[Kss-devel] Another CSS selector test

[Godefroid Chapelle]

Alec Mitchell wrote:
> This is indeed very useful, it's pretty clear that cssQuery is pretty
> sub-optimal.  I got similar results with mootools and prototype nearly
> tied in FF (in both cases mootools was slightly faster though), with
> dojo and ext 3-4x slower, and jQuery and cssQuery >15x slower.
> 
> In Safari, he results were a bit different.  Ext was the fastest, only
> slightly slower than it was in FF, dojo was 2x slower than than,
> mootools and prototype 3x slower, and  jQuery and cssQuery again
> unusable at >7x slower than ext.
> 
> I have no idea what this all means, and its interesting how different
> the relative perfomance of these libraries is across js
> implementations.  The only obvious conclusion is that cssQuery is a
> complete dog, and jQuery should be avoided if selector performance is
> a consideration (for KSS it's one of the primary considerations, IMO).
> 
> Alec
> 
> On 6/13/07, Jeroen Vloothuis <jeroen.vloothuis at xs4all.nl> wrote:
>> Just found a link to another CSS selector test. This might be
>> interesting because the CSS selector is one of the main performance
>> bottlenecks.
>>
>> http://mootools.net/slickspeed/
>>
>> The nice about this one is that you can test it in various browsers on
>> your own system. Unfortunenately it does not feature base2 which is
>> supposed to be a lot faster than cssQuery.
>>
>> In my case prototype was the fastest closely followed by mootools.
>>

I saw the same results on my FF under Mac OS.

The interesting stuff is that the code of slickspeed is actually pretty 
pluggable. IOW, it should be pretty easy to add base2 to the tests in 
order to evaluate it as well.

PS I never ran any PHP app so I have no idea how complex it would be to 
make a setup where we could plug base2.

--
Godefroid Chapelle (aka __gotcha)- BubbleNet  http://bubblenet.be