clive@ixi.UUCP (Clive Feather) (05/18/89)
There have been several postings recently implying that the .uucp
domain is not a real one, and that sites in this domain should get
themselves registered in a real domain.
Firstly, can someone please explain to me why .uucp is not a real
domain ? There seems to be some kind of registration mechanism to
prevent name clashes (at least, I had to register with uknet@ukc.ac.uk),
and UKC claims (in a monthly posting with "uk" distribution) that all
mail addressed to any .uucp site in the UK will be correctly routed
by them.
Secondly, is the only legal domain in the UK the .uk domain ? If
so, am I correct in thinking that UKC is the registration point for
this domain ? When I attempted to get a .uk registration from UKC,
I was pointed on to UMIST. Between the two, they implied that .uk
registration is equivalent to registration with JANET (a UK academic
network), and commercial organisations must pay UKL 100 to register.
Since I don't seem to be getting anything for my money, why pay ?
Incidentally, the .uucp registration was free.
Thirdly, is there anything to prevent UKC from constructing a new
domain (.uucp.uk or .gen.uk), and using this instead of the apparently
dubious .uucp domain ? If not, could anyone at UKC who is reading
this please explain why they don't ?
Fourthly, suppose that IXI were registered as "ixi.domain". If I
understand the domain system correctly, people here can invent names
of the form "user@machine.ixi.domain", or "user@machine.net.ixi.domain",
and, providing the one machine here with the modem can cope with
routing these internally, anyone else in the world can use these
addresses and rely on it being delivered because no-one outside IXI
looks at anything other than the "ixi.domain" bit to route it to here.
Am I correct ? If so, am I allowed to do the same thing with the
.uucp domain (i.e. use addresses of the form "user@machine.ixi.uucp" ?
If not, why not ?
If any of these questions are answered by a RFC, could someone please
e-mail me a copy.
I am sorry if anyone feels that this is trivial and/or treading on
someone else's toes (particulary UKC, who do a wonderful job) - a
relatively new sysadmin wants to know the right way to do things
(if Usenet has such a thing).
Please e-mail, and I'll summarise to the net.
--
Clive D.W. Feather clive@ixi.uucp
IXI Limited ...!mcvax!ukc!ixi!clive (riskier)
+44 223 462 131neil@ist.CO.UK (Neil Todd) (05/24/89)
From article <168@ixi.UUCP>, by clive@ixi.UUCP (Clive Feather): > There have been several postings recently implying that the .uucp > domain is not a real one, and that sites in this domain should get > themselves registered in a real domain. > > Firstly, can someone please explain to me why .uucp is not a real > domain ? One problem is that the uucp domain is a flat namespace, there can be exactly one ixi.uucp in the world. > There seems to be some kind of registration mechanism to > prevent name clashes (at least, I had to register with uknet@ukc.ac.uk), > and UKC claims (in a monthly posting with "uk" distribution) that all > mail addressed to any .uucp site in the UK will be correctly routed > by them. This is correct, UKC acts as a clearing house for UKnet (the UK part of the UUCP network) name registrations. > > Secondly, is the only legal domain in the UK the .uk domain ? The only TOP level domain for the UK is .uk, correct. This is registered with the NIC in the U.S. Beneath that there are .ac (Academic Community) .co (Commercials) .mod (Ministry of Defense - Military) and, I think, .rsgb (Radio Society of Great Britain, Packet Radio ??) > If so, am I correct in thinking that UKC is the registration point for > this domain ? No (life would be too easy otherwise :-) ) > When I attempted to get a .uk registration from UKC, > I was pointed on to UMIST. Strictly I think that they probably pointed you at the NRS ("Name Registration Scheme"), which is administered by the University of Manchester Regional Computer Centre (UMRCC) on behalf of the JNT (Joint Network Team). > Between the two, they implied that .uk > registration is equivalent to registration with JANET (a UK academic > network), and commercial organisations must pay UKL 100 to register. > Since I don't seem to be getting anything for my money, why pay ? > Incidentally, the .uucp registration was free. Well, you are not registering for JANET (Joint Academic Network). You are registering a name within the UK domain, strictly the CO.UK subdomain. This registeration will record details for how to reach your site for various services