Re: WGLC p1: MUST fix Content-Length?

[Willy Tarreau <[e-mail]>]

On Tue, Apr 30, 2013 at 11:53:28PM -0600, Alex Rousskov wrote:
> Hello,
> 
>     When talking about a Content-Length header field with multiple
> identical values, Part 1 Section 3.3.2 of HTTPbis says:
> 
> > the recipient MUST either reject the message as invalid or
> > replace the duplicated field-values with a single valid
> > Content-Length field containing that decimal value prior to
> > determining the message body length.
> 
> It is not clear whether "recipient MUST replace" (a requirement on the
> recipient) also implies that "a sender MUST replace [...] when
> forwarding the message" (a requirement on the sender). This issue has
> been raised on 2011/11/28, but the discussion diverged, and I could not
> tell whether there was a consensus on what the correct interpretation is.
> 
> Please decide whether a proxy MUST "fix" such Content-Length headers
> when forwarding the message and adjust the above text to clarify one way
> or another.

That's what the discussion converged to. I even modified haproxy in order
to do so. The idea is simple : if you receive a message with multiple
content lengths, either you can't deal with them and must reject the
message, or you can deal with them and then you know how to fix the
message before interpreting it or forwarding it, so you must do so.

Do you think the text needs to be adjusted ?

It can be seen as a constraint to fix a message before forwarding it,
but in fact it's a relaxation of a case where you were supposed to
reject the message.

Willy