Re: introduction to the TPE

Roy T. Fielding schreef op 2014-04-10 19:32:
> On Apr 10, 2014, at 9:07 AM, Dobbs, Brooks wrote:
> 
>> I know Roy has already shot this down, but I'd say this intro (along
>> with the current one) still has a deficiency.
>> 
>> As I've said before, the intro should set up what the spec does
>> without being misleading. The TPE is not a complicated spec in terms
>> of capabilities on the UA side; you can alter the status quo either
>> by sending a 0 or a 1. When we speak about expressing a preference
>> your options are pretty limited 0, 1 or choose not to decide. The
>> spec, as it sits, only requires the ability to communicate 1. It
>> would have been a trivial change to the language to mandate that UAs
>> MUST offer a DNT:0 option, but the consensus was not to mandate
>> this. Fine, that was what was decided, but having so decided you
>> need to reflect this decision in the introduction. This is not about
>> obtaining "a preference" it is about obtaining "the preference"
>> DNT:1.
> 
> Not mandating the option be present is not the same as forbidding
> that it be implemented. The protocol supports sending DNT:0 by
> configured choice -- it is a MAY on implementation by UAs.
> IIRC, Firefox already implemented it that way in its config,
> but I haven't checked if it actually sends DNT:0 on the wire.
> 
> ....Roy

Ff sends DNT:0 on the wire, which makes this preference a key-building 
block for a consent mechanism.

Rob

Received on Thursday, 10 April 2014 18:52:11 UTC