Ringobot log - RingoJS IRC channel: #ringojs on irc.freenode.net
2010-05-04:
[16:28] <robi42> fyi: just received lots of feedback from a guy who attended vienna.js, i'll wrap it all up a little and put on a wiki page[16:31] <robi42> btw, he's doing http://letsannotate.com/ and very interested in using ringo's cometd bayeux server wrapping package
[18:23] <hannesw> looks interesting
[18:26] <hannesw> i'm not sure if i should publish early benchmarks from my rhino companion closures branch
[18:26] <hannesw> numbers are breathtaking
[18:27] <earl> good to hear :)
[18:31] <hannesw> but it's only half done, many things not working
[18:31] <hannesw> so neither ringo nor v8-benchmark runs
[18:35] <hannesw> but on but on a simple closure benchmark such as this:
[18:35] <hannesw> http://ringojs.pastebin.com/xk5zepmG
[18:35] <hannesw> companion-scopes branch is more than 5 times faster than old rhino, and less than 5 times slower than v8
[18:36] <hannesw> (I'm careful, actual numbers are better)
[18:37] <hannesw> access to "x", "y", and "inner" is now a single JVM GETFIELD instruction rather than a hashtable lookup in a scope chain
[18:42] <hannesw> well i think i'll get ringo and possibly v8-benchmark running, and then do some benchmarking :)
[20:00] <earl> mhm, mhm
[20:00] <earl> looks good
[21:34] <earl> hannesw: does it run the rhino testsuite yet?
[21:34] <earl> looks very interesting, in any case
[21:35] <earl> from a few superficial looks, it reminds me of "flat closures" for scheme
[21:37] <earl> ah, if only i knew rhino a bit better :)
[21:38] <hannesw> no, it currently breaks on destructuring assignment
[21:38] <hannesw> because they seem to be using with() {} internally
