Re: Variant-ID proposal

Roy T. Fielding:
>
>Argh, no, you don't need any of this stuff.  

Yes I do.

>  The Alternates mechanism
>must be completely orthogonal to Vary.

I agree 100%.  That is why I think we need this stuff.  If we don't
have it, it will be difficult, if not impossible, to make transparent
negotiation (Alternates) orthogonal to opaque negotiation (Vary).

If you have two orthogonal mechanisms, you can't have them share the
same space in the Cval: header.  This is a general principle of good
design, and we had better stick by it.  That is why we need to make it

       cval-info = opaque-validator [ ";" variant-id ]

       variant-id = vary-part 
                  | alternates-part                  
                  | ( vary-part "/" alternates-part )

       vary-part = token
       alternates-part= quoted-string 


and no simpler.  

Roy, It is clear to me that, though we have the same requirements,
your mental model of how Alternates and Vary work orthogonally is
completely different from mine.  Your mental model may allow sharing
of space in the Cval header, but this sharing leads to horrendous
problems in mine.  I don't think we have enough time to get in sync on
mental models before the end of this month.

I'm asking you to trust my bad feelings.  We are walking on very thin
ice: we are trying to be compatible with a transparent negotiation
mechanism that is likely to change before it gets consensus.

I have proposed a minor syntactic extension to a header, an extension
that even _reduces_ the code needed to implement cache replacement.

I can't provide guarantees that we will absolutely need this extension
when we get to define opaque content negotiation on top of 1.1 in May.
If it turns out we don't need it, we can leave it out of 1.2, and
nothing much is lost.  All I can say is that I suspect that we _will_
indeed need it, and that we have very much to loose if we leave it out
while it had to be in.  As I indicated in my previous message, my
proposal above moves us on thinner ice already, and I have no
intention to make even bolder moves.

Again, I'm asking you to trust my bad feelings.

>........Roy

Koen.

Received on Monday, 15 April 1996 20:55:47 UTC