[jQuery] Re: Drastically reducing jQuery load time
Just use the minified version and send it over the wire gzipped and be done with it. The eval of packed jQuery does take some time, so even if the file is cached it probably will be slightly slower than the cached minified variation. Check this out: http://www.julienlecomte.net/blog/index.php?s=jquerysbutt=Go Eric On Nov 4, 2:13 pm, S. Robert James [EMAIL PROTECTED] wrote: While working on a web app, I thought of this method to drastically reduce jQuery load time. Every link follow causes jQuery to be reloaded again. Now, although the .js file is stored in the local cache, it still needs to be parsed and executed, which takes a lot of time. What if, instead, we use a FRAMESET, keeping jQuery in the master frame, and just changing a child frame (which takes up the entire screen). This way, jQuery never needs to be reloaded. What does everyone say?
[jQuery] Re: Drastically reducing jQuery load time
I would need to see some actual stats as to the performance hit of the jQuery file loading itself into memory before I had this concern. My personal experience, even with very slow computers, is that I haven't seen a lag when loading a cached page with a cached jQuery file into memory. I have no doubt that the solution you propose would improve performance, but if it went from 0.032 seconds to 0.0012 seconds, I can't imagine that would be worth using a frameset and the other hassles and problems it could cause. If you are using the packed version of jquery which uses this sort of setup: eval(function(p,a,c,k,e,r). You could instead use the minified/gzipped method, which will reduce any eval overhead (in additional to making a smaller download). JK -Original Message- From: jquery-en@googlegroups.com [mailto:[EMAIL PROTECTED] On Behalf Of S. Robert James Sent: Sunday, November 04, 2007 11:13 AM To: jQuery (English) Subject: [jQuery] Drastically reducing jQuery load time While working on a web app, I thought of this method to drastically reduce jQuery load time. Every link follow causes jQuery to be reloaded again. Now, although the .js file is stored in the local cache, it still needs to be parsed and executed, which takes a lot of time. What if, instead, we use a FRAMESET, keeping jQuery in the master frame, and just changing a child frame (which takes up the entire screen). This way, jQuery never needs to be reloaded. What does everyone say?
[jQuery] Re: Drastically reducing jQuery load time
On Nov 4, 3:04 pm, Jeffrey Kretz [EMAIL PROTECTED] wrote: I would need to see some actual stats as to the performance hit of the jQuery file loading itself into memory before I had this concern. Is there a good way to profile page load, and determine how much time is spent parsing/executing jQuery, how much is spent on CSS, how much is spent on screen paint, etc.? (This would have tremendous use outside of this particular question...)
[jQuery] Re: Drastically reducing jQuery load time
Hi Robert, You can try the YSlow extension. It piggybacks off of the Firebug extension for Firefox: http://developer.yahoo.com/yslow/ --Karl _ Karl Swedberg www.englishrules.com www.learningjquery.com On Nov 4, 2007, at 3:37 PM, S. Robert James wrote: On Nov 4, 3:04 pm, Jeffrey Kretz [EMAIL PROTECTED] wrote: I would need to see some actual stats as to the performance hit of the jQuery file loading itself into memory before I had this concern. Is there a good way to profile page load, and determine how much time is spent parsing/executing jQuery, how much is spent on CSS, how much is spent on screen paint, etc.? (This would have tremendous use outside of this particular question...)
[jQuery] Re: Drastically reducing jQuery load time
Hi Robert, I doubt the load time of jQuery is very significant. Downloading files take much much longer than it does to execute. Regardless, I personally would not use frames since then the url will never change. ~Sean
[jQuery] Re: Drastically reducing jQuery load time
I just tried looking through the various parts of Firebug, and while it profiles net traffic excellently, I don't see anyway to profile memory/CPU time on the different elements. I did test a script which captures the date/time, then loads the jQuery file, then captures the date/time again and calculates the millisecond difference between them. You can see this (using uncompressed jQuery): http://www.scorpiontechnology.com/Cobalt/speedtest.htm After clearing the cache and visiting the page (reflushing the cache each time), I got these results: 1078ms 1109ms 1078ms 1078ms Subsequence hits to the page (using the retry link) were: 16ms 15ms 0ms 16ms I then made another page that uses the packed version of jQuery which has to be eval'ed on load. http://www.scorpiontechnology.com/Cobalt/speedtest2.htm Initial load test: 641ms 656ms 656ms 640ms Subsequent visits using the retry button (with cached files): 47ms 32ms 46ms 47ms So there is definitely some overhead evaling the packed version of the script, on the order of magnitude of about 0.03 seconds. Then, for poops and giggles, I tried the same thing with a minified gzip jQuery: http://www.scorpiontechnology.com/Cobalt/speedtest3.htm Initial load test: 485ms 468ms 469ms 562ms Subsequent cached loads were: 15ms 16ms 0ms 16ms (identical to the initial uncompressed jQuery test). I'd be curious to see if others get similar results, but for now, I would say that using the minified/gzipped solution will result in the least amount of network overhead (duh) without the (albeit minor) performance hit of the packed version. And on the original point about framesets, 0.016 seconds savings in time is not worth the problems inherent in a frameset-based website IMHO. JK -Original Message- From: jquery-en@googlegroups.com [mailto:[EMAIL PROTECTED] On Behalf Of S. Robert James Sent: Sunday, November 04, 2007 12:37 PM To: jQuery (English) Subject: [jQuery] Re: Drastically reducing jQuery load time On Nov 4, 3:04 pm, Jeffrey Kretz [EMAIL PROTECTED] wrote: I would need to see some actual stats as to the performance hit of the jQuery file loading itself into memory before I had this concern. Is there a good way to profile page load, and determine how much time is spent parsing/executing jQuery, how much is spent on CSS, how much is spent on screen paint, etc.? (This would have tremendous use outside of this particular question...)
