> While playing around with your example though, I found that in a more
> 'practical' loop, like 1-4, Slick is actually much faster (running
> Chrome).
If you're still using the tiny test XML, that's still not "practical."
The high number of iterations was one way to amplify the performance
differences. The other way is to use a real XML file of some depth and
breadth and run with far fewer iterations.
Here we only run 100 times over a 256K XML file.
http://jsfiddle.broadleafsystems.com/FrTQc/41/show/light/
The differences are very dramatic, it's not wasting single-digit
milliseconds in this scenario but rather almost 1 to 5 full _seconds_.
In my FF, I get 0-1ms times for native getElementById yet ~1 second
times for Slick. In IE, it's 500ms times for native selectSingleNode
and ~6 seconds for Slick!
So I don't consider it premature nand/nor micro-optimization.
Especially when you take JS' single-threading model into account. It's
unlike other deployments where slow-running code can be left to crawl
along, or even lock up one CPU leaving other CPUs free. An
optimization that would be crazy-talk in a traditional desktop
environment can be quite reasonable in the browser.
And consider web apps that use XML as a "data island" to frequently
run complex lookups against the same doc, say online spreadsheets and
suchlike (after all, if we have a need for getElementById outside of
the DOM at all, the implication is that XML is not being used simply
as an opaque transport mechanism for HTML tags). I wouldn't want to
know I'm spending 70ms per lookup if I can get it down to 5ms pretty
easily.
True, you could make the point that optimization is unnecessary you're
just running getElementById a couple of times over the page's
lifetime. But even in a one-page site I just built, I pull in like 200
records over XML and query across them before insertion (because I
have to massage certain stuff as I don't own the back end). I wish I
could be assured of the best performance possible.
-- S.
--
---
You received this message because you are subscribed to the Google Groups
"MooTools Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to mootools-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
