On Mon, 16 Apr 2007, Philip Hazel wrote:
> On Fri, 13 Apr 2007, Tony Finch wrote:
>
> > This is incorrect. RFC 2920 says:
> >
> > The actual transfer of message content is explicitly allowed to be
> > the first "command" in a group. That is, a RSET/MAIL FROM sequence
> > used to initiate a new message transaction can be placed in the same
> > group as the final transfer of the headers and body of the previous
> > message.
>
> Oh rats! I'd overlooked that. OK, I'll have to back off some of it, but
> I can still catch the case when the client has not sent anything after
> the final dot, and the connection has then gone away before the server
> can respond. That will catch many (most?) of the common cases.
I have committed a revised patch that catches the case when the client
does not send anything after the final '.', but closes the socket before
Exim gets round to sending its reply.
Philip
--
Philip Hazel, University of Cambridge Computing Service.
