[MLton-devel] cvs commit: profiling: elimination of currentSource

Matthew Fluet fluet@CS.Cornell.EDU
Thu, 23 Jan 2003 07:58:37 -0500 (EST)

>   In making this change, I had to tweak similar parts of
>   x86-generate-transfers.fun and c-codegen.fun.  Just a thought: maybe
>   it would be better to move synchronization of gcState to the backend
>   and make it more explicit in the Machine IL (i.e. make FlushStackTop,
>   FlushFrontier and friends Machine IL statements)?  This might
>   eliminate some duplication of knowledge of when to flush from all the
>   codegens.

That's a decent idea, although I'm not quite sure how to put into
practice.  generate-transfers is relatively late in the native codegen, so
any extra info needs to be carried through the whole simplify pipeline.
I'll think about it a little.

This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
MLton-devel mailing list