RE: "Learning" about pages

On Mon, 27 Jul 1998, Williams Jon wrote:

> Maybe squid could compute decent checksums (MD5?) and store those with the
> documents in the cache. When squid goes to look at the page again, if the
> checksum changed, say, five times in a row, only then would it record that
> the document was uncacheable in the in-memory table.

Sometimes a page might change "for aesthetic reasons" (eg., different "witty
quote" at the bottom) but this is no reason to mark the page as uncacheable
and thus force it to be reloaded for every request.

> > From: Jordan Mendelson [SMTP:jordy@wserv.com]
[...]
> > page. This happens all the time, the best way to fix it is to have the
> > adminstrator add expires tags and what not, but let's face it... not
> > everyone is going to do that.

We need to encourage site admins that it is in *their* best interests to use
and maintain (correctly!) these headers; admins that don't use them, or abuse
them just to avoid being cached (to increase their hit rate) are forcing
unnecessary cost back onto the originating site (remember, not everyone in the
world has flat-rate Internet access - some of us pay to receive traffic so
proxy caching is a very serious business).

> > One of those web pages is yahoo, it seems to show a different web page if
> > you are using a text browser versus a graphical browser. I could go in and
> > disable caching, however in time I'm going to have more hosts than I know
> > what to do with.

Yahoo! should be shot:

  (http://www.yahoo.com/)
  HTTP/1.0 200 OK
  Content-Length: 10518
  Last-Modified: Tue, 28 Jul 1998 05:22:16 GMT
  Content-Type: text/html

  (http://www.yahoo.com.au/)
  HTTP/1.0 200 OK
  Last-Modified: Tue, 21 Jul 1998 08:07:44 EAST
  Content-Type: text/html
  Content-Length: 11146

Ouch, for oh so many reasons. :-(

dave
Received on Tue Jul 28 1998 - 05:12:37 MDT