I only saw an error caught and reported in a pink box (with no
pertinent info) on loading the first test (and only the first time).
The site doesn't seem to be recording results for FF 10 either.
Here is a typical result from FF 10:-
Chrome 13.0.782 2,885,834 898,102 78,951 1
Chrome 14.0.835 1,622,298 434,627 25,368 3
Chrome 15.0.874 2,450,642 807,696 68,140 3
Chrome 17.0.963 3,155,691 1,474,550 101,658 3
Firefox 6.0.2 1,129,588 435,858 32,674 3
Firefox 8.0 3,135,401 913,051 54,460 2
Firefox Beta 12.0a2 1,874,259 500,396 49,073 1
IE 9.0 720,506 375,765 3,689 1
Opera 11.51 340,953 201,199 2,333 1
Safari 5.1.2 6,167,981 1,614,097 114,380 1
http://jsperf.com/removing-element-attributes-optimized
Should be apparent which column is which. Safari 5 is over 6
*million* attributes removed for the cross-browser rendition. My
Library comes in second with over 1.6 million. jQuery is just over
100 thousand. Think of those numbers in terms of dollars. Not just
for effect, but those seeking to create the next "Facebook Killer" (or
a basic Website that is responsive on battery-powered mobile devices)
should realize that the jQuery authors have been asleep at the switch
for a long time (and this is the latest version, after numerous well-
publicized "optimizations" that reportedly increased performance
hundreds of times over).
Then there's the long since conceded point that jQuery is terrible at
attribute manipulation (i.e. it comes up with a lot of wrong
answers). Granted, most jQuery users don't understand that issue
(which is largely confined to IE 6/7) or how it can affect them.
And, of course, most of the jQuery examples are neither concise nor
easily understood at a glance.
Meanwhile, the "fast, concise, etc." line continues to be parroted by
neophytes who never tried anything else. Show them tests like that
and they'll show you an even less readable example (a lot of $.fn.*
calls) that is slightly faster. Then they'll blither about early
optimization, saving developer time, etc. Of course, a site is
developed once but used many times, but then Web developers generally
have disdain for their users and consider their time far more
valuable. Add up all of the time spent "upgrading", retesting,
redeploying, etc., plus the inevitable cost of pissing off users with
slow or broken pages and all of those "arguments" fall apart.
So how does this please so many? Marketing, plus pigeonholing all
scripts as either "perfect" (seen as impossible) or "imperfect" (all
scripts then) and mistaking empirical evidence (e.g. works today in my
installed browsers!) for proofs. Looking at the code is seen as a
waste of time as how many users are going to look at the code? :)
I've been saying the same thing since 2007 (when I first encountered
the script and before I wrote the "competing" library) while most
others took a "give it a few years and they'll get it right" stance
(which I imagine most regret at this point).
It was never about writing the most efficient and "perfect" code or
patching the most "cross-browser" issues. It's always been about the
reaction I got from the authors (and their fans) after that first
review. That alone told me they'd never get anywhere (and the same
for Prototype, MooTools, etc.) On the contrary, many argue that
convincing lots of beginners to adopt it as a "standard" is getting
somewhere, but it's not anywhere I ever wanted to go and hasn't
improved the code logic or performance a whit.
Seeing them in action (acting on tips I sent over) further convinced
me that they just didn't understand what they were doing and were
virulently opposed to learning (too busy bringing lousy code to
market).