In a previous episode Henrik Nordstrom said...
::
:: Patrick McManus wrote:
::
:: > 'equivalent requests' is a problem.. consider Accept headers and q
:: > values and their interactions with proxies.. a proxy may very well
:: > choose to give the same response back to two subsequent requests that
:: > would have generated different variants if presented directly to the
:: > origin server.
::
:: Not entirely true. HTTP/1.1 servers SHOULD use Vary: to indicate when
:: the response varies based on header values or other things derived from
:: the request, and caching proxies MUST to pay attention to Vary headers.
::
:: See RFC2616, sections 12.1 and 13.6.
::
that's true of course, but it isn't really germane to the phrase..
a] SHOULD is not MUST, and all responses are not 1.1.. even for
responses that are 1.1 a server that values cacheability (perhaps a
bandwidth starved sever) may not issue a varies: on accept: as varies:
is really just a medium-grained cache busting technique.
b] indeed judging the equivalency of a request isn't something that
can really be done in isolation from the server and the possible
responses... you'd certainly have to define in what way they are
equivalent (I can think of at least 2 interpretations) and if you mean
"multiple requests that can be satisfied by the same response" then
I'd suggest just saying that..
-P
This archive was generated by hypermail 2b29 : Thu Nov 18 2004 - 11:21:26 MST