It's ironic that Grub returns nothing but sites that want to sell me jerseys when I search for an NFL player, whereas the top Google links are always the player profile pages from NFL.com and other major sports sites, and the player's Wikipedia entry. Add to that the easy access to "news" for the player and there's little question which search engine is the more useful.

Google works because their ranking system works. If it stopped working well, they would lose market share very quickly.

This is probably due to sqlite and a severely fragmented / huge / whatever history+bookmarks.

Try a clean profile. If that does the trick, try backing them up and importing in a clean profile. Probably "prune" them a bit while you're at it.

h264 isn't going to be a practical problem for the vast majority of users, since Firefox can just use a system codec (non-Windows-users would have to make sure they have one, of course).

As for JS speed, Mozilla are very ardent in their speed claims, so it's hard not to believe they have something to back it up. It's difficult for users and external testers to figure out exactly how fast they are, despite being open source, because the Moz team is pursuing several parallel tracks to increase JS speed. There's "fat-val", "tracer JIT" and "method JIT". Each is currently significantly faster than the "normal" versions, but there hasn't been any public testing on a build that combines all three.

Mozilla claim they'll be faster than everyone else and while they may be scuppered by new advances from Google and Opera, it seems reasonable that they will at least be faster at launch than where everyone else is now. That alone would ensure "next-generation JS performance".

Where they trail Chrome is in "use speed". Chrome starts and shuts down a lot faster -- and I think that's going to be a problem for Firefox moving forward (more than it already is).

Spoofing as Firefox 3 on Vista worked for me.

Specifically:
Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.0.6) Gecko/2009011913 Firefox/3.0.6

With Novell Moonlight 2.3.

The idiots behind the site are using OS detection, so if you're using Moonlight on a non-Windows/OSX platform, you'll need to spoof your User-Agent string.

Other than that, it works just fine with Firefox & Moonlight on Linux.

They are complementary only in a way icecream and fish are complementary. They don't seem to have overlap, but you don't want fish with your icecream.

Since bunratty's reply provides a technical description of why they are truly complementary (possibly even orthogonal), I'm going to believe that argument over your rather weak ice cream simile.

I think you're missing the point of what is being benchmarked. Mozilla hasn't released benchmarks of their new JS engine with both "method" and "tracer" JIT combined. They are being evolved separately, but are (according to Moz) complementary. Thus, we don't know how far they actually are from their goal yet.

Check out http://www.arewefastyet.com/ for benchmarks and description.

From what I can gather from the associated bug report, the "fatval" optimizations are also not applied to the portions of JS code that is traced... which would imply that the better job the tracer engine does, the less the "fatval" optimizations are applied.

The result is that an unknown "free" speed increase is waiting in the wings. What the magnitude of this increase is... well, that's the question, isn't it?

Does 1 September seem like a really tight deadline? Yes, sure does, but more in terms of stability and robustness than actually getting to a specific speed milestone.