From: Charles O. N. <cha...@su...> - 2006-11-14 03:41:15
|
Samuele Pedroni wrote: > yes, shipping is the most important thing, together with compatibility > and catching up. > Then, in my opinion some aspects of the code base (especially related to > java integration) > need some cleanup before we can even think of working on improving > performance > further, unless we want to risk even less readable, maintainable code. This is also a very important area for JRuby. We're starting to scope out what a 1.0 release should look like, and having a complete store for Java integration syntax and semantics is certain to be finalized by then. There's a lot to be gained by agreeing on similar syntax and semantics in both Jython and JRuby so they end up "feeling" right side-by-side. We can both learn from what each others' impls are doing today and come up with a common path for the future (especially considering Java 5's many new language elements). > As I described I was blocked by having to fix the code and > .expose/.derived files being out of sync > and related infinite recursion bugs in the new-style classes code. I may > still get to it, but I cannot promise > anything specific. PyPy has its own deadlines and I'm "just" back from > quite a bit of work travel (OOPSLA > in Portland, and one week of PyPy sprinting in Duesseldorf). Btw someone > has started a JVM backend > for PyPy, this may be quite useful but only long term. Ah, so you are one of the PyPy folks? Very good...I believe a PyPy kernel for the JVM would be an outstanding item to work on in the future. I am interested in doing the same for JRuby, possibly based on Evan Phoenix's Rubinius. However as you say they are more long-term goals, and given that the JVM already provides most of the VM features a PyPy or Rubinius would implement anew, there's not yet a compelling motivation for dropping our "thinner" JVM-based implementations of either language. The challenge here is to leverage as much as possible of the JVM's own capabilities while keeping the dynlang experience true. I think we can do that, and there's still a window open for even better dynlang support under Java 7. -- Charles Oliver Nutter, JRuby Core Developer Blogging on Ruby and Java @ headius.blogspot.com Help spec out Ruby today! @ www.headius.com/rubyspec he...@he... -- cha...@su... |