Re: Choosing proxy on location

On Fri, Nov 12, 1999, Patrick McManus wrote:

[snip]

> :: in the IE5 i just installed (version says "5.00.2314.1003"), MSIE issues
> :: appropriate If-Modified-Since headers for both 'reload' and 'shift-reload'
> :: when no proxy is defined.
>
> too bad the problem is the lack of 'cache-control: no-cache' and/or
> 'Pragma: no-cache' which force an end to end reload even in the case
> of non-stale cache data.. that's what msie doesn't send when connected to
> an origin server.. I don't believe that situation is different in msie
> 5, though I can't confirm at the moment.. but it is definitely the
> issue: not IMS.
>
> the rationale of why when traffic is hijacked that was destined to
> another address that you expect the application level protocol to
> optimize for the hijacker instead of the address the packets are labeled
> for has always escaped me.

You can't be guaranteed that the origin server you're talking to *IS*
an origin server.

Take, for example, a cluster of http accelerators in front of a web
server. I know a lot of people who do this, and wondered why people
were seeing 'stale' data from the accelerators even with shift-reload.

So, yes, initially it doesn't make sense, but there's always more
to it than how it looks superficially.

Adrian

-- 
Adrian Chadd					Systems Engineer
<adrian@ip.versatel.net>			Versatel Telecom BV
						Amsterdam, The Netherlands
	"Music in the soul can be heard by the universe" - Lao Tsu