Basically, an index page is updated whever the archive is modified,
but if a user has cached an earlier copy of the index, that won't
refclect the changes. So, every time a user looks at the index,
should he/she reload it?
When no Expires or Cache-Control header is sent by the server to the
client, the behavior of the client will depend on how it is
configured. For example, IE and Netscape can be configured to not
check if page changed unless a new session is started (i.e. browser is
exited and restarted), never, or every time.
Or is the http protocol clever enough to ensure that if a page is
requested fom a proxy server that returns a cached copy, the proxy
will first ensure that the cached page is up-to-date, and request a
fresh copy if necessary?
I think it depends on how the request proxy server is configured
to deal with data not given an expiration.
I was wondering about using the "expires" META tag to expire pages.
But unfortunately, from what I understand of the specs, this tag only
allows specification of an absilute date, not of an offset, which is
a bit limiting. I would like to be able to specify a 24 hours"
value for "expires", because whilst an absolute day of now+24hours
would work fine whilst the archives are current, I don't like to
think of the ugliness of indexes to old periods being
permanently stamped as "expired".
You may want to see if you can specify a Cache-Control directive in a
META tag and see if it will work. It does assume that the client
supports HTTP/1.1 and that Cache-Control directive is honored if in a
META tag. Check the HTTP/1.1 spec on more information about
Cache-Control. There seems to be no relative time specification
BTW, if you get something that does work, post your solution to
this group. This topic looks worthy enough to include in the FAQ.