RE: [squid-users] Squid V2.5 and RedHat

On Sat, 2003-01-25 at 10:13, Robert Adkins wrote:
> Actually....
>
> They would claim this as a "Feature" as it does speed up the initial
> connection to Microsoft Internet Information Servers. They use some kind
> of non-standard (Really? MS being "non-standard" Who would have thought
> that?) way of initializing communication between IE and IIS. It skips the
> initial handshake and when the first access attempt is made by IE, IIS
> starts spitting out the page, as the ACK. (Or something like that... The
> article I read went into more detail then I am currently able to follow.)
>
> Anyway, when talking to a Squid proxy server, and everyone else's web
> servers, this is why IE is slow to receive pages. (Since it then has to
> start acting like a standard browser when it's initial BS communication
> attempt fails.) With Squid, you get that "Page cannot be displayed
> error."

Actually this is wrong. The fault affects the use of IE 6SP1 with IIS as
well, with any of basic, NTLM or kerberos authentication. The exact
behaviour depends on the authentication type used.

As for the article, if its the one I saw linked from slashdot, the
author:
*) Didn't provide tcpdump traces or equivalent
*) Was writing from memory of events a while back (according to the
article anyway).

I'd love to see traces showing the claimed behaviour, to be able to
effectively review it.

Rob

-- 
GPG key available at: <http://users.bigpond.net.au/robertc/keys.txt>.