kls@ditka.UUCP (Karl Swartz) (10/21/88)
In article <8299@alice.UUCP> debra@alice.UUCP () writes: >Sorry I cannot easily reply to you by mail, Bill, but with a site named >"Grumpy" your mail is doomed to go to "grumpy" whenever it goes through >a gateway which converts names to lowercase. The real problem is that mail auto-routed to "Grumpy" will go to 'grumpy' instead. If you specify an explicit bang path, there's no problem unless the path happens to go thru a "rabid router", in which case you'd better find a functional path. Changing the name is, of course, the *right* solution. -- Karl Swartz |UUCP {gatech!emoryu1,uunet!dasys1}!ditka!kls 1-505/667-2402 (work) |ARPA rt1!ditka!kls@hc.dspo.gov 1-505/672-3113 (home) |BIX kswartz "I never let my schooling get in the way of my education." (Twain)
david@ms.uky.edu (David Herron -- One of the vertebrae) (10/21/88)
uh, guys case is insignificant in mail addresses therefore Grumpy == grumpy == grUmPY == ... ad nauseum -- <-- David Herron; an MMDF guy <david@ms.uky.edu> <-- ska: David le casse\*' {rutgers,uunet}!ukma!david, david@UKMA.BITNET <-- <-- Controlled anarchy -- the essence of the net.
kai@uicsrd.csrd.uiuc.edu (10/24/88)
> /* Written 3:41 pm Oct 20, 1988 by kls@ditka.UUCP */ > > Changing the name is, of course, the *right* solution. Are we supposed to register EVERY site on EVERY local network with a name unique to the world, just so that some rude people will be happy? How long before we run out of unique names? I'd rather just register the gateway between our local network and the rest of the world, and let people route through it. If someone far far away tries to route mail through our registered gateway to a non-registered host on our local network, a re-routing host might incorrectly recognize that local name as someone else's registered host, and send the mail to the wrong place. Is that my fault for making sure every hostname we use is unique? A better solution would be to avoid and/or eliminate those obnoxious re-routing programs. Maybe this has been proposed before, but how about a new header, "NOreroute:". I see no benefit to forcefully re-routing all mail. If the address path given to your mailer correctly lists a host you talk to, please LEAVE IT ALONE! Patrick Wolfe (pwolfe@kai.com, uunet!kailand!pwolfe)
karl@triceratops.cis.ohio-state.edu (Karl Kleinpaste) (10/25/88)
kai@uicsrd.csrd.uiuc.edu writes:
Changing the name is, of course, the *right* solution.
Are we supposed to register EVERY site on EVERY local network with
a name unique to the world, just so that some rude people will be
happy? How long before we run out of unique names?
[a] Yes. [b] Ages ago.
If you're going to use a single-component name without any kind of
qualifiers, then Yes, you very much need to register a name which is
guaranteed to be unique in the world. You can't guarantee the
operation of anyone else's mailer. Whether you agree with Rutgers'
operation or not is of no relevance - they actively reroute. That
means that your non-unique hostnames are subject to misdelivery.
The ability to register hosts for UUCP purposes has existed for lo
these several years for exactly this reason. People started having
name collisions a LONG time ago, and the registration process was the
solution.
The only way to have local names that might not be unique is to
qualify them within some domain. Our gateway host is called Tut.
Would you care to guess how many "tut"s are out there? (At least 5
that I can name in 30 seconds or less.) The One and Only Real Tut is
in Finland - guess where mountains of OSU-bound mail ended up until I
fixed that glitch in our news configuration? Now every piece of news
and mail leaving OSU is fully-qualified, much to the consternation of
people who worry (needlessly) about the length of Path: headers and
such - but we have effectively registered ALL of our hosts by using
domain qualifiers. Your (e.g.) `banana' cannot possibly conflict with
my `banana,' because my `banana' is called `banana.cis.ohio-state.edu'
when given proper reference.
Realize also that news posted from a private, name-conflicting,
single-component-name host will not be seen by anyone on the real,
registered host by that name, because the news software's audit trail
of hosts in the Path: header will keep it from going there. This is
Bad.
