🚀 go-pugleaf

In article <57fcdv-sem@svusenet.ubs.ch>, Daniel G. Kluge wrote:

>scripts which should parse such silly beasts. According to RFC 882 (
>DOMAIN NAMES - CONCEPTS and FACILITIES) the parts of domain-names have
>to be of the following structure (as perl-regexp)
>[a..z][a..z\d-]*[0..9a..z]/i.

Ahem, should be   /[a-z]([a-z0-9-]*[a-z0-9])?/i
One-letter host names have been allowed since the dawn of humanity, in
fact, one of the top-level nameservers used to be "a.isi.edu".

But as I said, RFC 1123 allowed leading digits:

                  /[a-z0-9]([a-z0-9-]*[a-z0-9])?/i

The question was if some newer RFC exists that legalized '_' .


>But if you go and ask in comp.protocols.tcp-ip.domains, you will get
>no consensus, since a lot of people use '_' and '/' in their domain
>and hostnames.

Slashes!!! This is completely sick. Who would possibly use *SLASHES*
in hostnames? How should a program parse http://foo.ch/bar.ch/baz/ ?
Should it try to do a DNS lookup on "foo.ch/bar.ch" ?

"A lot of people" is an exaggeration, IMHO. In the past, this happened
within .ch when a new organization joined the net, and it usually
took them about a year to change that.
A few collected access logs from our machines shows:

                  30341 Fully qualified host names
                    544 with _ in the host part (153 from within .ch (*))
                      4 with _ in the domain part (all from outside .ch)
and indeed:           7 with / in the host part

The problem with this is that when NICs start to allow underscores
without sanctioning that practice in an RFC, how can they disallow
other funny characters? After all, if 3Com got their 3com.com (that was
the main reason for the change in RFC 1123); then why can AT&T not
have at&t.com ? What about

  lookhere!^[[2J.screenblank.org    (^[ == ASCII ESC)
  \0\0\0.zero.null.net
  not^H^H^Hmsn.com




 - Andi Karrer