Re: Issue-57

Xiaoshu Wang wrote:
> Don't ask people to make inference from *how* a message is received. Ask
> people to make assertions in the message *explicitly*, with more refined
> terms suited to their intended granularity.

It's interesting that this is exactly how I used to feel about HTTP-range-14, 
and to some extent still do.  I.e. that semantics of information should be 
independent of the protocol whereby it is obtained.

But there remains an important consideration.

Imagine an exchange:

   A: give me X
   B: I can't give you X, but here is Y

I think it's fair to say that A shouldn't feel licensed to make inferences about 
X (the thing asked for) from B's response.  After all, I don't suppose you'd 
suggest that A could make inferences about X if B's response was an HTTP 404?

As far as I can tell, that is all that HTTP-range-14 is providing:  a way to 
distinguish between a response that means roughly "here is what you asked for", 
and "here is something not what you asked for".

In this respect, it's hard to argue that the HTTP-range-14 resolution is wrong. 
  But I can appreciate that for many purposes it's not enough.  Which, I think, 
is all that Jonathan is trying to convey.

(At this point, I think I'll return the the cover of the nearest undergrowth...)

#g
--

Received on Friday, 24 June 2011 20:34:43 UTC