Re: [progress-events] stalled, lengthComputable, loadstart vs begin

On Tue, 31 Jul 2007, Charles McCathieNevile wrote:
> > > 
> > > * HTML 5 has an event called "stalled" that is dispatched after 
> > > there are three seconds of no progress at all so you do not have to 
> > > create your own timer scripts.
> > 
> > 3 seconds of no progress might not be out of the ordinary depending on 
> > connection speed and distance to the host. This seems pretty 
> > arbitrary.
> 
> I agree with Maciej on this one. I raised ISSUE-117 for it, but my 
> proposal is that we do not adopt it (and that we point out to HTML-WG 
> that 3 seconds is very arbitrary). This would be consistent with how we 
> dealt with ISSUE-107...

To clarify -- in the HTML5 spec the three seconds is indeed arbitrary, the 
spec just says "about three seconds" and it is expected that user agents 
will adapt that as appropriate based on the connection speed, distance to 
host, and so forth.

Furthermore, in the HTML5 case it's specifically for streaming multimedia, 
where the user does care about a few seconds with no buffering going on 
since it can mean that the media will skip.

I agree that it may well be inappropriate to have this event in the 
progress events metaspecification. If it's included at all, it should 
probably be listed as something that specifications should only include if 
it is considered useful for that particular case.

HTH,
-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Wednesday, 1 August 2007 00:30:54 UTC