wales@ucla-cs.UUCP (04/12/85)
This comment has been made previously by other people, I think, but I
feel prompted to mention it again.
It would really be nice if people could phase out the use of the colon
in mail addresses. Colons create all kinds of problems in ARPA-land for
at least three reasons:
(1) The colon has special meaning in ARPANET addresses under RFC822, in
connection with the "source-route" or "audit-trail" notation (such
as "<@MIT-MC:FRED@MIT-OZ>".
Hence, a colon in an ARPANET address is destined to lose big unless
the part of the address to the left of the at-sign is surrounded by
double quotes.
(2) Even using quotes may not always help, since quite a few mail sys-
tems currently in use don't always (or ever) handle quotes properly.
This is wrong, but it is also reality.
(3) Another reason quotes may not help with addresses containing colons
is that several mail systems queue up each outgoing ARPANET message
with a "control" line at the beginning of the message, specifying
the destination host and other parameters; these parameters are sep-
arated on the control line by -- you guessed it -- colons.
Again, this is bletcherously wrong -- but it is a reality, and get-
ting all sites in question to fix it is wishful thinking at best.
I am currently aware of at least two kinds of mail network which make
heavy use of the colon in addresses:
(1) The old "Berkeley network" ("Berknet"), which uses addresses of the
form "site:user".
(2) DECNET, which uses addresses of the form "SITE::USER".
If you are responsible for an environment which uses the colon as a
site-user delimiter in addresses, PLEASE give some consideration to the
possibility of using something else which will be more compatible with
RFC822. Your contribution can help eliminate colons in addresses in our
lifetime!
--
Rich Wales // UCLA Computer Science Department // +1 213-825-5683
3531 Boelter Hall // Los Angeles, California 90024 // USA
wales@UCLA-LOCUS.ARPA {ihnp4,ucbvax}!ucla-cs!waleshokey@plus5.UUCP (Hokey) (04/14/85)
I don't know what to say about Berknet, but I do know that decwrl accepts addresses of the form "user@site.dec". I don't know if they accept the uucp project form "site.dec!user". I also don't know how hard it would be for decwrl to change its sendmail.cf file to convert all recipient and sender fields to make it easier on users on the different nets. This is a very diicult problem. -- Hokey ..ihnp4!plus5!hokey 314-725-9492
johnsson@decwrl.UUCP (Richard Johnsson) (04/17/85)
Dotted addresses reportedly cause problems in some uucp implementations.
At decwrl, we try to provide mail gateway functions in the sender's native
syntax. Thus we accept:
user@site.dec from local senders
user%site.dec@decwrl.arpa from the Internet
<@decwrl.arpa:user@site.dec> also from Internet
...!decwrl!dec-rhea!dec-site!user from UUCP
We also take site::user from local senders just to make life interesting.
Some combinations/modifications of these addresses also work, but nothing is
guaranteed.
--
Richard Johnsson, DEC Western Software Lab, Palo Alto, CA
UUCP: {decvax,ucbvax,ihnp4,allegra}!decwrl!johnsson
ARPA/CSNET: johnsson@decwrl DEC ENet: rhea::johnsson
phone: +1 415 853 6676avolio@decuac.UUCP (Frederick M. Avolio) (04/20/85)
At DECUAC (we are on UUCP and DEC's Enet) any mail coming in in the
forms u@h.DEC, h!u.DEC, h::u goes to the ENET. At present we do not do
anything with adresses coming out. (IE, you could see an address
"...decuac!orphan::fuzbu".
I am very interested in making our sendmail.cf be as accommodating as
possible. In other words, I entertain any and all suggestions.
--
Fred Avolio {decvax,seismo}!decuac!avolio 301/731-4100 x4227