Re: [Zope] RE: zope/style sheets
On Wed, 02 Feb 2000 11:33:36 -0800 Sam Gendler <sgendler@impossible.com> wrote:
I would like to warn the community about serving multiple pages up based on the exact same request url (which is what the code below can contribute to).
This doesn't make sense to me on account of cookies. Many sites out there have the content of their pages dynamic on the presence of a specific cookie, without varying the base URL at all. -- J C Lawrence Home: claw@kanga.nu ----------(*) Other: coder@kanga.nu --=| A man is as sane as he is dangerous to his environment |=--
J C Lawrence wrote:
On Wed, 02 Feb 2000 11:33:36 -0800 Sam Gendler <sgendler@impossible.com> wrote:
I would like to warn the community about serving multiple pages up based on the exact same request url (which is what the code below can contribute to).
This doesn't make sense to me on account of cookies. Many sites out there have the content of their pages dynamic on the presence of a specific cookie, without varying the base URL at all.
If you read the spec for state management in HTTP using cookies, it specifically states that it is the responsibility of the content author to ensure that content that is served based on the value of a cookie is marked with cacheability headers correctly. Responses that include a Set-Cookie header are inherently uncacheable, but all objects that have a cookie in the request are to be considered cacheable unless told otherwise. Obviously, you can set a Varies: Cookie response header, but that becomes something of a pain in the ass, especially since most caches ignore it. This is very important, since all graphics and other static content will also have a cookie sent in the request, even though they are totally static. There were enough sites that abused this concept, that most caches have developed the same startegy for dealing with cookies. Most caches will cache all binary objects that are received for requests with a cookie (all content-types that don't start with 'text/'). All objects that are text are automatically not cached if there is a cookie in the request. Most of the cache vendors did this in response to my.yahoo, which had a tendency to serve the same my.yahoo page to all users once a singl person had logged in that day (it got cached). To the best of my knowledge, they still do this, despite numerous complaints. They are explicitly ignoring the cookie spec, and costing the rest of the world a lot of bandwidth in objects that could have beenm cached, but which are not. By the way, I have only ever seen one site that served different _binary_ objects based on a cookie (the graphics were different on a page). It was a Phish fan club site. I know others exist, but they are dooing a bad thing and should shape up their act! </rant> --sam
-- J C Lawrence Home: claw@kanga.nu ----------(*) Other: coder@kanga.nu --=| A man is as sane as he is dangerous to his environment |=--
-- Sam Gendler Chief Technology Officer - Impossible, Inc. 1222 State St. Suite 250 Santa Barbara CA. 93101 w: 805-560-0508 f: 805-560-0608 c: 805-689-1191 e: sgendler@impossible.com
participants (2)
-
J C Lawrence -
Sam Gendler