Re: Today's call: summary on user agent compliance

On Jun 11, 2012, at 2:07 PM, Rigo Wenning wrote:

> Roy, 
> 
> On Saturday 09 June 2012 16:06:35 Roy T. Fielding wrote:
>>> Exactly what information arrives
>>> so the Web server understands itıs broken?
>> 
>> Somebody tests the browser and says it is broken. 
> 
> This is stupid browser sniffing. And it doesn't buy you a thing.

It doesn't need to buy anything.

> Because if I have altered my preferences, even IE sends compliant 
> signals as they represent my choice. This means you just have 
> ignored a valid DNT:1 request because you second-guessed the wrong 
> thing.

No, it means I have ignored a header field because it came in
with another header field that matches a non-compliant UA.
Since I have stated that I will not honor DNT when set by
that UA, I have done exactly what I said I would do.  If you
have chosen to spoof the User-Agent header field for some other
UA, then I take that as an instruction that you want all of
the same behavior that I would have delivered for that UA,
including ignoring the DNT signal.

> Sane protocol design looks different IMHO

Sane protocol design doesn't start with an indication of user
preference that did not involve the user indicating a preference.

....Roy

Received on Monday, 11 June 2012 22:14:58 UTC