Re: Compiling with Continuations

bliss@sp64.csrd.uiuc.edu (Brian Bliss)
Fri, 31 Jan 92 18:54:49 GMT

          From comp.compilers
Related articles
Re: Compiling with Continuations/Andrew Appel eifrig@blaze.cs.jhu.edu (1992-01-27)
Compiling with Continuations wand@corwin.CCS.Northeastern.EDU (1992-01-29)
Compiling with Continuations appel@Princeton.EDU (Andrew Appel) (1992-01-30)
Re: Compiling with Continuations bliss@sp64.csrd.uiuc.edu (1992-01-31)
Re: Compiling with continuations xleroy@margaux.inria.fr (1992-02-02)
| List of all articles for this month |
Newsgroups: comp.compilers
From: bliss@sp64.csrd.uiuc.edu (Brian Bliss)
Keywords: optimize, Scheme, storage
Organization: UIUC Center for Supercomputing Research and Development
References: 92-01-101 92-01-123
Date: Fri, 31 Jan 92 18:54:49 GMT

>In particular, in the absence of call/cc and the like, continuations
>(activation records) can ALWAYS be stack-allocated. Heap allocation is
>only necessary for environments (closures), even if your compiler is
>arbitrarily stupid.


I don't see what the big deal about call/cc is. Yes, heap allocation is
necessary for closures which escape upward. the closure for the
continuation which call/cc creates is no different from any other; heap
allocation may or may not be necessary for it. any program written using
call/cc can be rewritten without it.


bb
--

Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.