Re: [PATCH] send HTTP/1.1 requests to HTTP servers for selected requests

On Tue, 02 Mar 2010 01:07:16 +0100, Henrik Nordström
<henrik_at_henriknordstrom.net> wrote:
> tis 2010-03-02 klockan 12:34 +1300 skrev Amos Jeffries:
>
>> The only risk I can really see is with compliant servers. If they see
us
>> as 1.1 relaying an Expect: there is a reasonably high risk they might
>> actually try to use it. As they should.
>
> Which won't hurt compared to today assuming we are swallowing the 100
> responses, which we already do due to broken servers sending them to us
> even in HTTP/1.0..
>
> It's very very rare servers will respond with 417 in response to Expect:
> 100-continue. And if they do then the applications talking to those
> servers can be assumed to not blame Squid when the application fails to
> understand 417...
>
> But yes, once we announce ourselves as HTTP/1.1 to servers then we will
> see a whole lot more 100 Continue responses which need to be swallowed.
> But that's nothing strange.
>
> Regards
> Henrik

Okay we seem to be in rough agreement.

I'll check the 100-continue swallowing is in 3.x properly and port the
ignore_expect_100 + 417 stuff from 2.7 later this evening.

For now the consensus seems to be that we don't have enough evidence of
brokenness when sending 1.1 ourselves to make down-grading version
worthwhile yet. That can be added as needed after the fact, like most
hacks.

Alex; If you want to go ahead and make patches to advertise 1.1 on server
side and client side separately please just say yes and do it. Otherwise it
will stay on my long-term TODO list for later in 3.2.

Aim for committing to trunk in about a week. Seeing how much we can get
tested and into 3.1 as well. Final RC for 3.1 is expected in 2 weeks.

Amos
Received on Tue Mar 02 2010 - 00:41:50 MST