Re: Meaning of Vary header applied to non-existent headers

On Mon, 23 Feb 2004, Jamie Lokier wrote:

> 1. When an origin server selects the response by looking at a request
>    header, and the header does not exist, is it correct for the server to
>    add the name of that header to Vary?

If cachable response depends on header X, the server must add a
	Vary: X
	Vary: *
header. This behavior should be independent from actual headers
present in the request (IMO).

> 2. How should a cache treat missing requests headers which are
>    mentioned in Vary, when deciding if a stored entity matches a
>    request?

As you know, RFC 2616 defines Vary matching based on the existence of
a LWS/merge transformation. Such transformation cannot add new or
remove existing headers. Thus, IMO, a present header can never match
an absent header. An absent header matches an absent header.

> ...
> I would imagine that in case a, the entity matches, and in cases b
> and c, the entity doesn't match and must be revalidated.

I agree.

>    The selecting request-headers from two requests are defined to match
>    if and only if the selecting request-headers in the first request can
>    be transformed to the selecting request-headers in the second request
>    by adding or removing linear white space (LWS) at places where this
>    is allowed by the corresponding BNF, and/or combining multiple
>    message-header fields with the same field name following the rules
>    about message headers in section 4.2.
> Is this supposed to mean that an absent header is equivalent to an
> empty header, or that an absent header is semantically a distinct
> value from an empty header, or that an absent header is not considered
> at all in the decision, or that an absent header cannot be matched
> even with an absent header in the original request for the cached
> entity?

If both requests lack a varied header X, then a null transformation
(i.e., transformation that does nothing) matches them.

To match a header without a field-value with an absent header one
would need to add a header name to one of the requests, which is
probably not a part of the allowed transformation. However, a header
without field-value and/or a header without field-content (RFC 2616
BNF is ambiguous in this context) are strange animals, and I would not
recommend counting on any specific implementation of related

> 3. Does this mean that server code which inspects a header value, and
>    finds the header missing, should look like this instead?
>        IF exists request_headers (Accept-Language) THEN
>            reponse_headers (Vary) = "Accept-Language"
>        ELSE
>            response_headers (Vary) = "*"
>        ENDIF
>        IF request_headers (Accept-Language) contains "ja" THEN
> 	   serve the japanese page
>        ELSE
>            serve the english page
>        ENDIF

I would add Vary header unconditionally. Whether you want to disable
caching depending on request properties is a question that goes beyond
protocol semantics. The right thing to do would depend on the
trade-off between more hits and fewer incorrect objects delivered.

Note that a valid Accept-Language cannot be empty. It is not clear,
however, how an invalid header is supposed to be treated by the
matching algorithm in the RFC.


Received on Monday, 23 February 2004 16:34:05 UTC