rsalz@bbn.com (Rich Salz) (11/08/88)
>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. Fair enough. Just don't complain when domain-only sites start turning up and they refuse to send mail to you. -- Please send comp.sources.unix-related mail to rsalz@uunet.uu.net.
blarson@skat.usc.edu (Bob Larson) (11/09/88)
In article <1165@fig.bbn.com> rsalz@bbn.com (Rich Salz) writes: >[quoting from someone else:] >>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. >Just don't complain when domain-only sites start turning up >and they refuse to send mail to you. ?? start ?? I'm postmaster on 4 such systems, and I don't plan on changing the mailer in that area. (Actually, they do use a table of local systems so non-domain names work on our primenet.) -- Bob Larson Arpa: Blarson@Ecla.Usc.Edu blarson@skat.usc.edu Uucp: {sdcrdcf,cit-vax}!oberon!skat!blarson Prime mailing list: info-prime-request%ais1@ecla.usc.edu oberon!ais1!info-prime-request
jc@minya.UUCP (John Chambers) (11/14/88)
In article <1165@fig.bbn.com>, rsalz@bbn.com (Rich Salz) writes: > >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. > > Fair enough. > > Just don't complain when domain-only sites start turning up > and they refuse to send mail to you. Time once again to point out that uucp is not an email package; it is a file-transfer package, with email layered on top of it. This isn't a facetious bit of pickiness. It is quite common to make transient uucp hookups (often via a null-modem cable) for the purpose of transferring a few Mbytes of files into a machine, and then unplug them and carry one of them off somewhere. If doing this required registering with some bureaucracy somewhere, it would change a few-minutes job to a few-weeks job. The fact that email comes along with such a transient uucp link is not of much interest to the people who set it up (other than the fact that "mail old!new!root" is a quick test of bidirectional connectivity if you have just linked "new" to "old"). It's also common for a set of machines in a lab to have ad-hoc uucp links (again, usually via null modem cables) that last for days, weeks, or months. The users usually don't know or care that there is a big world of email outside. So what if some turkey on one of the machines makes an outside link? Why should the users of the other machines be expected to register with the bureaucracy, when they never asked to be part of the email system? Why should their internal use of uucp for file transfer obligate them to fill out papers for some outside email authority that they've often never even heard of? -- John Chambers <{adelie,ima,maynard,mit-eddie}!minya!{jc,root}> (617/484-6393) [Any errors in the above are due to failures in the logic of the keyboard, not in the fingers that did the typing.]
rsalz@bbn.com (Rich Salz) (11/19/88)
Me: > Just don't complain when domain-only sites start turning up > and they refuse to send mail to you. jc@minya.UUCP (John Chambers): >Time once again to point out that uucp is not an email package; >it is a file-transfer package, with email layered on top of it. > >This isn't a facetious bit of pickiness. It is quite common to >make transient uucp hookups (often via a null-modem cable) for >the purpose of transferring a few Mbytes of files into a machine, >and then unplug them and carry one of them off somewhere. If >doing this required registering with some bureaucracy somewhere, >it would change a few-minutes job to a few-weeks job. Fine. Just don't complain when a domain-only site refuses to send mail to you. This is not a facetious repetition. I said "a" won't work, you said "but I wanna do b." Read carefully. NOBODY requires that the UUCP name match the domain name. It's a lot easier, but it is NOT required. Finally, there is no bureacracy somewhere that requires you to fill do ANY paperwork. When you unpack your new machine you call somebody INSIDE YOUR ORGANIZATION and say "anybody using the name 'host'?" > Why should their internal use of uucp for file transfer >obligate them to fill out papers for some outside email authority >that they've often never even heard of? It doesn't require any outside authority: No mail? No domain name necessary. Domain name necessary? Ask your local authority. Got it? /rich $alz -- Please send comp.sources.unix-related mail to rsalz@uunet.uu.net.
