- From: Jeffrey Mogul <mogul@pa.dec.com>
- Date: Thu, 02 Jan 97 11:21:10 PST
- To: Koen Holtman <koen@win.tue.nl>
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Regarding the apparent contradiction between 14.8 and 14.9.4 over
the meaning of proxy-revalidate:
>I think we've agreed that this is a bug in the specification. That
^^^^^^^^^^^^
>is, we did not intend that the meaning of this cache-control directive
>should change when another response header is present.
For the record, if you are using the word `bug' not just to mean `we
intended it differently', but also to mean `this is a fatal problem
that must be fixed', I do _not_ agree that this is a bug in the
specification. In my reading, 14.8 does *not* modify or contradict
14.9.4.
14.9.4 talks about restrictions implied by proxy-revalidate.
14.8 talks about restrictions implied by the Authorization header
which happen to be _lifted_ when proxy-revalidate is present.
Well, we *did* intend it differently, as Paul Leach has reminded
me, although the most serious consequence of this is for the
digest-authentication stuff.
It is true that one could read the language in 14.8 as being a
modification of the meaning of 14.9.4, although in that case I would
have to admit that I wrote a confusing and non-orthogonal
specification. for this directive.
But in fact, when I wrote these two paragraphs, I had different
(and incompatible) meanings in my own head for the same directive,
and so that's why I feel confident calling it a bug. It was my
error, so I get to say what kind of error it is! :-) (More to
the point, what I wrote does conflict with what the editorial
group wanted to see.)
-Jeff
Received on Thursday, 2 January 1997 11:32:46 UTC