I've heard all these amazing claims that "I shouldn't have to register
my hosts" but I've never yet seen a good explanation for why it isn't
A Good Idea.
--Karl
rja@edison.GE.COM (rja) (10/26/88)
In article <42300005@uicsrd.csrd.uiuc.edu>, kai@uicsrd.csrd.uiuc.edu writes: > Are we supposed to register EVERY site on EVERY local network with a name > unique to the world, just so that some rude people will be happy? How long > before we run out of unique names? The correct thing to do is to obtain a REAL domain name ( .UUCP is not a real domain ) from the good folks at nic-sri.arpa. Then if you always use your fully-qualified domain names including subdomain names for each machine you might have, no name conflicts can arise. Small sites can join the .US domain wihtout dealing with the NIC directly and should contact westine@isi.edu for more info. > I'd rather just register the gateway between our local network and the rest > of the world, and let people route through it. The key thing is that you should use your valid domain name for each system. A gateway into your domain is common and fairly easy to do. Using the .UUCP kludge just leads to more problems. > If someone far far away tries to route mail through our registered gateway to > a non-registered host on our local network, a re-routing host might > incorrectly recognize that local name as someone else's registered host, and > send the mail to the wrong place. Is that my fault for making sure every > hostname we use is unique? If you use real domain names that are fully qualified this cannot happen. > A better solution would be to avoid and/or eliminate those obnoxious > re-routing programs. Maybe this has been proposed before, but how about a > new header, "NOreroute:". The real solution is for all of the sites to get real domain names. The .US domain removes any former objections based on complexity. I'm glad that people are beginning to think in terms of domainising their site names even in the uucp-only world. I hope everyone gets switched over.
ewiles@netxcom.UUCP (Edwin Wiles) (10/26/88)
In article <25641@tut.cis.ohio-state.edu> karl@triceratops.cis.ohio-state.edu (Karl Kleinpaste) writes: > >I've heard all these amazing claims that "I shouldn't have to register >my hosts" but I've never yet seen a good explanation for why it isn't >A Good Idea. > I've only seen one article that could be interpreted as above, and that one was simply asking whether he *had* to do it. 99.9% of the articles that I have seen have not railed against registration, but instead have spoken out against *agressive rerouting*. The basis for speaking out against it was that you *can* *not* count on the map data being accurate; EVEN IF EVERYONE REGISTERED! EVEN IF THEY SENT MAP UPDATES IN IN A TIMELY FASHION! YOU CAN'T COUNT ON HAVING ACCURATE MAP DATA! There are always local conditions, delays in passing the new map data around, system failures, mistakes in the map data, etc..... Thus, agressive rerouting is a BAD IDEA! It will continue to be a BAD IDEA as long as 99% accurate map data cannot be guaranteed 100% of the time! Be Careful! -- ...!hadron\ "Who?... Me?... WHAT opinions?!?" | Edwin Wiles ...!sundc\ Schedule: (n.) An ever changing | NetExpress Comm., Inc. ...!pyrdc\ nightmare. | 1953 Gallows Rd. Suite 300 ...!uunet!netxcom!ewiles | Vienna, VA 22180
cik@l.cc.purdue.edu (Herman Rubin) (10/27/88)
In article <1016@netxcom.UUCP>, ewiles@netxcom.UUCP (Edwin Wiles) writes: > In article <25641@tut.cis.ohio-state.edu> karl@triceratops.cis.ohio-state.edu (Karl Kleinpaste) writes: [Much omitted; not relevant to this comment.] > Thus, agressive rerouting is a BAD IDEA! It will continue to be a BAD IDEA > as long as 99% accurate map data cannot be guaranteed 100% of the time! The current data is far more than 99% accurate. But considering the tens of thousands of relevant items, that is no good. Aggressive rerouting is a bad idea unless the rerouter takes on the obligation to unreroute in the case of nondelivery. -- Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907 Phone: (317)494-6054 hrubin@l.cc.purdue.edu (Internet, bitnet, UUCP)
snoopy@sopwith.UUCP (Snoopy T. Beagle) (10/30/88)
In article <25641@tut.cis.ohio-state.edu> karl@triceratops.cis.ohio-state.edu (Karl Kleinpaste) writes: | You can't guarantee the operation of anyone else's mailer. | Whether you agree with Rutgers' | operation or not is of no relevance - they actively reroute. That | means that your non-unique hostnames are subject to misdelivery. One can also look at this from the other direction: there have been, there are, and there will always be unregistered machines. You cannot change this no matter how much you implore everyone to register every last machine in the entire universe. Therefore active rerouting is unsafe, and should not be done. _____ /_____\ Snoopy /_______\ |___| tektronix!tekecs!sopwith!snoopy |___| sun!nosun!illian!sopwith!snoopy
