Rhino is About to Get a Lot Faster

infoq.com

22 points·bokchoi··6 comments

They mention work on getting things inlined and generally making calls faster but the article is surprisingly free of any real information on what they are actually doing.

Try this, try that, [laundry list of VM techniques]. we'll make it faster somehow. Just wait!

(snooze)

I do find Charlie's jab at end-to-end JS apps amusing though. It's the kind of comment I'd expect flame wars to emerge from on most forums.

Well the answer is "InvokeDynamic" of course :) FWIW I've heard that term brandied around for 3-4 years at this point, and I still have no idea what it means, and I'm a VM implementer. Personally, my feeling is that implementing a language on the JVM basically involves learning far too much about the JVM internals, and then changing your language significantly to optimize for that. But then, I have a competing horse in the race :)

Invokedynamic basically gives you access to the same kind of optimizations the JVM does for virtual method invocations except you do the method resolution. John Rose has some great articles on it.

Seems like a no-brainer given that a lot of the JIT and tracing techniques from the JVM were taken from pioneering VM research in Self, a language which also pioneered the prototype-based OO for which JS is famous.

I think it's great that there more end-to-end JS options emerging. JS is in many ways just as sane and capable to write server-side code as Python, Ruby, or PHP. It is obviously capable client side because it's basically the language that powers Adobe AIR apps. Funny enough, Javascript is about the only common language that every major tech company can get behind.

this feels like it might be fueling the flame war from the below post. Even though I work with Node all day.

Reply on news.ycombinator.com