Getting silly was Re: using an ip in an MX record

[Simon Waters]

Leon wrote:
> 
> For what its worth, Jims argument that if it doesn't appear in the
> Cricket book it is not correct doesn't wash. It seems reasonable to
> expect that the Cricket book may not have covered all circumstances (
> and it didn't expressly forbid dotted quads)

P533 copies the information from RFC1035 
>>>
owner ttl class MX preference exchange-dname

EXCHANGE A domain-name which specifies a host willing to act as
a mail exchange for the owner name.
<<<

So it is perfectly well covered, standards tend only to define
what you can do, not the myriad of ways you could get it wrong!

Of course a dotted quad IP address could be a "valid" domain
name !?!!?! (although lobbying ICANN to delegate you 256 new top
level domains could take a while(!), and they might wonder at
your motives). So from an input validation perspective, a dotted
quad should not cause the parser to stop, alas.

There is nothing stopping the authors of BIND (or other DNS
programs) printing a warning if they see an "all digits and
dots" domain name where the standards mandate a "domain name",
but they would still have to serve it to be standard conforming.
Perhaps an "allow-dottedquad-dnames" option is needed, so people
would have to switch it on.

I wonder what sendmail does if an IP address appears in the MX,
but this corresponds to a genuine domain name? I don't have
sendmail around anymore....

Perhaps I'll lobby for 256 TLD's with the alternative root
co-ordination guys.... Okay 0 and 255 probably won't get me much
mail, but you never know when they'll come in handy. 

I bet a couple of well placed wildcard MX records in or around
the "root" domain could clear up a multitude of e-mail problems,
once people realise it is all getting blackholed. I suspect the
management lack sufficent of the bastard operator from hell
mentality.....