Path: utzoo!attcan!utgpu!news-server.csri.toronto.edu!clyde.concordia.ca!uunet!mcsun!ukc!dcl-cs!aber-cs!odin!pcg From: pcg@cs.aber.ac.uk (Piercarlo Grandi) Newsgroups: comp.protocols.tcp-ip.domains Subject: Re: CS top-level domain and its impact on the UK? Message-ID: Date: 28 Jul 90 16:55:00 GMT References: <9007101320.AA03269@dockside.mitre.org> <1990Jul23.065901.8741@ifi.uio.no> Sender: pcg@aber-cs.UUCP Organization: Coleg Prifysgol Cymru Lines: 178 In-reply-to: enag@ifi.uio.no's message of 23 Jul 90 06:59:01 GMT "enag" == Erik Naggum writes: enag> Originator: enag@slembe.ifi.uio.no enag> In article , enag> pcg@cs.aber.ac.uk (Piercarlo Grandi) writes: pcg> The fact that in the Internet the top level domain country code is GB pcg> for the United Kingdom enag> It's also my understanding that it was the then administrators of enag> NRS or whatever who decided to implement their own hack with enag> respect to domain names, and ignore ISO 3166. Which they are perfectly entitled to, because Janet and the NRS are totally independent of the DNS and Internet. The use of ISO 3166 is an Internet convention; other networks can use other conventions. pcg> When I see discussions by people quoting RFCs and wishing that pcg> everybody conform to them, e.g. register in the DNS, etc..., they seem pcg> to constantly forget that the Internet is *one* net, even if the pcg> largest. enag> If you want to talk to or with the Internet, follow Internet rules. enag> It's as simple as that. Say that to the X.400 people. They sure will listen. :-) :-). enag> If you continue to insist that you should be allowed to follow enag> whatever rules you fancy on the Internet, you will have to be enag> taken off the net, either by force from the Internet side, or by enag> default, since you won't be reachable. That's not the problem, the problem is whether the Internet has ways to address of issue of communicating with networks that do not follow Internet rules. As far as I am aware, the % hack is the only half baked way, and all its critics clamor instead for its removal and everybody registering with the DNS, and having MXs point at the relevant gateways. This is clearly *impossible*. It will never happen, not even within the USA. It is not even desirable. But a lot of people in this newgroup forget about that, as I have already remarked. They complain about Janet/NRS names not complying to RFC syntax. Well, precisely -- they have nothing to do with Internet syntax, they belong to a different world, like the UUCP ones (to which exactly the same ideas apply). This is forgotten by all the Internetcentrics that for example think that everything that has dots in it must be an Internet name, and would therefore do rabid name resolution based on that delusion. The % hack is not an hack to do source routing; it is a way to say @ in a non Internet sense within an Internet address. If we had a network out there where the dot had a special meaning, we would need a dot synonym as well probably. Should a gateway rewrite any addresses in NRS format to DNS format? even if in signatures? Even if they make no sense in the DNS? I think not, IMNHO, because in general it is not *possible*, because ther emay be no equivalent. Even in the easiear cases, it may be better not to provide translations between two different naming syntaxes/conventions. Suppose that I send a message to the Internet from Janet; I should address it (and I do) as enag%slembe.ifi.uio.no@uk.ac.nsfnet-relay because I must address the gateway using NRS conventions (and this explains what is at the right of the @), and I cannot use the DNS address the target while using the @ contained in it because that has meaning for the NRS. Almost by chance the Janet-Internet gateway uses the percent hack to resolve the overloading of @, but it could decide to accept e.g. '#' as a synonym for the DNS @, or provide some form of quoting (which would be best, or my idea to use a Then-To: header for gateway use). When I get a reply from the Internet, it might/should be addressed as (assuming that the Internet side of the gateway uses % as a synonym for @ as well): pcg%uk.ac.aber.cs@nsfnet-rely.Janet.net pcg%uk.ac.aber.cs@nsfnet-rely.NRS.org pcg%uk.ac.aber.cs@nsfnet-rely.ac.gb (which one do you think would be most appropriate, incidentally?) but not as pcg@cs.aber.ac.uk which is not a legal Internet address (ironically it may be currently valid though, because of the double registration of Janet names in the NRS and the DNS, if I remember correctly). Of course there is no reason to assume that I need type all this myself; the MTA could apply heuristics (which break down in the CS case for example, but that is not a problem with naming schemes, only with the heuristics) and detect when the supplied address is outside the locally valid naming zone and generate the reference to the appropriate gateway in the apporpriate form, e.g. using tables, as we seem condemned to do, both Janet and Internet, when addressing somebody in the X.400 world. pcg> If the people on the Internet harbor the delusion that everybody pcg> must respect the relevant RFCs, or else, they will lose pcg> connectivity with not just the UUCP world, but also the UK, BITNET, pcg> and many others. [ ... avoiding ... ] enag> This is largely achieved by having gateways that accept and follow the enag> rules of both the networks they're on. You seem to think that the enag> Internet should accept all sorts of standards, and have failed to enag> grasp the idea of a gateway. This is not the Internet's fault, enag> either. I think you have failed to understand the context of this discussion at all, which was why the % hack is necessary. Let me repeat here in yet another form: it is necessary to hide @'s that have meaning in non Internet naming zone addresses from over eager Internet resolvers, and thus avoid breakdowns in the heuristics used to convert addresses between two different conventions. Were it true that gateways were properly designed! And were it true that gateways *could* be properly designed, that they could convert all addresses meaningfully. On a fairly different subject: enag> This doesn't scan well. Are you saying that many sites are migrating enag> as quickly away from ISO/GOSIP networks as they can? Oh yes. enag> I wasn't aware of any ISO/GOSIP networks that any site could enag> migrate away from to begin with, Well, in continental Europe the *only* WANs available are ISO oriented (and eventually GOSIP oriented). What is happening that the level of service is crude enough and expensive enough that many sites are running a largish segment of the Internet in Europe using these WANs as transport, with TCP/IP tunnelling on X.25 (or ISDN when available, or even on point-to-point lines, even switched ones, where not available). Norway is different because it was on the ARPAnet since SATNET days. enag> and it seems to me that this ISO/GOSIP stuff is being migrated enag> _to_ by people who want to win contracts with the GOSIP folks. That is something that is happening with the *government*. Where people have to pay with their own funds, the move away from ISO and to TCP/IP is obvious to the naked eye. In Europe it is slowest in the UK, where Janet is (nearly) 100% government subsidized, and for some EEC projects, where the same idea applies. enag> I know I would be very happy to be wrong, but the trade rags haven't enag> made me feel that way for a long time. No, no. You can be happy, try to figure out the number of names in the Internet that have a .de .dk .nl .fr .se .fi top domain and their increase in the last few years. De facto, the Internet is today the dominant international WAN in continental Europe, at least as far as the research community is concerned. enag> Give me three years, and you'll have Internet enag> all over the United Kingdom. Any takers? :-) If only... I think it could be done in three months. I reckon virtually all Janet sites actually run Internet protocols locally, and by buying off-the-shelf switching equipment and reusing the current lines a switch to TCP/IP interconnection among them could be done fairly quickly. -- Piercarlo "Peter" Grandi | ARPA: pcg%cs.aber.ac.uk@nsfnet-relay.ac.uk Dept of CS, UCW Aberystwyth | UUCP: ...!mcsun!ukc!aber-cs!pcg Penglais, Aberystwyth SY23 3BZ, UK | INET: pcg@cs.aber.ac.uk