etc (Mail, FTP, X29, etc). Other people in the UK can use this information to communicate with your site, rather than routing through a third party (UKC in the case of mail). This information is kept in an online database ("the" NRS) can can be transfered to User sites. You have the ability to update your entries at any time. This is only part of the story, As I mentioned earlier, .uucp is a flat domain space, somebody somewhere has to know where to send the message so that it arrives at clive@ixi.uucp, what tells whether the site is in the U.S. or the U.K. ? The advantage of the domain scheme is that there are many more clues as to where to send the message, indeed on the Internet there are things floating around called "MX" (Mail eXchanger) records which indicate how to get to a particular domain (so mail to anything ending in .uk can be routed to a site that knows much more about the details of sites within the .uk domain. > > Thirdly, is there anything to prevent UKC from constructing a new > domain (.uucp.uk or .gen.uk), and using this instead of the apparently > dubious .uucp domain ? If not, could anyone at UKC who is reading > this please explain why they don't ? > (I'm not from UKC, but....) A .uucp.uk goes against the spirit of multilevel domains - I also doubt whether they've got the time and money to do such a thing. > Fourthly, suppose that IXI were registered as "ixi.domain". If I > understand the domain system correctly, people here can invent names > of the form "user@machine.ixi.domain", or "user@machine.net.ixi.domain", > and, providing the one machine here with the modem can cope with > routing these internally, anyone else in the world can use these > addresses and rely on it being delivered because no-one outside IXI > looks at anything other than the "ixi.domain" bit to route it to here. > Am I correct ? If so, am I allowed to do the same thing with the > .uucp domain (i.e. use addresses of the form "user@machine.ixi.uucp" ? > If not, why not ? Well, yes and no, you really should ensure that if you allow internal names to be seen by the outside world, then you should register them in the NRS (no they don't cost 100 pounds each!). Most mailers that understand domain based addressing could cope if you didn't - whant you should do is to use the abilities of you mail system (sendmail, MMDF, etc) to hide the internal names from the world and support a aliasing scheme on your mail gateway machine ("the one with the modem on"), this means that Users can change machines without having to tell everybody they exchange mail with that they have got an new address. > > If any of these questions are answered by a RFC, could someone please > e-mail me a copy. > > I am sorry if anyone feels that this is trivial and/or treading on > someone else's toes (particulary UKC, who do a wonderful job) - a > relatively new sysadmin wants to know the right way to do things > (if Usenet has such a thing). Unfortunately it has lots of "right ways"! > > Please e-mail, and I'll summarise to the net. I read this, but you cover so many points that I feel a followup was justified Neil Todd (neil@gid.co.uk, neil@gid.uucp) GiD Ltd +44 491 671964 > -- > Clive D.W. Feather clive@ixi.uucp > IXI Limited ...!mcvax!ukc!ixi!clive (riskier) > +44 223 462 131
dg@lakart.UUCP (David Goodenough) (06/06/89)
From article <1393@istop.ist.CO.UK>, by neil@ist.CO.UK (Neil Todd): > One problem is that the uucp domain is a flat namespace, there can be > exactly one ixi.uucp in the world. This is not a problem in itself - .bitnet is a real live domain, and it's as flat as a millpond. No, the _REAL_ problem is that .UUCP is an anarchy: there is no central authority. For example, look at the recent fulcrum uproar, and there are two edsel's in .UUCP. To take things to their logical extreme, my home machine is pallio.uucp (unique as far as I was able to check, and I looked very carefully). However, there is nothing except my concience (and good sense) to prevent me from calling it mcvax, or decvax, or uunet or anything else. It is left as an excercise to evaluate the level of confusion and trauma that would result :-) -- dg@lakart.UUCP - David Goodenough +---+ IHS | +-+-+ ....... !harvard!xait!lakart!dg +-+-+ | AKA: dg%lakart.uucp@xait.xerox.com +---+
wisner@mica.Berkeley.EDU (Bill Wisner) (06/08/89)
(David Goodenough) >This is not a problem in itself - .bitnet is a real live domain, and it's >as flat as a millpond. Huh? As a domain, .BITNET is every bit as bogus as .UUCP. As a network, though, BITNET is rather smaller than UUCP.