kai@uicsrd.csrd.uiuc.edu (11/01/88)
So, if uucp domains are the answer to the active rerouting problem, how come more people aren't using them? We registered KAI.COM over a year and a half ago, about half of our direct uucp connections still don't support domains, and cannot handle replying to messages we generate because of this. The UCB Mail program still generates bogus addresses for replies when the original sender address is in domain format and any other receivers (carbon copies) aren't. Patrick Wolfe (pwolfe@kai.com, uunet!kailand!pwolfe)
karl@triceratops.cis.ohio-state.edu (Karl Kleinpaste) (11/02/88)
kai@uicsrd.csrd.uiuc.edu writes:
So, if uucp domains are the answer to the active rerouting problem, how come
more people aren't using them?
Probably because the UUCP registration process has been reasonably
successful, and most hosts to and from which mail goes are in fact
registered and unique. It's just when you run into a duplicated,
hence unregistered, private host, or when the connectivity info of the
maps is out of date, that you run into serious problems.
--Karl
jc@heart-of-gold (John M Chambers) (11/05/88)
> >I've heard all these amazing claims that "I shouldn't have to register > >my hosts" but I've never yet seen a good explanation for why it isn't > >A Good Idea. > > > > I've only seen one article that could be interpreted as above, and that > one was simply asking whether he *had* to do it. Well, I continually run into a simple, straightforward answer to that question. I routinely deal with a fair number of computer-ignorant users (secretaries, managers, and so on) who of course all have a computer of some sort on their desk. They have no interest whatsoever in becoming computer gurus, for mail or for any other purpose. When the machine gets installed, they just want to turn it on and use it with as little hassle as possible. They tend to see the first few pages of whatever installation guide comes with the machine, and when it does what they need, they totally use interest in anything else. In particular, if someone tells them that they should register their hostname with some self-proclaimed authority somewhere, their response typically is to give an incredulous "What ARE you talking about, turkey?" look and then casually go about the job they are being paid for. In other words, they are NOT going to listen to you. If you expect their workstation to be registered, YOU will have to do the registering. Maybe you can persuade the vendor(s) to include some software that automates this, though I kinda doubt it. Maybe you can persuade me to take care of a few of them, though I myself DO have a job that I'm being paid for, and it only includes babysitting PCs as a side interest. One thing you especially have going against your demand is that they must have a name that doesn't conflict with names used halfway around the world. Their response to that is quite predictable: "Who cares, when I'm just going to exchange files with Sally down the hall?" OK, so you don't think such novices should be allowed to buy machines with email on them. Tough; they're buying them right now, and there's no way you can restrict email to only gurus like yourself. Of course, the arrogance with which much of the email community approaches the issue is also a major turn-off to many users. Perhaps if you could find ways of helping people, rather than insulting them, they'd be more willing to cooperate. Note that arguments of the form "The email system requires that you..." are not convincing. Few users care to spend their time/money solving someone else's problem. If you want cooperation, you must find a way of saying "If you want to use email easily, you should..." Then perhaps they'll get motivated. -- From: John Chambers <mitre-bedford.arpa!heart-of-gold!jc> From ...!linus!!heart-of-gold!jc (John Chambers) Phone 617/217-7780 [Send flames; they keep it cool in this lab :-]
allbery@ncoast.UUCP (Brandon S. Allbery) (11/08/88)
