SSA IL

[Matthew Fluet]

> > Some of them also fail with "Raise must have one var".
> 
> This is a bit more troubling, but I think will go away once everything
> makes it to SSA.

I'm

> > and a segFault on exnHistory2; various other run-time errors.
> 
> These are a bit more scary.  Maybe you could track one down?

Yeah; I didn't get a chance to look at them, I was just watching the
regression output.  Unfortunately, there are some other bugs becoming
manifest: mutex and prodcons aren't terminating.  I'll try to track down
one or more of these tomorrow.

> I have one more minor hack on my todo before the shrinker.  Hopefully
> I will start on that tomorrow.

Great.  Just some notes: I've been mostly assuming some think like
shrinkBlocks: Statement.t vector -> Block.t vector -> Block.t vector
in the code (although they are all one liners), but when I was thinking
about doing a really simple shrinker, I put
shrinkFunction: Statement.t vector -> Function.t -> Function.t
into shrink.sig, which seems more reasonable.