tag:blogger.com,1999:blog-7094652.post7296257791824760360..comments2024-03-23T14:36:09.980+00:00Comments on Neil Mitchell's Blog (Haskell etc): Supercompilation for HaskellNeil Mitchellhttp://www.blogger.com/profile/13084722756124486154noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-7094652.post-1401015121093161832007-12-13T10:22:00.000+00:002007-12-13T10:22:00.000+00:00Josef: I've tried some CPS programs, and they were...Josef: I've tried some CPS programs, and they were dealt with fine - an earlier version of Supero even converted the programs to continuation passing style first. However, it is a trickier case than normal, so the results may not be as good - I'll need to benchmark that in detail.<BR/><BR/>Thanks for the comments about GRIN - I'll incorporate them into my related work section.Neil Mitchellhttps://www.blogger.com/profile/13084722756124486154noreply@blogger.comtag:blogger.com,1999:blog-7094652.post-72748515593024574132007-12-13T00:59:00.000+00:002007-12-13T00:59:00.000+00:00First of all I want to say you've done a terrific ...First of all I want to say you've done a terrific work with Supero. I'm looking forward to a public release.<BR/><BR/>I'm curious about how Supero would handle programs in continuation passing style. Those things are typically hard to deal with and often require some kind of flow analysis. It'd be interesting to see if supercompilation somehow supersedes even that.<BR/><BR/>Also, a comment on your related work section. I think you've missed the point about GRIN. The major thing about GRIN is the flow analysis + defunctionalization + eval inlining that happens as the first step. This step completely removes all unknown function call (trading them for case analysis) and enables many of the later optimizations.Josefhttps://www.blogger.com/profile/13272830598221833253noreply@blogger.com