wcwells@ucbopal.CC (William C. Wells) (10/24/84)
Forwarded from Usenet "net.mail" news group:
---------
>From gnu@sun.uucp Tue Oct 16 23:14:12 1984
Newsgroups: net.mail
Subject: Re: parsing host1!user@host2 -- the real fix
Article-ID: net.mail/528
Several people have recently suggested that ( and ) be used in
addresses to define precedence. While this would be a great idea
if it had been standardized, instead most of the left- and right-
matching characters have been standardized in the Arpanet formats
to mean various other things:
address (comments)
comments <address>
We could perhaps use [ ] or { } to accomplish this, but this will not
work unless every host we go through uses the same conventions. The
only precedence conventions defined in the Arpanet RFC's do it by
quoting; if you want to nest quoted things you have to double all the
internal quotes. Needless to say this makes for very messy looking
addresses which are easy to type wrong. This is one of the failings
of the Arpanet RFC's and one I hope they'll address in the future.
Most people have come around to agreeing that if we all standardized on
one naming convention, rather than mixing two or more conventions
using ! @ : ^ . % etc, we could all talk to each other. Furthermore,
that the Arpanet domain naming convention is reasonable enough to use.
The rest of the question as far as UUCP mail is concerned revolves
around two implementation questions:
* How do we take an address like mark@cbosgd.att.uucp and deliver
the mail to the right place (routing)
* How do we implement the above without breaking all the existing
uucp sites? (eg, avoid routing mail which should not get
rerouted; avoid generating return addresses old sites can't handle)
There's a task force (cbosgd!uucp-mail) working on these issues now and
software is being readied for eventual release in net.sources. Stay tuned.