Re: ETags vs Variants, was: Revising RFC2616 - what's happening

sön 2006-11-05 klockan 17:13 -0500 skrev Robert Sayre:

> You're deprecating a feature people actually use, so the result is
> much worse than the current reality. I understand why the header is
> difficult for Henrik, because Squid is designed to be placed between
> parties it knows nothing about. Caches that have more out-of-band
> information, such as AOL's or Django's, have an easier time with it.

I don't have a problem with understanding Vary. Apart from the small
detail what change in the Vary seen for an URI is actually supposed to
mean, which I choose to read as if the rules for that URI has changed.
If server authors reads it differently there will be a performance
penalty at worst due to stuff which in theory could have been served
from our cache no longer is and the request gets forwarded to the origin
server, and a bit of trashing in our store.

I'd like to see the specs clarified on what a new Vary value for an URI
means, but I can live with the current specifications just as server
authors have to live with what we implement if they want to benefit from
it.

Regards
Henrik