Re: Initial feedback on the well-known URI Proposal

On Wednesday 07 March 2012 11:39:55 Roy T. Fielding wrote:
> > The smart thing about the DNT-header approach is that a resource is
> > requested  with a DNT header and the response concerns exactly that
> > resource. This is scoping the semantics nicely and naturally without
> > having to describe everything in advance.
> 
> If that were true, the header field would have no value because
> it only defines what is claimed to have happened in the past.

And you can consent to things that happened in the past. Additionally, it is 
not correct to assert that something happened in the past. The user sends a 
DNT header to the service and the services sends a response header back with 
the content. This way the user knows whether the service has accepted his DNT 
preference. I know that you will say the user can not expect this. And yes, 
this is the disadvantage of having the protocol started by the user instead of 
having it started by the service like in P3P. There you know what the service 
will do and can take it or leave it. DNT is the other way around. The user 
declares something and the service may take it or leave it. If you scope that 
declaration in a very general way, you will hit the wall of the granularity 
needed by internet services. The risk is that you get a general DNT=1 for 
everything IMHO instead of getting a DNT=1 for the things the users don't 
want. IMHO, the missing granularity is to the detriment of competition between 
the services and also to the detriment of the third parties who will be in an 
all or nothing position together will all other third parties. If one evil guy 
is there, everything gets blocked. I don't see this as a future proof 
solution. 

See, in WAP, the industry tried to force people to always start with the 
portal page. What a royal failure. This was not accepted. We can fail here 
too, not to the detriment of the users, but to the detriment of the industry 
IMHO.

Rigo

Received on Wednesday, 7 March 2012 20:56:42 UTC