davidsen@steinmetz.ge.com (William E. Davidsen Jr) (11/23/88)
In article <1218@fig.bbn.com> rsalz@bbn.com (Rich Salz) writes: | Finally, there is no bureacracy somewhere that requires you to | fill do ANY paperwork. When you unpack your new machine you call | somebody INSIDE YOUR ORGANIZATION and say "anybody using the | name 'host'?" What organization? With the advent of cheap unix-pcs and <$300 versions of UNIX which include uucp, there are thousands of sites which are not in an organization. Or more accurately they are in usenet. If there were no usenet maps, I would agree that no one should look for a free lunch and ask that maps be created. Given that the information is there, why not treat .uucp as an organization? | > Why should their internal use of uucp for file transfer | >obligate them to fill out papers for some outside email authority | >that they've often never even heard of? | | It doesn't require any outside authority: | No mail? No domain name necessary. | Domain name necessary? Ask your local authority. The local athority is the usenet maps for many sites. That is what registering .uucp would make official. There are just too many sites who are on usenet for many of us to ignore, even (or particularly) commercial sites. Domains were intended to be a solution, not a religion. The idea that some domains be formed to represent organizations is obvious; group all of the GE sites as a whole and call us something.ge.com and let GE handle delivery. This has nothing to do with geographic location, just connectivity. The .us domain is purely geographic, and there is little reason to think that sites in the same area would be in some way connected. Trying to use location as a domain is for the gratification of domain gurus, since there is no reason to think that sixwbn.albany.ny.us would even know that bilzvax.albany.ny.us exists. That is, machines are allowed to *register* but there seems to be no reason to believe that they can connect. How much better, then, to use .uucp as the domain for the usenet. Machines in the map can connect, and I can assume that joesbar.uucp will somewhat be reachable from armpitz.uucp. A domain should be composed of a group of machines which are all known and deliverable from a gateway. What gateway do we have for .us? If I want to register bilzvax.albany.ny.us and I'm the first one in that domain do I have to become a gateway? Do I agree to connect to every machine in the area, or must they connect to me? Fortunately good ideas are hard to keep down. As long as sites use .uucp in a way that's meaningful to them it will be workable, even if not blessed by an athority. I realize that uunet would like to get a registration fee and connect fee from every site on the net, but I doubt that it will happen. ________________________________________________________________ This one is a personal opinion, mine alone. -- bill davidsen (wedu@ge-crd.arpa) {uunet | philabs}!steinmetz!crdos1!davidsen "Stupidity, like virtue, is its own reward" -me
jim@eda.com (Jim Budler) (11/23/88)
In article <12649@steinmetz.ge.com> davidsen@crdos1.UUCP (bill davidsen) writes: | In article <1218@fig.bbn.com> rsalz@bbn.com (Rich Salz) writes: | [...] | | somebody INSIDE YOUR ORGANIZATION and say "anybody using the | | name 'host'?" [...] | What organization? With the advent of cheap unix-pcs and <$300 [...] So fine, in order to get on UUPC this cheap unix-pc has to arrange a connect with SOME OTHER MACHINE. | Domains were intended to be a solution, not a religion. The idea that | some domains be formed to represent organizations is obvious; group all They can't be a solution if people don't use them. | The .us domain is purely geographic, and there is little reason to | think that sites in the same area would be in some way connected. Trying | to use location as a domain is for the gratification of domain gurus, | since there is no reason to think that sixwbn.albany.ny.us would even | know that bilzvax.albany.ny.us exists. That is, machines are allowed to | *register* but there seems to be no reason to believe that they can | connect. As much reason as there is to believe they cannot connect. This one paragraph is the reason I am responding to this posting. Why use that statement "for the gratification of the domain gurus" at all. Because they are trying to find a solution to a difficult problem, you talk about it as if they were working hard at the problem for their own jollies, not to solve a problem. And .uucp is an unworkable solution. It is world-wide in scope, therefore to pick a name, you have to search the world to make sure the name you pick is unique. The point you miss is that the entire purpose of domains, whether organizational, or geographical, is to reduce the size of the area YOU have to search for uniqueness. | How much better, then, to use .uucp as the domain for the usenet. | Machines in the map can connect, and I can assume that joesbar.uucp will They can? Well if you are registered in a domain, you are in the map, aren't you? So your two hypothetical machines in .ny.us can connect. | somewhat be reachable from armpitz.uucp. ^^^^^^^^ key word and concept. | A domain should be composed of | a group of machines which are all known and deliverable from a gateway. | What gateway do we have for .us? From d.Top: apple .arpa, .com, .gov, .mil, .edu, .org, .net, .us, .ca