Re: Proxying PHP Pages

From: C. Jon Larsen <jlarsen@dont-contact.us>
Date: Thu, 10 Dec 1998 23:29:40 -0500 (EST)

I've had good luck with mod_perl apps and MSIE only when I set the
pages I don't want cached with an expiration date in the past in the HTTP
headers. Netscape works with the pragma nocache just fine, IE gets all
confused and starts caching everything sometimes, whether a proxy is in
the middle or not. Once I set the expiration date, everything works just
great, so I'd say its an issue of the application developer not testing
their application completely. I bet the problem crops up when a browser
issues an HTTP GET that has the same URL and query params as a previous
query that is cached. Again, expire dates "fix" this.

On Fri, 11 Dec 1998, Henrik Nordstrom wrote:

> Steve Judge wrote:
>
> > A couple of questions:
> > 1) I didn't think that squid cached cookies ?
>
> It doesn't cache cookies. It caches the page if allowed, but not the
> cookie (cookies are uncacheable).
>
> > 2) Dynamic PHP pages, are these the same as cgi's and should we
> > include them under the stoplist part not to cache or fetch from
> > neighbors etc ?
>
> PHP is a alternative to ASP.
> Yes, you could say that PHP kind of like CGI.
> Yes, you should probably include .asp and .php to your
> hierarchy_stoplist.
>
> Dynamic pages are NOT cached by Squid unless you have a refresh pattern
> with a min age other that 0. You SHOULD NOT have a min age other than 0
> on unknown or HTML files as this is how dynamic content usually is
> identified.
>
> If I am not mistaken some versions of IE gets totally confused when
> using a proxy and caches everything locally vithout revalidation (not
> even reload works). This could also be the cause here.
>
> What is said in the Squid access log? *MISS, *HIT or nothing?
>
> > I've also emailed the company concerned suggesting that their
> > product should be more proxy aware as they are a fact of life on
> > the internet now instead of adding isp's to a ban list that are
> > running " aggressive cacheing practices " as they put it.
>
> If their pages are unshareable then they should have a
> Cache-control: private
> header
>
> If their pages are one shot only (not even the end user should be able
> to redisplay the same page without refetching it) then
> Cache-control: no-cache
> and/or
> Pragma: no-cache
> should be used.
>
> Pragma: no-cache is defined in HTTP/1.0 and works in all caches and
> browsers. It is replaced by the more flexible Cache-Control directive in
> HTTP/1.1.
>
> ---
> Henrik Nordstrom
> Spare time Squid hacker
>
>

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
C. Jon Larsen Email: jlarsen@ford.ajtech.com
Systems Engineer Voice: +1.804.353.2800 x118
                          Cell: 357.3040
                          Pager: 219.3406
A&J Technologies http://www.ajtech.com

PGP Key fingerprint: 8A 62 4C 6E 1E 3C CD 63 B3 16 1A 1B D2 61 EE 97
PGP Public key available at: http://intranet.ajtech.com/~jlarsen/CJL.txt
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Received on Thu Dec 10 1998 - 21:27:05 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:43:36 MST