Re: First draft of negotiation requirements document

Ted Hardie:
>
>On Jun 17, 11:04pm, Koen Holtman wrote:
>> Section 5.2.1 end of first paragraph:
>>
>> | [TCN] describes a standard method for delineating
>> |the axes along which a resource varies and a set of methods by which
>> |caches can participate in the negotiation process.
>>
>> I assume that you mean that remote variant selection algorithms are in
>> this set of methods.  In that case, it would be better to write
>>
>>  `a set of methods by which origin servers and proxy caches can
>>  optimize the negotiation process.'
>
>I think we still need to test whether or not they do optimize the
>negotiation process. I think it is very clear that they allow the proxy
>caches to participate in the negotiation, but that we will need actual
>data on hit rates using proxy-negotiated selections before we can
>say that it works.

Actually, I don't care that much about proxies running remote variant
selection algorithms.  I think that it will be far more significant in
the short term that TCN allows origin servers to run remote variant
selection algorithms.

You start out by saying that 

    Elective negotiation relies on the idea that the first response to
    a request for a resource which has variants should be a statement
    naming the choices and/or the axes along which they vary.'

and a remote variant selection algorithm would optimize this.  Would
it be inoffensive enough if it would say that TCN has

  `....methods by which origin servers (or proxies) could
   optimise the basic elective negotiation scheme'?

>  I agree that the current language is a bit weak, but
>I think it would be more useful to find language which concisely describes
>*how* they participate.  Any help you can provide there would be
>much appreciated.

I don't think that we could manage to describe all forms of
participation in half a sentence.  There is not just remote variant
selection algorithms, TCN also allows for a number of caching
optimisations which go beyond those in HTTP/1.1, and some of these may
even, in the short term, have a greater roundtrip-saving effect than
remote algorithms.  Maybe it is better to just include a reference to
section 4.4 of the TCN draft.

Koen.

Received on Wednesday, 18 June 1997 12:03:38 UTC