decwrl .arpa, .com, .gov, .mil, .edu, .org, .net, .us, .ca harvard .arpa, .com, .gov, .mil, .edu, .org, .net, .us, .ca rutgers .arpa, .com, .gov, .mil, .edu, .org, .net, .us, .ca talcott .arpa, .com, .gov, .mil, .edu, .org, .net, .us, .ca ucbvax .arpa, .com, .gov, .mil, .edu, .org, .net, .us, .ca uunet .arpa, .com, .gov, .mil, .edu, .org, .net, .us, .ca ^^^ | | Fortunately good ideas are hard to keep down. As long as sites use | .uucp in a way that's meaningful to them it will be workable, even if | not blessed by an athority. I realize that uunet would like to get a | registration fee and connect fee from every site on the net, but I doubt | that it will happen. [repeated] | How much better, then, to use .uucp as the domain for the usenet. | Machines in the map can connect, and I can assume that joesbar.uucp will The only reason this works FOR ATTEMPTING TO MAIL TO SITES WITHIN THE INTERNET is that the same listed as the root for .us, included above, are also listed in the maps as the domaing root for, TA-DA .UUCP!!! So all the time you have been complaining that the domain gurus are trying to take away your beloved .uucp, (you're right they are), they have been quietly *keeping* it working until they have a *working* replacement in hand. Because, despite your arguments, .uucp could cease to work at a moments notice if the Internet decided to disallow it as they have threatened to in the past. The effect on you? Well you suddenly couldn't get mail to a lot of sites you used to get mail to, and they suddenly couldn't get mail to you, as they used to. And one of the sites 'volunteering' to carry this traffic to YOU, even though you are not paying them a penny is the very site you accused of trying merely to increase their connect fees. My opinionated opinions are the sole property of me. jim -- Jim Budler address = uucp: ...!{decwrl,uunet}!eda!jim OR domain: jim@eda.com #define disclaimer "I do not speak for my employer" #define truth "I speak for myself" #define result "variable"
dhesi@bsu-cs.UUCP (Rahul Dhesi) (11/24/88)
.UUCP is already a domain name. This is my de facto observation. -- Rahul Dhesi UUCP: <backbones>!{iuvax,pur-ee}!bsu-cs!dhesi
dpz@dorm.rutgers.edu (David P. Zimmerman) (11/24/88)
In article <4859@bsu-cs.UUCP> dhesi@bsu-cs.UUCP (Rahul Dhesi) writes: > .UUCP is already a domain name. This is my de facto observation. My de facto (?!?) observation is that the Internet root nameservers don't know diddly-squat about you or how to get to you. David -- David P. Zimmerman, the Dorm Networking Pilot Project, the UUCP Project, etc dpz@dorm.rutgers.edu rutgers!dpz dpzimmerman@zodiac.bitnet
lear@NET.BIO.NET (Eliot Lear) (11/24/88)
In article <4859@bsu-cs.UUCP> dhesi@bsu-cs.UUCP (Rahul Dhesi) writes: > .UUCP is already a domain name. This is my de facto observation. Not according to my nameserver. -- Eliot Lear [lear@net.bio.net]
pcg@aber-cs.UUCP (Piercarlo Grandi) (11/24/88)
In article <12649@steinmetz.ge.com> davidsen@crdos1.UUCP (bill davidsen) writes: What organization? With the advent of cheap unix-pcs and <$300 versions of UNIX which include uucp, there are thousands of sites which are not in an organization. Or more accurately they are in usenet. If there were no usenet maps, I would agree that no one should look for a free lunch and ask that maps be created. Given that the information is there, why not treat .uucp as an organization? This is of great interest to me, as I had and will have just one of those cheapo unixes at home. Actually, usenet maps are abolutely necessary. Usenet is a voluntary, cooperative, no obligations network. There must be a way to know how to reach another site, and the best way is for everybody to publish their links in some sort of directory; this implies that everybody is giving everybody permission to use them as a relay, in exchange for the same privilege. I mean, everybody but AT&T, that wants everybody to believe they are a leaf node... There are others that do cheat, trying to masquerade as leaf nodes, but thank goodness this is not terribly widespread. The problem of what you do in a cooperative environment when everybody takes advantage of your connectivity is real, and the only solution so far is the uunet one (i.e. you get a bit less cooperative). Domains were intended to be a solution, not a religion. The idea that some domains be formed to represent organizations is obvious; group all of the GE sites as a whole and call us something.ge.com and let GE handle delivery. This has nothing to do with geographic location, just connectivity. Actually, bangistas and domainistas are two factions that will be forever at odds, like blanco and colorado parties in latin american countries :-). Domains, in their strictest interpretation, have nothing to do with either location o connectivity; they just provide some means to help generate unique names, and keep lists of them. Associating routers/nameservers or gateways (the two are different things) with levels of the domain tree maybe a convenience, but is not mandatory at all. In other words, a domainista world can be as flat an address space as you want. I do not see why ever bangistas would be opposed to domain based NAMING. In my opinion the best possible world is one in which every machine has a domain based name, bang routing is used, and every machine on the bang route undertakes to do its best to