contification paper

Stephen Weeks MLton@sourcelight.com
Sun, 11 Mar 2001 12:30:41 -0800 (PST)


>   I'm still working on the benchmarks section; but here is the where I
> currently stand.  I should be back in my office and online by 1:30 EST, so
> I should make some mork progress this afternoon.  Drop me a line when you
> get in and we'll sync up again.

I'm online.  

> I made a couple of changes to the CPS operational semantics (replaced some
> fun k = e  with  fun f = e, becuase I think that's what you meant).
> Everything else that I actually touched was just some spelling or word
> duplication.  I haven't read it with an editting eye, but otherwise I
> think it's really good -- now I finally understand how the CPS IL works!

Cool.

> Two things to think about:
>  - make the main function "stand out" in the CPS IL;
>    i.e., something like
>     P ::= let fun f(<x>) = e ... in f_main() end
>  - I think we need a sentence along the lines, "As usual, we assume all
>     variables are unique" somewhere.  In particular, we should note that
>     function names and continuation names are distict, so the
>     transformation doesn't really need to do renaming.

Makes sense.

Send me a new snapshot if you need to, and let me know what sections I can work
on right now.  If you're just working on the benchmarks, maybe I could touch
everything else?