[Bug 16442] Extreme length of single page spec causes long load time/unresponsive browser, decreases effectiveness of spec.

https://www.w3.org/Bugs/Public/show_bug.cgi?id=16442

--- Comment #4 from Roy <tchalvakspam@gmail.com> 2012-03-20 10:41:27 UTC ---
I decided to run some timing tests, all performed with a pre-primed cache:

Computer specs:
notebook computer (Lenovo ideapad U350 notebook) running ubuntu linux, firefox
10.
This is pretty much the only website that I've seen give this computer trouble.
 It's not a powerhouse by any means, but apart from heavy image processing, it
never complains on any other sites.

Testing http://www.whatwg.org/specs/web-apps/current-work/
Full initial load:
70 seconds
(Shortcut to sectioned spec opened after the first 10 seconds, I ignored it to
see how long the full spec would take to load.)

Loaded page and then clicked through to multi-page version:
60 seconds.
Second load test: 50 seconds.
20 seconds for initial page to load, another 30 seconds for it to respond to
the click and 10 to load the next page.

http://dev.w3.org/html5/spec/single-page.html
Firefox: 46 seconds to fully load.
Second load test: 56 seconds to fully load.
20 sec to display content on page.

Google chrome, 26 seconds to fully load.
Second time through, ~20 seconds to load "slow loading?" window, by the time it
loaded the page was fully loaded.


Droid 2 Global, Android 2.2 browser:
110 seconds to display any content.
~150 seconds to finish loading the page.
(I actually recommend against testing this on mobile, it made mine very
unresponsive)

Firefox loads google.com , in under a second, as usual.  Ditto msn.com.

So what does this suggest to me:  The "loading slowly?" box cuts down on the
overall time slightly, and you get a more usable result out of it, but it's by
no means a silver bullet.  I suspect that there's something going on here where
web pages rarely use so much raw html or something so browsers aren't optimized
for using memory efficiently with raw html, as I can, say, run flash and
javascript games on this machine just fine.  Obviously avoiding that "all the
spec in a single page" load would be ideal, and that may only be accomplishable
via some kind of SEO or use of canonical urls targeted at specific elements.

(These tests were by no means empirical, and obviously this machine isn't a
powerhouse, but I'm sure there are people using worse, and on mobile it pretty
much thrashes the device.)

-- 
Configure bugmail: https://www.w3.org/Bugs/Public/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.

Received on Tuesday, 20 March 2012 10:41:36 UTC