lipman%rhea.DEC@decwrl.ARPA (04/16/84)
In my previous submission to header-people, I noted that I had eliminated identifying RHEA.DEC as RHEA.ARPA in the "Received: " headers. I had chosen to identify RHEA.DEC as DEC-RHEA.ARPA at least partially due to the way 4.2 BSD sendmail generates that Received header message. A second reason for changing the host name and not just the domain was the argument that user@rhea, user@rhea.ARPA, and user@rice-rhea.ARPA should all be equivalent. I based this feeling on the argument that I and my user community wanted user@Shasta, user@Shasta.ARPA, and user@SU-Shasta.ARPA to all be equivalent. Consistency might even make some of this obscure junk understandable. But Brian Reid points out in the note enclosed below that neither user@rhea, nor user@rhea.ARPA should be arriving at decwrl from the outside world. Nicknames are not supposed to leave the local machine on which they are used, they are supposed to be converted into the real host name. Note that I believe 4.2 BSD Sendmail violates this rule. If this is indeed the case, then it is only my local user community I should be concerned about and for them, user@rhea should mean dec-rhea. I'll bet there is a mailbag worth of opinions out there! Peter Lipman DEC Western Research Laboratory 4410 El Camino Real Los Altos, California 94022 (415) 949-0776 uucp: {decvax, ucbvax, ihnp4, allegra} decwrl!lipman ARPA net: lipman@decwrl.ARPA DEC-Enet: RHEA::LIPMAN --------------- Note from Brian Reid ------------------------ From: RHEA::DECWRL::"reid@Glacier.ARPA" "Brian Reid" 15-APR-1984 16:21 To: Peter Lipman <lipman@decwrl> Subj: rhea and domains Received: from DECWRL by DEC-RHEA with SMTP; Sun, 15 Apr 84 16:21-PST Received: from Glacier (su-glacier.arpa.ARPA) by decwrl.ARPA (4.22.01/4.7.16) id AA24254; Sun, 15 Apr 84 16:18:03 pst Date: Sun, 15 Apr 84 16:16:11 pst Cc: Forest Baskett <baskett@decwrl> Message-Id: <8404160018.AA24254@decwrl.ARPA> Peter, I saw a message on acetes that there is a naming conflict between rice-rhea and DEC's rhea. I don't think there should be a need to panic. One of the main purposes for having domains in symbolic addressing is to provide a way of circumventing this problem. The main use, so far, of domain addressing in the internet has been to assist subnet routing of mail. Perhaps this has obscured the more important use of domains for name qualification. Thus, the name RHEA.DEC is a different name from RHEA.ARPA. As it turns out, if you read the fine print in RFC822, it is illegal to put domain qualifications on nicknames, e.g. RHEA.ARPA is not a legal name, but RICE-RHEA.ARPA is a legal name. Furthermore, the standard requires that all exported text contain only fully-qualified names. This means that once a message leaves the .ARPA domain, it cannot contain the name "RHEA" anywhere in it, or even "RHEA.ARPA". This important restriction provides the basis for making your system do what you want. When you are on, say, Acetes, and you reference a name "RHEA", you are allowed to assume that you are in some domain besides the .ARPA domain. Specifically, you are in the .DEC domain. You can think of the task of name resolution as being like looking down a search path of domains: first you check your own home domain, then you check neighboring domains, and so forth. If you did it this way, then the name "rhea" would always resolve to RHEA.DEC and not RICE-RHEA.ARPA when it was used in an unqualified context on any machine in the DEC domain. The moral: you don't have to change the name of your machine unless you want to. Brian
wcwells%ucbopal.CC@Berkeley.ARPA (William C. Wells) (04/17/84)
In reply to: Message-Id: <8404162100.AA01284@decwrl.ARPA> Date: Monday, 16 Apr 1984 11:47:08-PST From: lipman%rhea.DEC@decwrl.ARPA To: header-people@mit-mc.ARPA Subject: More on RHEA.DEC ..... Nicknames are not supposed to leave the local machine on which they are used, they are supposed to be converted into the real host name. Note that I believe 4.2 BSD Sendmail violates this rule. ..... Peter Lipman DEC Western Research Laboratory 4410 El Camino Real Los Altos, California 94022 (415) 949-0776 Let's stop blaming the Unix sendmail program and work toward getting local sendmail configuration files fixed so gatewaying is handled properly. At Berkeley we permit users to use different nicknames and top-domain names with sendmail. Our mail system also transforms non-Internet addresses into Internet addresses of before transmitting the message to an Internet mail site. Bill Wells wcwells@Berkeley.ARPA