Why don't they just use v8, tracemonkey, or squirrelfish? It's nice they continue to develop their own technologies (opera is awfully nice on cell phones), but, in this case, writing a js engine is a problem with lot's of pre-existing high-quality solutions with very active developer communities.
Opera writing their own next-generation js engine smells a little bit too much like not-invented here syndrome. Considering how small their market share is, it seems like this money could be much better spent elsewhere.
Maybe for garbage collection or some efficiency reason. Walter Bright had a short write up about shared gc and dlls somewhere in the D site. I'm not sure how much of that would apply to Opera, though.
I personally like that they are creating a new implementation. I suppose that understanding the code and having full control over its direction is a big thing for them.
Sometimes, writing your own is the best way to understand the problem.
Side note: I think people said the same thing about Google developing v8. After a few months, people will accept the new engine (if it works) and it will go on to the next engine.
> we are implementing compilation of whole or parts of ECMAScript programs and functions into native code.
This seems to be a new, popular approach to JS optimalization. For an open-source implementation it's a reasonable approach, but for a proprietary system it usually just delays builds for new architectures.
I think what he is inferring is that it will tend to be delayed on alternate architectures because the company puts manpower on things that will deliver the most bang for the buck(i386, ppc maybe). When it is opensource the people that are affected by the delay have the opportunity to help out.
Its great the opera mini gave my old cell phone a feel of a smart phone. And opera mobile supports flash.