campbell@redsox.bsw.com (Larry Campbell) (06/08/89)
In article <560@lakart.UUCP> dg@lakart.UUCP (David Goodenough) writes:
-No, the _REAL_ problem is that .UUCP is an anarchy: there is no central
-authority. For example, look at the recent fulcrum uproar, and there are
-two edsel's in .UUCP. To take things to their logical extreme, my home
-machine is pallio.uucp (unique as far as I was able to check, and I looked
-very carefully). However, there is nothing except my concience (and good
-sense) to prevent me from calling it mcvax, or decvax, or uunet or anything
-else. It is left as an excercise to evaluate the level of confusion and
-trauma that would result :-)
But there _is_ a central authority: the uucp map project. I thought the
map folks (Mel et alia) were supposed to check your entry and reject it if
it's a duplicate name. Now, you can still go and pick a dumb name and not
list yourself in the map, but as far as I and many others are concerned (we
run a fascist mailer here), if you're not in the map, you don't exist.
You might have trouble getting mail, but you won't mess anyone else up.
--
Larry Campbell The Boston Software Works, Inc.
campbell@bsw.com 120 Fulton Street
wjh12!redsox!campbell Boston, MA 02146mcb@ncis.tis.llnl.gov (Michael C. Berch) (06/09/89)
In article <560@lakart.UUCP> dg@lakart.UUCP (David Goodenough) writes: > From article <1393@istop.ist.CO.UK>, by neil@ist.CO.UK (Neil Todd): > > One problem is that the uucp domain is a flat namespace, there can be > > exactly one ixi.uucp in the world. > > This is not a problem in itself - .bitnet is a real live domain, and it's > as flat as a millpond. BITNET is *not* a "real live domain" at all. Its status is exactly the same as UUCP, that is, a pseudo-domain understood by some mailers. The "real live domains" (i.e., top-level domains) consist of the set {COM, EDU, GOV, MIL, NET, NET, ORG, US} for the U.S., and the ISO 2-letter country abbreviations for other countries (of which there are approximately 35 or 40 currently organized and registered). The listing of registered top-level domains is in the file DOMAINS.TXT in the directory NETINFO on the host SRI-NIC.ARPA, available for anonymous FTP. -- Michael C. Berch mcb@tis.llnl.gov / uunet!tis.llnl.gov!mcb
lear@NET.BIO.NET (Eliot Lear) (06/09/89)
I believe BITNET's long term plans are to move to DNS. -- Eliot Lear [lear@net.bio.net]
dg@lakart.UUCP (David Goodenough) (06/12/89)
campbell@redsox.bsw.com (Larry Campbell) sez: > In article <560@lakart.UUCP> dg@lakart.UUCP (Who, me?) writes: > -For example, look at the recent fulcrum uproar, and there are > -two edsel's in .UUCP. > > But there _is_ a central authority: the uucp map project. I thought the > map folks (Mel et alia) were supposed to check your entry and reject it if > it's a duplicate name. Looks as if it isn't quite working out. (see above) The problem is that I don't have to register pallio or anything, just add the line: # lakart pallio(DAILY) to lakart's map entry. Now, to add (say) a bogus mcvax, I just add: # lakart mcvax(DIRECT) to my map entry, and: mcvax Never to L.sys, and the dirty deed is done. The software on pallio would need a recompilation, but I'm not going to do it since pallio only has 20 Megs of storage. BTW _I_STAND_CORRECTED_ - _.BITNET_IS_NOT_A_REAL_DOMAIN_ - _50_LASHES_ _WITH_A_WET_NOODLE_WILL_BE_ADMINISTERED_ -- dg@lakart.UUCP - David Goodenough +---+ IHS | +-+-+ ....... !harvard!xait!lakart!dg +-+-+ | AKA: dg%lakart.uucp@xait.xerox.com +---+