I just randomly came across this Ajaxian podcast episode from a while back that answers the very question I posed.
http://ajaxian.com/archives/audible-ajax-episode-20-project-tamarin Interestingly, Adobe open-sourced their JIT implementation of Javascript after a 3 year complete rewrite, and that's what Mozilla, Adobe, and others are collaborating on for the Tamerin Project. Some say it can increase speeds up 1ox. While Tamarin won't be in Firefox 3, it will be a later addition, likely Firefox 4 sometime in late '08. In typical Microsoft fashion, they are currently rewriting their own Javascript engine (JScript) and they intend to match, or exceed Tamarin's speed. Why doesn't MS just use Tamarin as well since it is open-source? Here's one instance where I'm glad MS is sticking to their own sandbox cause it just means more competition. If they can't beat Tamarin when they can look at Tamarin's code as much as they want, that's pretty sad. So, to completely answer my question, it looks as if the interpreter is the current bottleneck, but one interviewee said the bottleneck for most AJAX apps will be network speed in the near future. On 10/2/07, Robert Koberg <[EMAIL PROTECTED] > wrote: > > > On Tue, 2007-10-02 at 11:26 -0500, Derek Gathright wrote: > > Thanks for the link, interesting stuff. > > > > After looking through info on Rhino, I was left with the question... > > why build the JS core engine in Java and not a non-interpreted > > language? > > You can compile them to byte code and create classes (I am guessing that > is why rhino performs much better in the 'real world' test cases). I use > them for a webapp in a servlet container. > > best, > -Rob > >