lynx-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: lynx-dev '\' reloading documents


From: David Woolley
Subject: Re: lynx-dev '\' reloading documents
Date: Thu, 29 May 2003 21:30:55 +0100 (BST)

> Yes, and irritating it is too.  I wish I could get more browsers to
> ignore the gratuitous no-cache pragmas that some sites send.

They are typically invalid and largely ineffective as well.  As far
as I know, HTTP/1.1 only define Pragma: no-cache in the client to
server direction.   At a HTTP/1.0 compatibility level, I think Expires
is what they should use.  I suspect that big 2 browsers ignore the 
standard in this respect, although it is also possible that Pragma:
no-cache from server to client is totally apocryphal.

Moreover, shared caches totally ignore the payload of the HTTP response,
so don't see them (they are probably more likely to not cache because of
Cookies in the request, or the lack of a correlator (Last-Modified, or
ETag) in the response).

I think the basic reason that starting web designers always ask how
to frustrate caching is that it avoids the need to think, including
learning more than one technology (HTTP as well as HTML) and choosing
exactly which cache control directives are needed to maintain function,
but maximise performance.  Thinking, and employing people that can
think costs businesses money.

A secondary reason is probably the desire to get cookies back, even on
static pages, for market research purposes.   (I can't remember whether
HTTP/1.1 caches are supposed to infer a Vary: Cookie header, when there is
a cookie in the request, or actually require it to be present, to be
effective.  Either way, that's a two technology situation in a one
technology world. (If you just want to track with HTTP/1.1, you should
use cache-control: must-revalidate, so that only If-Modified requests go
end to end.))



; To UNSUBSCRIBE: Send "unsubscribe lynx-dev" to address@hidden

reply via email to

[Prev in Thread] Current Thread [Next in Thread]