From: CRDGW2::CRDGW2::MRGATE::"SMTP::CUNYVM.CUNY.EDU::POSTMAST%YMIR.BITNET" 16-SEP-1989 00:22 To: MRGATE::"ARISIA::EVERHART" Subj: RE: PMDF and All-in-1 Resent-From: @CUNYVM.CUNY.EDU:postmast@YMIR.BITNET Received: from YMIR.BITNET by CUNYVM.CUNY.EDU (IBM VM SMTP R1.2.1MX) with BSMTP id 1896; Sat, 16 Sep 89 00:22:54 EDT Resent-Message-Id: <527B1C05549FE01D0F@YMIR.BITNET> Message-Id: <528A3643A19FC000B9@HMCVAX.BITNET> Received: from HMCVAX.BITNET by YMIR.BITNET; Fri, 15 Sep 89 20:17 PDT Resent-Date: Fri, 15 Sep 89 20:30 PDT Date: Fri, 15 Sep 89 19:13 PDT From: Ned Freed Subject: RE: PMDF and All-in-1 Resent-To: INFO-PMDF-LIST2@YMIR.BITNET To: DPMSYS@RITVAX.BITNET, INFO-PMDF@YMIR.BITNET Errors-To: postmast@YMIR.BITNET X-Vms-To: IN%"DPMSYS@RITVAX.BITNET" X-Vms-Cc: IPMDF Let's backtrack a little on this issue. PMDF has to be able to deal with stuff like: MRGATE::"A1::All-in-none user" that it gets from VMSmail. PMDF elects to convert such an address into forms like "MRGATE::\"All-in-none user\""@localhost because it is unable to otherwise unwind this mess (at least not without know all the ins-and-outs of A1 mailing). Now, this address causes problems with mailers that don't support RFC822 properly. A second issue comes when such an RFC822 address must be present to VMS MAIL. This is a different issue, but PMDF solves it too. An RFC822 address of the form "funny user name"@host will be converted to IN%"'funny user name'@host" This is legal VMS MAIL. The MRGATE mess above, if received by a PMDF system, would be converted into: IN%"'MRGATE::\'All-in-none user\''@localhost" which is legal and replyable. In other words, I claim that VMSMAIL-PMDF-RFC822-PMDF-VMSMAIL works completely correctly in all cases (at least insofar as address translation is concerned). The address may get incredibly ugly, horrendous, and otherwise they may smell very bad, but they do work. This is because the translation takes care to follow the restrictions of both VMS MAIL (the primary one here is that double quotes cannot appear inside the IN%"" construct no matter what) and of RFC822 (which, frankly, is no less ideosyncratic), and to be able to completely invert the translations it does do. Apart from a few bug fixes along the way, I've never seen anything that indicates this stuff is busted or incomplete. You can say there's problems with the parsing in VMS MAIL. I would agree -- the quoting is at odds with both common sense and with the rest of VMS. BUT IT WAS NEVER DOCUMENTED TO WORK A PARTICULAR WAY. I don't feel right saying it is busted when I don't know the design criteria that went into making it what it is. (Now the fact that leaving the trailing " off one of these suckers sends VMS MAIL into an infinite loop -- that's a bug.) So where are the problems? Well, they arise mostly in mailers that don't follow RFC822. There are a lot of them. I recently had another glitch pointed out to me (which I have not verified) -- it seems that Pony Express makes the assumption that doubled quoting works in VMS MAIL addresses. Sorry Peter, it does not. As I've said before, I sure wish it did!!! If anyone can find an example where this stuff breaks down that cannot be traced to a broken mailer of some other ilk, I'd like to see it very much. In the meantime, I'm sticking by my statement that PMDF can handle anything I've ever seen VMS MAIL throw at it, and in a way that won't upset VMS MAIL. I felt bad about treading so close to the edges of RFC822 in all of this. The problem is, if I avoid the edges of the RFC, I cannot come up with any quoting scheme that will work. There's only barely enough quoting to make it work now. Ned