W3C home > Mailing lists > Public > public-webapps@w3.org > January to March 2009

Re: [XHR2] Progress events during synchronous requests

From: Sergey Ilinsky <castonet@yahoo.co.uk>
Date: Thu, 12 Feb 2009 10:21:54 +0000 (GMT)
To: Jonas Sicking <jonas@sicking.cc>
Cc: Webapps WG <public-webapps@w3.org>
Message-ID: <566139.60270.qm@web26901.mail.ukl.yahoo.com>

I am sure a developer would like to make use of a consistent API. This would simplify application logic modification: using XHR1 a code would need to be written differently when executing synch/asynch calls.

Also, dispatching only "checkpoints" events would probably be sufficient for that matter.

Sergey Ilinsky/


--- On Thu, 12/2/09, Jonas Sicking <jonas@sicking.cc> wrote:

> From: Jonas Sicking <jonas@sicking.cc>
> Subject: Re: [XHR2] Progress events during synchronous requests
> To: "Anne van Kesteren" <annevk@opera.com>
> Cc: "Webapps WG" <public-webapps@w3.org>
> Date: Thursday, 12 February, 2009, 4:52 AM
> On Thu, Feb 12, 2009 at 1:19 AM, Anne van Kesteren
> <annevk@opera.com> wrote:
> > On Wed, 11 Feb 2009 22:44:17 +0100, Jonas Sicking
> <jonas@sicking.cc> wrote:
> >>
> >> Is there a reason why the spec says to not send
> progress events
> >> (upload and download) during synchronous requests,
> while still saying
> >> that readystatechange events should be fired?
> >>
> >> I guess it can be argued that progress events
> (especially the load
> >> event) is less useful for synchronous loads and
> that not dispatching
> >> them is a performance optimization. Is that the
> reason?
> >>
> >> I don't really feel strongly either way,
> though I do think that from a
> >> web developers point of view it would be more
> useful to dispatch the
> >> events than not to dispatch them. And being
> consistent with regards to
> >> readystatechange is always nice.
> >
> > It is a performance optimization, yes. Firing a lot of
> progress events for a
> > large file that is synchronously downloaded in a Web
> Worker seems like
> > waste. The reason readystatechange events are
> dispatched is because Internet
> > Explorer already did that.
> 
> At least in gecko it is not a big performance win. In fact,
> I doubt
> that it's even measurable. I'd be surprised if this
> was significantly
> different in other implementations.
> 
> That said, I'm personally fine with waiting to add
> these events until
> authors start asking for them. But I suspect they will.
> 
> / Jonas


      
Received on Thursday, 12 February 2009 10:22:34 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 18:49:30 GMT