[jQuery] Re: Drastically reducing jQuery load time
Firebug even has its own code profiler built in (Profile button on the Console tab), I have used it and YSlow, I think firebugs is better on its own. On 11/4/07, Karl Swedberg [EMAIL PROTECTED] wrote: Hi Robert, You can try the YSlow extension. It piggybacks off of the Firebug extension for Firefox: http://developer.yahoo.com/yslow/ --Karl _ Karl Swedberg www.englishrules.com www.learningjquery.com On Nov 4, 2007, at 3:37 PM, S. Robert James wrote: On Nov 4, 3:04 pm, Jeffrey Kretz [EMAIL PROTECTED] wrote: I would need to see some actual stats as to the performance hit of the jQuery file loading itself into memory before I had this concern. Is there a good way to profile page load, and determine how much time is spent parsing/executing jQuery, how much is spent on CSS, how much is spent on screen paint, etc.? (This would have tremendous use outside of this particular question...) -- http://morglog.alleycatracing.com Lets make up more accronyms! http://www.alleycatracing.com LTABOTIIOFR! ROFL! ROFL! ROFL! Upcoming alley cats, reviews, touring logs, and a general congregation of bike nerdity.
[jQuery] Re: Drastically reducing jQuery load time
I benchmarked also, using a homebrew console.debug(new Date().getTime()), and consistently got about 90ms to load jQuery, plus another 100ms to load all the plugins and custom JS. That's a lot of time - in the sense of a browser based web app that is supposed to be highly reactive. One alternate idea to frames is to load jQuery at the end of the page. That way, the user can see the entire page while jQ loads in the background, See http://developer.yahoo.com/performance/rules.html#js_bottom (thanks for the tip, Karl). On Nov 4, 5:11 pm, Jeffrey Kretz [EMAIL PROTECTED] wrote: I just tried looking through the various parts of Firebug, and while it profiles net traffic excellently, I don't see anyway to profile memory/CPU time on the different elements. I did test a script which captures the date/time, then loads the jQuery file, then captures the date/time again and calculates the millisecond difference between them. You can see this (using uncompressed jQuery):http://www.scorpiontechnology.com/Cobalt/speedtest.htm After clearing the cache and visiting the page (reflushing the cache each time), I got these results: 1078ms 1109ms 1078ms 1078ms Subsequence hits to the page (using the retry link) were: 16ms 15ms 0ms 16ms I then made another page that uses the packed version of jQuery which has to be eval'ed on load. http://www.scorpiontechnology.com/Cobalt/speedtest2.htm Initial load test: 641ms 656ms 656ms 640ms Subsequent visits using the retry button (with cached files): 47ms 32ms 46ms 47ms So there is definitely some overhead evaling the packed version of the script, on the order of magnitude of about 0.03 seconds. Then, for poops and giggles, I tried the same thing with a minified gzip jQuery: http://www.scorpiontechnology.com/Cobalt/speedtest3.htm Initial load test: 485ms 468ms 469ms 562ms Subsequent cached loads were: 15ms 16ms 0ms 16ms (identical to the initial uncompressed jQuery test). I'd be curious to see if others get similar results, but for now, I would say that using the minified/gzipped solution will result in the least amount of network overhead (duh) without the (albeit minor) performance hit of the packed version. And on the original point about framesets, 0.016 seconds savings in time is not worth the problems inherent in a frameset-based website IMHO. JK -Original Message- From: jquery-en@googlegroups.com [mailto:[EMAIL PROTECTED] On Behalf Of S. Robert James Sent: Sunday, November 04, 2007 12:37 PM To: jQuery (English) Subject: [jQuery] Re: Drastically reducing jQuery load time On Nov 4, 3:04 pm, Jeffrey Kretz [EMAIL PROTECTED] wrote: I would need to see some actual stats as to the performance hit of the jQuery file loading itself into memory before I had this concern. Is there a good way to profile page load, and determine how much time is spent parsing/executing jQuery, how much is spent on CSS, how much is spent on screen paint, etc.? (This would have tremendous use outside of this particular question...)
[jQuery] Re: Drastically reducing jQuery load time
On Nov 5, 2007, at 12:11 AM, Jeffrey Kretz wrote: of network overhead (duh) without the (albeit minor) performance hit of the packed version. The performance hit is bigger for clients which have slow refrigerator cpu (read iPhone). -- Mika Tuupola http://www.appelsiini.net/
