latest contify.fun

[Stephen Weeks]

> So, we can drop the lookup everywhere and do special checks for (Root,j)
> edges.

Yep.

> Sort of a side question, but does MLton do "unused record slot removal"?

Yes.  cps/useless.fun

That and constant propagation are probably the most complicated optimizations.

> I was going to have both an addEdge and an addEdge' (the latter doing the
> linear lookup), and was wondering if MLton would determine that the
> addEdge' function is never called, so conceivable could be removed from
> the record.  But that would mean chaning the return type of
> newJumpFuncGraph, which started seeming overly complicated.

That should get caught.  You can always do a -keep cps and find out :-)