Re: Received header not parsed

[Justin Mason <jm...@jmason.org>]

Ben Wylie writes:
> >> Received: (from localhost [24.180.47.240])
> >>  by server. (NAVGW 2.5.2.12) with SMTP id M2006060503484615455
> >>  for <us...@domain.co.uk>; Mon, 05 Jun 2006 03:48:47 +0100
> >
> > OK, we specifically skip received headers that start with "(" at line 
> > 387 of Received.pm.  Annoyingly, we don't provide debug info for when 
> > that happens.
> 
> Is there a logical reason for this? If the idea is that SA doesn't pass the
> ALL_TRUSTED rule if there are malformed Received headers, why would you make
> a special exception for badly formed Received headers which begin with an
> open bracket?

There are two subtypes of Received header:

- a server-to-server handover, where the mail crosses a network somehow

- an internal-to-server handover, where the mail travels inside a single
  machine (eg. from a local-delivery subsystem to the SMTP subsystem)

There *was* a pretty solid convention -- "real" server to server handovers
start with "from", and internal messages use the RFC-822 header comment
format, ie. "(enclosed in brackets)".  The latter are irrelevant for
trust path determination, since they happen entirely within one machine.

NAVGW is being really quite broken here, by using that comment format
for "real" transmission over a network.

However I guess we really need to support this, since it sounds like
trust path resolution is screwed unless we do so for NAVGW users :(

--j.

> > What broken-ass software is producing these received headers?  "by 
> > server." certainly isn't too descriptive.  I suppose I could write you a 
> > patch to fix this, but I'm not sure that it'd be included in the 
> > standard code base.
> 
> NAVGW stands for Norton Antivirus for Gateways.
> 
> If you had the time to write a patch for this I would be very grateful, and
> let me know what I would need to do to apply the patch, and if I would need
> to make changes every time I upgraded.
> 
> Thanks
> Ben