As quoted from <1689@edison.GE.COM> by rja@edison.GE.COM (rja): +--------------- | In article <42300005@uicsrd.csrd.uiuc.edu>, kai@uicsrd.csrd.uiuc.edu writes: | > Are we supposed to register EVERY site on EVERY local network with a name | > unique to the world, just so that some rude people will be happy? How long | > before we run out of unique names? | | The real solution is for all of the sites to get real domain names. The | .US domain removes any former objections based on complexity. | | I'm glad that people are beginning to think in terms of domainising their | site names even in the uucp-only world. I hope everyone gets switched over. +--------------- It'll never happen. Case in point: tdi2, which talks to ncoast and is listed in the maps because *occasionally* mail is sent to/from them from outside of ncoast. Every time the topic comes up with the de-facto administrator of the system (and its little brother tdi1) the response is always the same: "Why should we? We don't use it enough." And yet they use it enough to warrant a map listing. (I have to agree that they don't use it enough to warrant even a .US domain name.) You'll *never* get *all* UUCP sites registered. Face the fact, don't just stand there and tell us that it's the solution to all our problems. ++Brandon -- Brandon S. Allbery, comp.sources.misc moderator and one admin of ncoast PA UN*X uunet!hal.cwru.edu!ncoast!allbery <PREFERRED!> ncoast!allbery@hal.cwru.edu allberyb@skybridge.sdi.cwru.edu <ALSO> allbery@uunet.uu.net comp.sources.misc is moving off ncoast -- please do NOT send submissions direct Send comp.sources.misc submissions to comp-sources-misc@<backbone>.
lear@NET.BIO.NET (Eliot Lear) (11/09/88)
There will come a day (that is still pretty far away) when the rest of us get sick and tired of non-domainist sites, and we will simply refuse to route to them anymore. Then people at sites like tdi2 will no longer be able to receive mail. At that point will the administrator pick up a domain. -- Eliot Lear [lear@net.bio.net]
vixie@decwrl.dec.com (Paul Vixie) (11/10/88)
# There will come a day (that is still pretty far away) when the rest of # us get sick and tired of non-domainist sites, and we will simply # refuse to route to them anymore. Promise? -- Paul Vixie Work: vixie@decwrl.dec.com decwrl!vixie +1 415 853 6600 Play: paul@vixie.sf.ca.us vixie!paul +1 415 864 7013
lear@NET.BIO.NET (Eliot Lear) (11/11/88)
My article comes across a little harsh, so let me rephrase it a little better. The point is that I (and a lot of other people, I think) view the uucp maps as a form of backward compatability for sites that do not use domains. (Are there any sites that cannot have domain names for one reason or another?) In any event, like all obsolete technology, there will probably come a time when there are so few sites using the maps that it will not be cost effective for people to maintain the software. This point in time is quite a ways away, but I have to believe it is there. -- Eliot Lear [lear@net.bio.net]
kai@uicsrd.csrd.uiuc.edu (11/14/88)
> /* Written 11:55 pm Nov 10, 1988 by lear@NET.BIO.NET in uicsrd.csrd.uiuc.edu:comp.mail.uucp */ > The point is that I (and a lot of other people, I think) view the > uucp maps as a form of backward compatability for sites that do not > use domains. ... In any event, like all obsolete technology, > there will probably come a time when there are so few sites using > the maps that it will not be cost effective for people to maintain > the software. As I understand it, domains will only shorten, not eliminate the maps, since you only have to know how to get to the gateway for each domain, instead of each and every host. Unfortunately, to maintain compatibility with sites who don't have registered domains (yet), lots of people are registering all their hosts AND their domain, so the maps are getting larger, not smaller. Luckily, Kailand is connected to Uiucuxc, a very smart host at the University of Illinois, so we don't have to maintain the UUCP maps (or pathalias) at all. Our smail database lists only the sites with which we have direct UUCP connections, and everything else goes to uxc for further delivery. Maybe if more sites used smart-hosts instead of trying to maintain the maps themselves, it wouldn't be so difficult to keep the maps correct and in sync. Not that I LIKE the idea of the maps, I just see it as an unavoidable reality. Patrick Wolfe (pwolfe@kailand.kai.com, uunet!kailand!pwolfe)
chip@ateng.ateng.com (Chip Salzenberg) (11/15/88)
According to lear@NET.BIO.NET (Eliot Lear): >The point is that I (and a lot of other people, I think) view the >uucp maps as a form of backward compatability for sites that do not >use domains. I disagree. If UUCP is being used, you (the sender) still have to know how to get to the *gateway* for the destination domain! Thus maps will be necessary as long as we use UUCP for our mail transport. -- Chip Salzenberg <chip@ateng.com> or <uunet!ateng!chip> A T Engineering Me? Speak for my company? Surely you jest! Beware of programmers carrying screwdrivers.
