Firefox 3 is now using expiration policies in its memory caches. The developers' thinking is that if you haven't retrieved a previously viewed page in half an hour or so, the savings in memory by dropping the page from your cache are more important than the small possibility your page will load faster if you retrieve the stale document. (For more explicit information on how memory issues have been tweaked in Firefox 3, a good resource is this blog entry by Mozilla developer Stuart Parmenter.)
The result is that, regardless of any other improvements, Firefox 3 is faster and more stable than its predecessor. I found that, on average, opening and closing tabs on Firefox 18.104.22.168 used up about 5 per cent more RAM per browser tabbing session compared with Firefox 3. And in the weeks I've been running Firefox 3 on multiple systems on the same exact same PCs doing the same work as I was doing with Firefox 2.x, I haven't seen a single freeze-up.
Besides the memory improvements, I found Firefox to be both faster and more stable than its predecessor for other reasons. Thanks to the vastly improved Gecko 1.9 Web rendering platform, Firefox makes complex pages -- like Computerworld's own front page, with its text, graphics and animations -- pop, rather than be painted, on the screen.
To test that, I looked at a group of pages, first on Firefox 22.214.171.124 and then on 3.0. (In all cases, I cleared the cache first.) I saw a 20 per cent to 35 per cent reduction in the time from when a Web page was summoned to when it completely appeared on the screen.
Here Firefox 3 scored 71, which isn't exactly a prize-winning rating; the latest version of Safari for the PC, Safari 3.1.1, scored 87. On the other hand, the other browsers I tried, such as IE 7 and IE 8 beta 1, turned in even worse results. Since Safari's security could well be described with the word awful, I'll stick with Firefox.
For practical purposes, the only Web pages that are likely to give you trouble are the same ones that always have: Web pages that were designed specifically with Internet Explorer and ActiveX in mind. But I wouldn't worry too much: In the months I've been using Firefox 3, first as a beta and then as a release candidate, Firefox had no trouble rendering any of the thousands of Web pages I visited.
Often, when it comes time to look at a product's new features, I end up writing what amounts to a laundry list of functions that no one is ever likely to use in the real world. That's not the case with Firefox 3.