Re: Why Microsoft's authoritative=true won't work and is a bad idea

Dave Singer wrote:
> 
> At 17:46  +0200 7/07/08, Julian Reschke wrote:
>>> An alternative might be to add a header "look, I guessed it" when 
>>> Apache adds a 'guessed' text/plain as the content-type, and say that 
>>> browsers might take text/plain+IGuessed as something to sniff. This 
>>> is like the proposed Microsoft header but the other way up...
>>
>> I agree that's nicer in theory. But how do we get all the existing 
>> installations to change their DefaultType?
> 
> Well, someone has already said that the IE upgrade rate is less than 
> 50%.  There are a lot more clients (browsers) than servers out there, 
> and system admins tend to be better at upgrading than users (because 
> they get security fixes along with bug fixes).  So fixing the server end 
> of the equation with an extra "look, I guessed, I might be wrong" might 
> get us closer to a better position faster than fixing the browser end.  

But if we can get server admins to change their default to "text/plain; 
butIguessed", we probably can also get them not to send a default, 
right? (*)

> The "I guessed" message has, as a famous politician once said, the 
> advantage of being true, which means there is no need to work around it 
> etc.

It would be something like "authorative=no", and yes, it would be the 
superior solution if we could get it working.

> Browsers could then detect that they (a) are talking to a server with 
> this feature and (b) they get text/plain *without* "I guessed" and 
> therefore (c) not sniff, but believe.

BR, Julian

(*) Assuming for a moment that the httpd fix is getting backported...

Received on Monday, 7 July 2008 16:28:24 UTC