Caterpillar

Call stack hints are straight-forward enough, if you just look at simple cases. But I'm not so sure how they would work in the presence of mutual recursion. In that case, it might be necessary to search more deeply into the stack to find a hint, causing more runtime overhead.

I honestly don't know, if that's going to be a problem, or if there are clever ways to keep the overhead minimal. I need more time to think about this. And once I start implementing call stack hints, I'll probably come up with new insights too.

One last case we haven't looked at yet are anonymous functions. Those can also have branches, which might necessitate call stack hints. I can't think of a reason why that case would be different from named functions though. So I think this won't require any special handling.