uncurrying

[Stephen Weeks]

> On an unrelated note, Henry mentioned a while ago that failure
> to uncurry could be the cause of the slowdown in the Kit (because
> environment records were growing too big).  Is this your belief
> as well?

I think it's possible, but not probable.

> If so, I would be willing to take a shot at implementing
> a simple uncurryer before the meeting.  Presumably, the uncurryer
> works over XML, right?

Yes.  Sxml would be sufficient.  It's the same signature as Xml except 
no need to handle PolyVal and Exception declerations.  Look at
src/xml/call-count.fun for an example, since it has to handle currying 
in order to count curried applications as one call.

In any case, I think it's simple enough to implement that it's worth
doing the experiment.