Re: empty quoted strings and other oddities

[[e-mail] (Kai Henningsen)]

[e-mail] (Keith Moore)  wrote on 02.10.02 in <[e-mail]>:

> > > so? there are lots of addresses which are valid syntax but
> > > not valid because the domain is not defined or the IP
> > > address (in the domain literal) does not exist or is not
> > > assigned to a host.
> >
> > I'm not sure I buy that reasoning.  You seem to be saying that since it's
> > not possible to detect all invalid addresses syntactically that the
> > ability to trap some syntactically (or more than currently trapped) is
> > irrelevant.
>
> no, I'm saying that
>
> - there is useful separation of function here between detecting syntax
>   errors and detecting addresses that are invalid for other reasons that
>   should not be discarded

Aren't you both missing a point here?

These particular bad domain literals certainly *should* be detected  
syntactically - not by the SMTP checker, but by the convert-to-IP-address  
routine. (Or by the convert-to-other-transport-address routine, if  
applicable. And then it might even be legal.)

That is exactly the same case as for the "" localpart - only those parties  
which actually need to understand it should decide if it is syntactically  
valid, the rest of the world can just as well handle this as an opaque  
token.

MfG Kai