W3C home > Mailing lists > Public > public-tracking@w3.org > October 2012

Re: Multiple DNT Headers (ACTION-283, ISSUE-150)

From: Shane Wiley <wileys@yahoo-inc.com>
Date: Fri, 12 Oct 2012 14:14:33 -0700
To: "ifette@google.com" <ifette@google.com>
CC: Walter van Holst <walter.van.holst@xs4all.nl>, "public-tracking@w3.org" <public-tracking@w3.org>
Message-ID: <E1999CD5-D72E-4252-8451-134C395EED73@yahoo-inc.com>
Ian,

That's fair - instead of "page load" perhaps we say "object load" or something similar.  This would help differentiate initial page state and subsequent activities before another full page load where I'm assuming even widgets are refreshed.

Sent from Shane's mobile

On Oct 12, 2012, at 2:08 PM, "Ian Fette (イアンフェッティ)" <ifette@google.com<mailto:ifette@google.com>> wrote:

I would say the effect of changing mid-session on a long-lived session would be rather undefined.

Imagine you load a webpage that loads a social widget. This social widget has a web-wide exception. It loads all of its application code, and every minute polls its backend (via XHR, for instance) to see if there's anything interesting to show. A half hour later, the user somehow revokes the DNT exception for this widget. The widget is already loaded and is still polling (the user left the tab open with the widget on it), and seeing DNT:0 on the initial request the server sent the version of the widget that does "tracking" (whatever that means, given we haven't defined it...). Now, on one of the update calls, that update call gets sent with DNT:1 because the user has changed their settings. The widget is already loaded though, it's not pulling updated application code, just data. Any requests this widget makes now would get sent with DNT:1 but the widget was loaded with DNT:0 and may have made perfectly valid assumptions based on that. I would say it would be an unrealistic expectation that DNT:0->1 would kick in until the next navigation (that destroys whatever widgets are currently loaded), as opposed to the next request made by an already loaded widget/piece of content/.... user closes the tab, sure the next time they browse to a page with the widget it will get sent dnt:1 and have to deal with that, but to expect that already loaded content is going to change in response to the user changing their browser settings is a very high bar.

-Ian

On Fri, Oct 12, 2012 at 1:55 PM, Shane Wiley <wileys@yahoo-inc.com<mailto:wileys@yahoo-inc.com>> wrote:
Walter,

As the DNT signal is sent with each header a user can change their mind between pages (intra-session).  This issue is focused on a single page request header carrying 2 or more conflicting DNT signals.

Sent from Shane's mobile

On Oct 12, 2012, at 1:13 PM, "Walter van Holst" <walter.van.holst@xs4all.nl<mailto:walter.van.holst@xs4all.nl>> wrote:

> On 10/12/12 10:03 PM, Adrian Bateman wrote:
>> This subject has nothing to do with Internet Explorer. The question is
>> what happens if a broken user agent sends multiple DNT headers, which
>> violates the definition of the DNT header (which MUST only appear once).
>> Options include a) assume DNT:1; b) assume DNT:0; c) ignore all the
>> headers; d) if the multiple headers all have the same value use that
>> value otherwise one of the previous options; etc.
>
> Do I understand you correctly that this would mean that the user cannot
> change his or her mind mid-session?
>
> Regards,
>
> Walter
>
>


Received on Friday, 12 October 2012 21:15:16 UTC

This archive was generated by hypermail 2.3.1 : Friday, 21 June 2013 10:11:36 UTC