contification paper: benchmarks

Matthew Fluet fluet@CS.Cornell.EDU
Mon, 12 Mar 2001 12:54:45 -0500 (EST)


> I added the following one
> 
> @techreport( TofteEtAl98,
>   author = tofte # and # birkedal # and # elsman # and # hallenberg # and # olesen # and # sestoft # bertelsen,
>   title = "Programming with Regions in the {ML Kit} (for Version 3)",
>   number = "98/25",
>   institution = "University of Copenhagen",
>   year = 1998)
> 
> Is that enough, or which one did you want for 1997?

That's fine.

> > I'm working on the formal transformation description.  Not bad at all;
> > I've got all the rules now, and I just need about an hour to write up some
> > textual description and justification.
> 
> Great.  I thought about the transformation some last night, and I think I
> figured out a way to compute the continuation that each function returns to
> based on A.  Then, walkExp doesn't have to recursively pass the continuation
> around, which I think is the most confusing part.  Anyways, I'll be interested
> to see what you do.

Well, the continuation will just be recursively walking back up the
A(A(A(...(f)...))) until you get either a cont (which is the continuation)
or Unknown (in which case there isn't one).

Let me finish up what I'm working on, which doesn't go that route, but we
can edit it.

> I assume I have the token to work on sections 5 and 6, while you have the token
> for 1-4.

Hold off for a couple minutes and I'll send the updated MLton stats and
fix the reference for the ML-kit.