new release of MLton, an optimizing SML compiler

[Daniel Wang]

"Stephen Weeks" <sweeks@intertrust.com> writes:

> > So can you give me a hacker perspetive on what has changed? i.e. IR
> > wise.. have the variant types gone away in the IR?
> 
> Yes. :-)  Case statements now implicitly destructure their constructed
> args.  That was the only IL change (as far as I remember).
> 
> > There were lots of silly
> > things that forced me to hack my code in before CPS I'd love to hack my code
> > in after CPS and am wondering if the new release has made my life eaiser...
> 
> Was that due to variant types or something else?
> 

Mainly the variant types... the other reasons was the globals.. but that I
can fix my IL for in anycase...

> As to other changes:
> * lots of bug fixes (compiler and basis library)
> * basis library additions
> * a couple new Cps optimizations
> * somewhat different runtime to handle threads and signals
> * more conservative register allocator (due to context switching at
>    any limit check point)
> * the generated C code now looks very different -- more like assembly
> 

hmm.. okay.. I'm hacking up a proof and doing other related TeXing for my
thesis. At somepoint I have to go back and run more experiments on 
non-trivial ML benchmarks.. the code I have right now just barely runs life
so I think, the right thing for me to do is to rehack all my changes into
the newer version and see if I end up with a more robust system... 

Anyway.. I'm still in TeX hacking mode but should switch over to SML hacking
some time in August...

BTW whats the deal with hosting everything on  sourcelight have you been
job hopping in the valley?

If you looking for hosting space.. it might be worth taking a look at

http://www.sourceforge.net

They export a CVS repository which is quite useful in my experience.