Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!ames!think!barmar From: barmar@think.COM (Barry Margolin) Newsgroups: comp.unix.questions Subject: Re: Internet address syntax (and semantics, for that matter) Message-ID: <29792@news.Think.COM> Date: 20 Sep 89 20:27:15 GMT References: <2274@ttardis.UUCP> <29725@news.Think.COM> <4270@buengc.BU.EDU> Sender: news@Think.COM Organization: Thinking Machines Corporation, Cambridge MA, USA Lines: 43 In article <4270@buengc.BU.EDU> bph@buengc.bu.edu (Blair P. Houghton) writes: >In article <29725@news.Think.COM> barmar@think.COM (Barry Margolin) writes: >>In article <2274@ttardis.UUCP> jcf3703@ttardis.UUCP (chap flack) writes: >>>cfctech!sharkey!chenry@carleton.edu. >> cfctech! ((sharkey!chenry) @carlton.edu) >The way a bang-at address is interpreted (is it really undefined) >is to send to the at-machine, then start down the bang-path. It really is undefined. Many systems even parse it differently depending on the source. For instance, if the address foo!bar@baz comes in from the UUCP mail server the "!" is parsed first; if the same address were to come from an SMTP mailer server the "@" would be parsed first. >It's obvious that our friend only has to use the address "chenry@carleton", >since the mail got through carleton to cfctech, and then to sharkey, >which bounced it. That's a possibility. In fact, that was my first thought. But I assumed that he would know whether his own machine knew how to route SMTP addresses automatically, and wouldn't have bothered with the UUCP addressing if he didn't need it. >What I'd like to know is, why not just fix mailers to accept the address as > > "carleton to cfctech to..." > >instead of all this !%@. That wouldn't be any easier than getting them to all have smart routers so users could all just say "user@domain"; they both involve fixing thousands of mailers. And during the transition period (five to ten years), you'll have the problem of THREE different mail address formats instead of just two. The general direction of mail routing is away from requiring users to specify routes explicitly (e.g. the MX records in the domain system); computers are supposed to be good at figuring out those kinds of things. Barry Margolin Thinking Machines Corp. barmar@think.com {uunet,harvard}!think!barmar