I've started down the SSA route for Cornerstone, too.  This is why I'm not back 
to using your BS 
library yet.  I'd be very interested in working with other people to do SSA 
work, though.

For the work I've done so far, I've been using Scala, and it's...okay.  Calling 
Scala code from 
Java/Groovy requires understanding of the non-obvious compiler mappings in 
Scala, so once you're in 
Scala, it's best to stay there.  Also, I'm not sold that the Scala pattern 
matching is really 
gaining me much over polymorphic method calls in this particular case -- and 
this is coming from a 
once-upon-a-time OCaml hacker/evangelist.  So redoing what SSA library stuff 
I've got in raw Java 
might be better, at least in terms of garnering contributions.

I'm not particularly interested in a dynamically-typed SSA library since 
correctness is critical in 
the SSA, and if I didn't have the type safety, I'd want to implement it.  
Although I could be sold 
another direction on that point.

I'm also less excited about CPS, since that's going to tend towards optimized 
code resulting in 
non-recursive function tail calls, which I'm trying to avoid until we get a 
version of the JVM which 
can optimize them.

~~ Robert Fischer.
Grails Training        http://GroovyMag.com/training
Smokejumper Consulting http://SmokejumperIT.com
Enfranchised Mind Blog http://EnfranchisedMind.com/blog

Check out my book, "Grails Persistence with GORM and GSQL"!
http://www.smokejumperit.com/redirect.html


Charles Oliver Nutter wrote:
> I'm looking to do a transformation of JRuby's AST into SSA or CPS so I 
> can more easily apply optimizations to it. Currently I'm walking the AST 
> and gathering what info I can, mostly based on isolated nodes. It seems 
> that an SSA graph would be easier to inspect, and I'd get more 
> optimizations almost for free.
> 
> My questions are:
> 
> * Anyone know of a compiler toolchain for JVM which accepts an SSA 
> graph, or tools that might help with this process?
> * Does this sound like a good idea? Perhaps there's a shared library 
> possibility here?
> 
> - Charlie
> 



--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "JVM 
Languages" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at 
http://groups.google.com/group/jvm-languages?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to