[MLton-devel] icfp?

[Stephen Weeks]

> I was speaking with Olin Shivers last week and he
> strongly encouraged us to submit something for
> ICFP this  year; they are actively looking for 
> implementation-related papers. 

Here are three things I see that we could write about:

SSA IL
	linearization algorithm
	type system
	exceptions
	interaction with optimizations
	compare with CPS
limit check insertion algorithms
how the profiler works
	insertion of Enter/Leave at source
	interaction with optimization
	nice interaction with function duplication (due to
		defunctorization, monomorphisation, polyvariance,
		and inlining) and flow analysis produces accurate call
		graphs

> I was thinking of something along the lines of Shao's 98 icfp paper
> on Implementing Typed Intermediate Languages where we use profile
> information to guage the cost of various optimization phases.

Nice idea.  We could measure the fraction of time/bytes spent on type
manipulation throughout the compiler.  It'll take some playing with
the profiling output, but I'm all for advertising the new profiler.

Engineering-wise, we are pretty much like that paper.  We use
hash-consing, dags instead of trees, and memoization.  But I guess
it's nice to confirm that the approach works well.  We certainly have
a much easier type language.



-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
_______________________________________________
MLton-devel mailing list
MLton-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mlton-devel