get the message to the next machine, and NO more than that. In this way people that like other guys do routing for them would say a.b.c.d!user and the local map would turn that to e.f!a.b.c.d!user and e.f would turn that into g.h!a.b.c.d!user and so on, each stage adding only the name of the next stage; people that do prefer source routing would specify as complete a path as they can think of with their maps, and all nodes in between would only optimize the path to the next element in it. If any stage were to know of any gateway into a domain, if one existed, they could make use of that. This organization would allow the full continuum between static and dynamic routing, and between hierarchical, gateway based, and flat, map based, oath finding. The .us domain is purely geographic, and there is little reason to think that sites in the same area would be in some way connected. Trying to use location as a domain is for the gratification of domain gurus, since there is no reason to think that sixwbn.albany.ny.us would even know that bilzvax.albany.ny.us exists. That is, machines are allowed to *register* but there seems to be no reason to believe that they can connect. Exactly! the point is that domains need not have gateways at all (ehr, in my opinion, certinaly not in that of the guys that run the internet). Domains are quite useful exactly to allow orderly registration and cataloging of unique names. How much better, then, to use .uucp as the domain for the usenet. By virtue of the discussion above, no. Using .uucp simply reproduces the problem of generating unique names easily. Unless of course you mean to have subdomains under .uucp; but then it has been persuasively argued that top levels that are network names ought to disappear (the crucial points are that networks are paths, and you can be connected to many networks). -- Piercarlo "Peter" Grandi INET: pcg@cs.aber.ac.uk Sw.Eng. Group, Dept. of Computer Science UUCP: ...!mcvax!ukc!aber-cs!pcg UCW, Penglais, Aberystwyth, WALES SY23 3BZ (UK)
jc@minya.UUCP (John Chambers) (11/26/88)
> > Why should their internal use of uucp for file transfer > >obligate them to fill out papers for some outside email authority > >that they've often never even heard of? > > It doesn't require any outside authority: > No mail? No domain name necessary. > Domain name necessary? Ask your local authority. Actually, what I was talking about is perhaps best exemplified by a scenario (with names changed to protect the innocent ;-): While working in their lab, Joe & Mary make a few uucp links via some null modems, including links to their development machine ("dev") that is a Sun or some other BSD system on which /bin/rmail has been replaced with a little monster that sends uucp mail to sendmail. They need to get some stuff from test machine "x" to test machine "y", which aren't directly connected. They know that "uucp foo dev!y!tmp/bar" doesn't work, though they haven't spent the time trying to figure out why, and it's just an ASCII source file, so they decide to try "mail <foo dev!y!root" instead. On dev, as usual, nobody has yet quite figured out how to get sendmail to work right, so when the mail reaches dev and rmail drops it in sendmail's lap, sendmail bounces it to some SNMP forwarder in Texas or Palo Also or Hong Kong, because there's a "q.y.foo.bar" in /etc/hosts. Thirteen hops later, the mail has passed through Bitnet and Decnet routers, and falls into the grip of one of the UUCP Mapping Project's mailers, who notes mail coming from a machine "x" at a place it hasn't seen an "x" before. Somehow, the mailer figures out a path back, and a couple days later, after they've forgotten all about the puzzle of the lost mail, Joe and Mary get a notice that they are supposed to register their machine "x" with the mapping project. They also get similar mail from each of the other email systems, and spend some time being somewhat amused before bringing it to the attention of the guru@dev, who also finds it amusing. Of course, the main reaction of Joe and Mary is to ask "What the @#*$@&^ is going on here; who ARE these people?" and then to go back to work (or drinking coffee or playing hack or whatever). Meanwhile the flames rage outside. But they are oblivious to it all, and will go on to offend the email world a few more times before the project's done. Eventually someone installs usenet on dev, and they read a few articles in some of the email newsgroups, don't find them interesting, and unsubscribe without realizing that people were talking about them. Perhaps guru@dev figures out a way to make it work. Likely he/she discovers that saying "y.uucp" corrects the problem, and they use that, except when they forget. After a while, the uucp links are torn down and/or switched around to other configurations. The test machines have their board shuffled, and their network names change 8 or 12 times. Some of the new names elicit requests for registration, some don't. The folks in the lab spend a little time digging through manuals for an explanation, don't find it, shrug and go about their business. As I said, it's not at all hypothetical. However, it is at least mildly amusing. -- John Chambers <{adelie,ima,maynard,mit-eddie}!minya!{jc,root}> (617/484-6393) [Any errors in the above are due to failures in the logic of the keyboard, not in the fingers that did the typing.]
