** Keith Moore wrote: **
> I just looked at the proposed agenda for the webrepl BOF.
>
> I conclude that the BOF agenda needs some rethinking.
The BOF agenda is being rethought.
The draft agenda was based on the items with known Internet-Drafts,
or commitment from protocol developers. My lack of knowledge in the
replication area (I could not find any draft, and no protocol developers
contacted me) was the only thing keeping replication out.
> b) we're not likely to approve a working group to address caching
> without understanding how replication will fit in to the architecture
I am, on principle, rather disappointed by this. You are saying that no
working group on caching will be approved unless there is an understanding of
replication, even if there are concrete short term protocols emerging from the
group. I am not convinced that we will have an understanding of replication
within the next 6 months, or even within the next year. Understanding should
here be translated as rough consensus and running code.
We do need protocols for client-cache communication
and for networkelement-cache communication real soon.
There is push and committement for some of these protocols, and
competent people with running code willing to work within the IETF.
Some of the problems we have had so far in this group have been linked to,
in my opinion, lack of focus and a fuzzy feeling that "everything in the
general field of non-server and non-client webrelated issues" should be
part of the charter. I do not object to warm and fuzzy feelings, but I'd like
them to stay out of the charter. I do not mind taking on the world, but I'd
prefer to take on one part at a time (this may be due to engineering training).
I would rather we produced protocols for
- client-proxy configuration
- network element/proxy interaction
than reach understanding on replication. If we can do both, fine. If we try
to do both and get bogged down in philosophical discussions on how the world
_should_ be, not fine.
Bottom line on this:
If there are _concrete proposals_ on replication,
I am happy to include them as part of the charter
> c) since there's no time on the agenda to discuss replication,
I have a suspicion that I am confused on terminology (and
that some of the first things we need to do is to agree on
terminology for this area). Replication is "non-origin server copy"
as opposed to caching which is "non-origin server copy" with a client
pull flavor. And I am still confused on "intermediaries".
Ingrid
This archive was generated by hypermail 2b29 : Thu Nov 18 2004 - 11:21:25 MST