lear@NET.BIO.NET (Eliot Lear) (11/16/88)
Chip Salzenberg writes: > I disagree. If UUCP is being used, you (the sender) still have to know how > to get to the *gateway* for the destination domain! WElllllll....... There are two problems here, Chip. First, be careful to understand that domains don't have to have gateways, but they do have to have name servers. An illustrative example: Introducing FOO.COM, a multinational enterprise that has offices everywhere you can think of. I want to mail to XYZ.FOO.COM, a branch located in Mountain View. The name server for FOO.COM, located in Tarytown, NY hands me back an MX record saying send it to AMES.ARC.NASA.GOV, located 10 blocks from me. So really what we need is a way for uucp sites to have access to name servers. Yes, this would require extensions to UUCP. However, we are talking near term so such a software solution should not be out of the question. (Can you say DIAL-A-DOMAIN? ;-) Also, keep in mind that one of goals of DNS is to AVOID tables for other sites. A software solution should reflect that theme. -- Eliot Lear [lear@net.bio.net]
Makey@LOGICON.ARPA (Jeff Makey) (11/17/88)
In article <Nov.15.14.30.07.1988.25541@NET.BIO.NET> lear@NET.BIO.NET (Eliot Lear) writes: >So really what we need is a way for uucp sites to have access to name >servers. [implementation details omitted] >Also, keep in mind that one of goals of DNS is to AVOID tables for >other sites. A software solution should reflect that theme. Yes, but you would *still* need tables to tell you how to get to at least the top-level domain servers, and the domain-name resolver would probably keep its own cache (read: table) of paths. Tables cannot be entirely eliminated, but they can be made as small as one entry. There is a tradeoff between local table size and real-time network traffic to resolve names. Has anyone done any modeling to determine the optimum balance between these two? :: Jeff Makey Department of Tautological Pleonasms and Superfluous Redundancies Department Disclaimer: Logicon doesn't even know we're running news. Internet: Makey@LOGICON.ARPA UUCP: nosc!logicon.arpa!Makey
lear@NET.BIO.NET (Eliot Lear) (11/18/88)
I am willing to accept a small table of 12 servers that changes once every two or three months. It is a hell of a lot better than pushing around thousands of entries. It may not even be necessary to keep all the root servers. All you would need would be a list of neighbors that would be willing to do the lookups for you. As to keeping the traffic down, essentially, you would be running a uucp flavored named, so you would want to do things that the real named does, like caching, and preloading of that cache for sites that you tend to speak more to. $BEGIN PIPEDREAM$ Of course, the real solution would be to see everyone on the Internet. Maybe in our lifetimes? $END PIPEDREAM$ Eliot -- Eliot Lear [lear@net.bio.net]
chip@ateng.ateng.com (Chip Salzenberg) (11/26/88)
According to lear@NET.BIO.NET (Eliot Lear): >I am willing to accept a small table of 12 servers that changes once >every two or three months. Can you say, "Long distance charges for each name resolution"? I knew you could. -- Chip Salzenberg <chip@ateng.com> or <uunet!ateng!chip> A T Engineering Me? Speak for my company? Surely you jest! Beware of programmers carrying screwdrivers.