[Date Prev][Date Next][Thread Prev][Thread Next]
- From: David Given <dg@...>
- Date: Wed, 03 Sep 2008 22:02:45 +0100
> twins to me. The syntax is different but the basic idea is very
> long time I have said that it would be incredibly hard or impossible
> define "fast" as faster than Lua). If someone manages to pull it off
> I will be impressed.
I've actually thought that it would be very interesting to build a
enough semantically that the mapping is probably quite easy.
parse with traditional systems like Yacc --- semicolon insertion
algorithms, yuck. I'm not even sure it can be simply parsed using
recursive descent parsers. Either way, not fun.
If we can come up with a JS implementation based on Lua that's got
competitive performance and is also smaller than the alternatives, it
would probably prove something-or-other...
Just out of interest, TraceMonkey (thanks for the pointer) gets a score
of 6.6 in my benchmark which is dismal --- but as the score goes *up* to
17 if I turn the JIT *off*, I suspect I'm simply not driving it properly.
┌─── ｄｇ＠ｃｏｗｌａｒｋ．ｃｏｍ ───── http://www.cowlark.com ─────
│ "All power corrupts, but we need electricity." --- Diana Wynne Jones,
│ _Archer's Goon_
Description: OpenPGP digital signature