2003-12-10, 12:37 | Link #1 |
Administrator
Administrator
Join Date: Jan 2001
Location: Netherlands
Age: 45
|
HTML experiment
I'd like to run a small HTML experiment. Right now I'm using PHP to generate the
HTML that displays the large tables full of links you see on the site. The result of this is a lot of HTML which in turn uses a lot of bandwidth, eventhough a lot of the elements are re-used (html and links). By using JavaScript to generate the pages I'm hoping to very significantly reduce the bandwidth usage of the site, which, just to give you an example, was up to 20 Gb on December 4th alone. Now the problem with this is that I'm asking Javascript to do a lot, and before I actually go and modify the scripts, it might be useful to test first if it is even feasable. I've created a test script which does nothing more than generate a table of 1000 rows. It takes about 1-2 seconds to display on my notebook, which is quite accepitble, certainly as no page on AnimeSuki is actually showing more than 150-250 rows max. But... my notebook has a Athlon XP 2000 CPU and I'm wondering what would happen on systems that are a bit slower than that. Now of course on 486's etc it will probably take much more time to generate the page, but then again, those systems can't be used to view anything that can be downloaded of AnimeSuki anyway. Also how acceptably slow a page loads appears to depend on the browser. On the browser I've tested, Mozilla appears to be the slowest. IE is at an acceptible normal speed and Opera (v7) and Safari (Mac OS browser) are quite fast. You may get the following message when using MS Internet Explorer 6.0: A script on this page is causing Internet Explorer to run slowly. If it continues to run your computer may become unresponsive. Do you want to abort the script? When you see this message, hit Yes and please report in this thread the row number of the last row that was displayed. On my notebook it is "row573" right now, but I've got a feeling this message may depend on the speed of your computer. Anyway... enough talk. The URL of the test page is: http://animesuki.mangaitalia.com/javascript.html As you can see it generates quite a huge page. But if you look at the source code and at the filesize as reported by IE -- the page is less than 1200 bytes. So what do I want exactly? Well, if you want to help me out, please test this on: * Any browser you may have installed on your PC that supports JavaScript to see if it works * See how fast the page is actually generated -- no stopwatch required, but did it take very long, or is it an acceptible delay? * If you use MSIE 6.0 and see the message mentioned above: mention row number please. |
2003-12-10, 13:22 | Link #2 |
…Nothing More
Administrator
Join Date: Mar 2003
Age: 44
|
FireBird - 0.6.1 - It works fine and loads very fast for me; this is the browser I normally use for the site and forum too, so no problems here.
Opera - 6.01 - Very slow, 2 minutes 4 seconds and I'm still waiting for the page to stop "loading". IE - 6.0.2 - Average speed, acceptable. Stops with the script warning between ~400 and 800 depending on what other stuff my computer is doing. FYI: Tested on Athlon, 1.4 GHz, 1Gb Ram with a normal selection of applications loaded and in use. Edit: Opera had loaded the page by the time I completed my post, but the speed is still unacceptable at > 60 Seconds. Edit 2: Ever considered checking the version of the browser and sending an XML page with associated stylesheet for the formatting? Thats got to be cleaner and quicker than JavaScript... Last edited by NightWish; 2003-12-10 at 13:45. Reason: Update... |
2003-12-10, 13:53 | Link #4 | |
Don't use animesuki now..
Join Date: Nov 2003
Location: London, UK
Age: 34
|
Quote:
EDIT i just tested it again and this time it said row486 ... is it suppsed to change?! ^_^ P.S i use internet explorer on windows ^_^ |
|
2003-12-10, 14:00 | Link #5 |
r00t for life
Join Date: Jun 2003
Location: /dev/null
|
row999 and 1.80123 sec of loading time on Opera 7.23. I myself HATE JavaScript coz I sometimes use elinks to get torrents from AS and if you'll be using JS that couldn't be possible. To make bandwidth usage smaller I suggest using gzip compression of HTML documents with mod_gzip (if Apache 1) and mod_deflate (if Apache2).
It seems that www.animesuki.com uses Apache/1.3.29 (Unix) PHP/4.3.4. I suggest installing mod_gzip and enabling compression on all text based documents. That will save you about 30-60% of bandwidth depending how much CPU power you're ready to donate (I suggest using level 1 compression since it requires almost no additional CPU power). Also how about random redirect to some mirror? T:Forse P.S. I tested the JS page with opera 7.23 in windows on my work machine ( 1gig DDR400 and AMD XP 2200+ ) |
2003-12-10, 14:21 | Link #6 |
Afflicted by the vanities
Fansubber
Join Date: Feb 2003
Location: Fish-shape Paumanok
Age: 36
|
I am using Firebird at school, but my Firebird .exe file is remotely stored on my school's student drive, rather than on this computer. It took quite a while to load (I thought it had frozen) but it displays just fine.
__________________
|
2003-12-10, 14:26 | Link #7 |
nerfelwap
Join Date: Feb 2003
Location: Richmond, VA, USA
|
A quick test with gzip shows that the main page would go from ~55k to ~8k with gzip compression. (85% compression, that's very good.)
I tested the JS with my mozilla, and it worked in 4-5 seconds, but it locked up the browser entirely for that time, (I use many tabs, so I noticed it. very irritating) -- -billy- |
2003-12-10, 14:39 | Link #8 | |
r00t for life
Join Date: Jun 2003
Location: /dev/null
|
Quote:
|
|
2003-12-10, 15:06 | Link #10 |
Administrator
Administrator
Join Date: Jan 2001
Location: Netherlands
Age: 45
|
Damn. Forgot about Gzip. Of course, then I can just forget
about JavaScript AND it will keep loading without problems on loads of browsers without problem. But... installing Gzip is not up to me -- Mb81 has to do it. I think that at first Gzip was enabled, but it caused severe trouble with caching images (that is, Gzip caused nothing to be cached, which is a pain when images have to be reloaded with each new pageview). With this, this experiment is over until futher notice. Thanks for participating, and sorry for "waisting" your time |
Thread Tools | |
|
|