James, First off, nice work. I think it is important to have tests like slickspeed but its also important not to get caught up in the results. I keep an eye on the the "orange" I really don't like the thought of different matches across different frameworks.
The biggest performance increase anyone can obtain is switching to a modern browser. Running these tests in IE6 & 7 runs multiple times slower regardless of the framework. With that said, Ext seems to be the fastest in IE. YUI and Prototype are really hurting when running in IE. The tests that slickspeed performs often differs from real world usage on several levels. 1.) Document size. 2.) The number of selections 3.) The lack of repetition of the same selector (minimizes the effectiveness of caching) 4.) Dom updates prevent some of the caching techniques as well. Nathan On Tue, Oct 28, 2008 at 2:51 PM, Guillermo Rauch <[EMAIL PROTECTED]> wrote: > Peppy in my tests was noticeably faster than MooTools, jQuery and Dojo. > Only rivaled by Sizzle followed by Ext. Nice work > > On Tue, Oct 28, 2008 at 4:40 PM, JamesDonaghue <[EMAIL PROTECTED]>wrote: > >> >> Peppy doesn't cheat, it was a bug (in IE only) and it has been fixed >> in the 0.1.2 release: http://jamesdonaghue.com/static/peppy/. >> >> Also, if selector engines speeds aren't important then why even have a >> tool like Slickspeed? >> In fact why bother developing a selector engine at all if it is so >> rarely used or is so unimportant? >> >> The selector engines that exist are already fast, however it has been >> shown that they can in fact be faster. I don't understand the >> argument against this. >> The importance of web applications is very great and anywhere that we >> can achieve speed gains we should. >> >> >> On Oct 22, 12:27 pm, davidwalsh83 <[EMAIL PROTECTED]> wrote: >> > I agree with Bango. Is there really a difference between 2 and 5 >> > milliseconds? They're all fast and what should be avoided is a >> > "pissing contest." >> > >> > Great work Tom, Jan, and the rest of the team! >> > > > > -- > Guillermo Rauch > http://devthought.com >
