On Fri, 2 May 2008 10:46:30 +0100 Nigel Metheringham
<nigel.metheringham@???> wrote:
> As I understand it, addresses with trailing dots and other odd
> characters in the local part are not RFC valid - are they
> realistically used across the network?
>
> On that assumption, should the general local part validation of
> addresses accept those forms, be modified to reject them (which
> obviously *can* be done in ACLs), or modified to reject them with an
> override (like the helo characters one)?
>
> Modifications would be an incompatible change - working configurations
> could then be stopped working, which might be a good thing (unless you
> are the one caught out by it).
>
> Nigel.
I would support an option to opt in to strict checking. The don't support
doing it in ACLs because that adds to the complication of configurations and
would be far less efficient.
(This takes us back to the VERY big mistake of the early Internet maxim of
being strict in what you send but liberal in what you accept. If the maxim
was to be strict in what you accept then we would not find ourselves with a
legacy of systems generating invalid messages. Those systems would have failed
from the start and would have been fixed before going into service, so Yahoo
Groups would not be allowing these addresses now.)
Phil.
---------------------------------------
Phil Chambers (postmaster@???)
University of Exeter