davidsen@steinmetz.ge.com (William E. Davidsen Jr) (11/30/88)
In article <372@eda.com> jim@eda.com (Jim Budler) writes: | In article <12649@steinmetz.ge.com> davidsen@crdos1.UUCP (bill davidsen) writes: | | Domains were intended to be a solution, not a religion. The idea that | | some domains be formed to represent organizations is obvious; group all | | They can't be a solution if people don't use them. Look at your return addresses. Do they say ".UUCP"? People are using it, I not only have no complaint with domains, I have been working on mail problems here for two years, trying to get 500 (more or less) computers with different vendors, operating systems, and physical connects to talk to one another. I love domains, it's just that .UUCP works practically as a domain, and I can't see trying to break it and forcing people to convert their mailers, maps, user interfaces, mailing lists, forwarders, and whatever else just because having a network be a domain offends someone. It meets three criteria: there are gateways to other networks, there is a naming athority (map project) and there is a standard way to perform routing (maps, pathalias, smail). | | [ paragraph trimmed here ] | As much reason as there is to believe they cannot connect. This one paragraph | is the reason I am responding to this posting. Why use that statement "for the | gratification of the domain gurus" at all. Because they are trying to find a | solution to a difficult problem, you talk about it as if they were working | hard at the problem for their own jollies, not to solve a problem. Not a question of jollies, it's a question of trying to take a workable solution (internet domains) and make it fit all cases, even when some other form of domains is also workable. "When all you have is a hammer, everything looks like a nail." | And .uucp is an unworkable solution. It is world-wide in scope, therefore | to pick a name, you have to search the world to make sure the name you pick try "usenet maps" ^^^^^ | is unique. I admit I'm a pragmatist; if it works it not unworkable. | The point you miss is that the entire purpose of domains, whether | organizational, or geographical, is to reduce the size of the area | YOU have to search for uniqueness. This is true, but is NOT the only intent of domains. They are supposed to reduce or eliminate the host table concept, and to allow me to reach a machine without having a complete route to it. | Because, despite your arguments, .uucp could cease to work at a moments | notice if the Internet decided to disallow it as they have threatened to | in the past. The effect on you? Well you suddenly couldn't get mail | to a lot of sites you used to get mail to, and they suddenly couldn't | get mail to you, as they used to. Interesting thought... if a site shows us as a gateway from internet to .UUCP, and we are willing to act as a gateway, how does anyone make it "cease to work?" Will we be thrown off the internet if we accept mail for an unregistered domain? | And one of the sites 'volunteering' to carry this traffic to YOU, even | though you are not paying them a penny is the very site you accused | of trying merely to increase their connect fees. I'm sorry you read it that way. Rereading it, I'm sorry I put it quite that way. My point is not that any site is trying to get rich on registration fees, but that users won't pay it because their mail works without it. Large sites may well register and connect, but the guy in his atic won't, not should he have to. As for paying fees, if you mean uunet, while I don't sign the check we do, as an organization, pay substantial fees for their services, and no one (particularly me) has been suggesting that they are not worth it *to a commercial organization*. As long as .UUCP works, as long as there is a legitimate reason for reaching sites addressed in that way, and as long as that is a domain name in common use, I will support it both as an individual on the machines I run personally, and as a needed service to insure prompt and correct delivery of mail to serve the needs of the company for which I work. I think we have beat this to death, but would be glad to continue the discussion by mail if someone feels it waranted. -- bill davidsen (wedu@ge-crd.arpa) {uunet | philabs}!steinmetz!crdos1!davidsen "Stupidity, like virtue, is its own reward" -me