Comments on: What we have, what we want, how to get it http://tream.dreamhosters.com/lisp/what-we-have-what-we-want-how-to-get-it.html Just another blog about programming, Lisp and the like Mon, 22 Jun 2009 09:33:57 -0700 http://wordpress.org/?v=2.8.4 hourly 1 By: Juan Jose Garcia Ripoll http://tream.dreamhosters.com/lisp/what-we-have-what-we-want-how-to-get-it.html/comment-page-1#comment-30 Juan Jose Garcia Ripoll Fri, 08 May 2009 09:07:20 +0000 http://tream.dreamhosters.com/?p=10#comment-30 There are two goals that motivated this work. One was to have a more robust, simpler and better coded compiler. The other one was to eventually target LLVM as backend, which is precisely what you suggest. There are two goals that motivated this work. One was to have a more robust, simpler and better coded compiler. The other one was to eventually target LLVM as backend, which is precisely what you suggest.

]]> By: Jason Nielsen http://tream.dreamhosters.com/lisp/what-we-have-what-we-want-how-to-get-it.html/comment-page-1#comment-29 Jason Nielsen Thu, 07 May 2009 20:57:54 +0000 http://tream.dreamhosters.com/?p=10#comment-29 Very cool! Have you thought about perhaps using llvm (http://llvm.org/)? If you can reduce things to SSA and do some type inference it would be pretty easy to emit llvm ir and get good performance. I've been playing around with llvm recently and it does an amazing job optimizing the ir when you run he full suite of opt passes at it. Cheers, Jason Very cool! Have you thought about perhaps using llvm (http://llvm.org/)? If you can reduce things to SSA and do some type inference it would be pretty easy to emit llvm ir and get good performance. I’ve been playing around with llvm recently and it does an amazing job optimizing the ir when you run he full suite of opt passes at it.

Cheers,
Jason

]]>