>2) Squid recognizes a broken request (with a header that it can't
>handle)...
1- Do you reference "broken request" because you want to say that there is
written both "expect:100 continue" header and http/1.0 protocol ?
If it was so, Squid should break connection if I don't use
"ignore_expect_100 on" option
> 2) Squid recognizes a broken request (with a header that it can't handle)
> and terminates the request with a visible error explaining exactly what is
> going on
if Squid ignored broken request (with a header that it can't handle +
http/1.0 specified ) HTTP Server always would give 417 error ?
So Squid only would predict behavoiur of server... What do you think ?
----- Original Message -----
From: "Chris Robertson" <crobertson_at_gci.net>
To: <squid-users_at_squid-cache.org>
Sent: Friday, February 19, 2010 9:04 PM
Subject: Re: [squid-users] R: [squid-users] HTTP/1.0 417 Expectation failed
> Riccardo Castellani wrote:
>> I ask if Squid previous versions ignored Expect header forwarding request
>> to
>> server.
>> But now why Squid code (2.7 version) detects this header and it handles
>> it
>> as error ? Why has this Squid behaviour been modified ?
>
> Given the following scenario...
>
> Some random software, upon receiving notification that there is a HTTP/1.0
> link in the chain decides to ignore that fact and use a HTTP/1.1 method
> (against protocol)
>
> ...which outcome would you prefer:
>
> 1) Squid accepts the traffic, unrecognized request header (Expect:
> 100-Continue) and all, and silently fails when the response comes in, or
> returns a partial response as if it was complete.
>
> or
>
> 2) Squid recognizes a broken request (with a header that it can't handle)
> and terminates the request with a visible error explaining exactly what is
> going on.
>
> Personally, I like to know why things break when they inevitably do.
>
> Chris
>
Received on Sat Feb 20 2010 - 10:08:32 MST
This archive was generated by hypermail 2.2.0 : Sun Feb 21 2010 - 12:00:05 MST