Re: Is forwarding hop-by-hop headers a MUST-level violation?

On Fri, 23 Jul 2004, Jamie Lokier wrote:

> > > I wouldn't be surprised to find some old products check for Connection
> > > == "close", or !strncmp(connection, "close") if you see what I mean.
> >
> > I saw some _new_ products that do that. That is one reason why I am
> > not pushing for (2).
>
> Ouch.  Do the products you've seen check for (effectively)
> !strcmp (connection, "close") or !strncmp (connection, "close", 5)?
>
> If the former, is transmitting _two_ Connection headers, one of which
> is literally "Connection: close", an adequate workaround for those products?

Implementations and their bugs vary, of course. As you may suspect,
some cannot handle multiple Connection: headers (some will only see
the first "line", some only the last one, and some will see none).

It is probably impossible to change any RFC 2616 functionality without
breaking some buggy implementations. I am not sure whether we should
take those implementations into account. If we should, then we should
not require listing known hop-by-hop headers in the Connection header.

Alex.

Received on Friday, 23 July 2004 18:18:03 UTC