Re: What is a specification for? [was Re: Calculating Age Question]

>Thus, I'd like to ask that if ANYONE ELSE in the WHOLE WORKING GROUP
>agrees with Roy that the proposal for calculation of Age in the
>HTTP/1.1 specification is incorrect, would you please at least send me
>(if not the whole group) a message indicating that you think this is
>something that should be addressed.

FWIW, I think the age calculation is conservative to a fault and would
prefer to see it change, but I don't share Roy's passion.

I consider the issue of whether or not to add an Age: header (with any
value) to a response that was NOT served from cache to be a different issue
entirely, and feel more strongly that a proxy should not add an Age: to a
fresh response.

If a client makes a must-revalidate request to a proxy and sees the Age:
header (worse - one miscaluclated with a time larger than the initial
resposne time minus request time) how can it distinguish between whether 1)
it got a fresh response, or 2) that the proxy ignored it's wishes and gave
it a stale copy from a few seconds prior?

Now that I reference the spec again, I'm confused in the ascertation made on
this list that all proxies must add an Age: header.

In 14.6:  "HTTP/1.1 caches MUST send an Age header in every response."

That's caches -- not proxies.... ?

Daniel DuBois, Traveling Coderman
   o  The alpha of Heroes of Might and Magic II Bible is here!
   o  Free Trial: Now you can play Meridian 59 FREE for two days.
      Get all the details at

Received on Thursday, 28 November 1996 13:15:50 UTC