Re: CWS paper

Top Pagina
Bijlagen:
Bericht als e-mail
+ (text/plain)
Delete this message
Reply to this message
Auteur: Stephen Weeks
Datum:  
Aan: MLton
Onderwerp: Re: CWS paper

> Ok, so your example I translate to
>
>     let fun f () = if ...
>                       then h ()
>                       else g ()
>         and g () = if ...
>                       then h ()
>                       else f ()
>         and h () = ...
>     in K (f ())
>     end

>
> where I filled in something to make the calls actually be tail calls.
>
> In this case, if you do NOT contify f or g, then I don't see how you can
> contify h. Would would it mean to do so since f and g each want to get to
> it.


I agree.  You cannot contify h without contifying both f and g.  The problem
with your original analysis is h will not be contified because its continuation
set is the two element set
    {"return to f", "return to g"} 
Fixing the analysis so that f's, g's, and h's continuation set are all the
singleton {K} while getting all the previous examples is what I'm working on.