Hook {
    boolean result;
        @Override
        public IConstruct onFailure(IContext env) throws InterpreterException {
            result = false;
            return null;
        }
        @Override
        public IConstruct onSuccess(IContext env) throws InterpreterException {
            result = false;
            return null;
        }
}

Shouldn’t onSuccess set result to true, instead of false? (If not, this at least deserves a big comment next to it, as in other places onSuccess does appear to set a result to true ;-))

Didn’t test yet as I don’t know a way (yet) to test whether this is actually a bug.

Submitted by Tobi Vollebregt on 7 January 2011 at 17:24

Yes, that seems to be a bug. Fixed in r21644. It was probably introduced as the code was changed at some point to get rid of some of the exception-based control flow… The method is very rarely called though, so that may explain why it hasn’t caused any real problems yet. And yes, there’s still a case of exception-based control flow left, which probably explains at least some of the performance difference compared to the compiler, but it’s certainly not the only explanation.