sean@garfield.UUCP (Sean Byrne) (01/18/84)
>Yes, there is another Garfield, a research VAX-750 at the CS department of >Columbia University, New York City. It was not dial-upable a few months ago, >and must not be yet. Garfield has sibling machines named "Sylvester" and >"Heathcliff", by the way. It appears that a problem that has never occurred on Usenet before has come to be. We here at garfield (Memorial University) would like to hear more about this "other" garfield (Columbia University). The problem has never occurred before, apparently because we (garfield) do not talk to the same sites (Us: ihnp4,allegra,akgua,utcsrgv etc.) as `they'. We would appreciate hearing from someone at the ~other~ garfield, or someone who knows when they joined the net, or anything else. Hanxs. -- Sean Byrne Memorial University of NF St. John's, Nfld UUCP: {allegra, ihnp4, utcsrgv}!garfield!sean
mark@cbosgd.UUCP (Mark Horton) (01/25/84)
Policy on conflicting names is that the first site to get the name REGISTERED gets it. As the only current registry is Usenet, and since the garfield in Newfoundland has been registered and well known on Usenet for some time now, it seems pretty clear that Newfoundland got there first. This isn't the first time this happened. There are two machines calling themselves iuvax - one at Indiana University (they have been registered for a long time) and one at SRI (on which Kashtan did Eunice) that has never really been announced. Of course the one in Indiana has priority. I also seem to recall another collision, with one of of the sites somewhere in Florida, but I can't recall who. Does anyone remember? Note that a UUCP registry is being started up, and once that gets going, it will be the authority which gives out names. As UUCP mushrooms, collisions are going to become far more commonplace. To get a name, a site will have to request it from the registry and confirm that it isn't already registered. Mark Horton
north@down.UUCP (Professor X) (01/26/84)
How can there possibly be an "authority" that decides what uucp name you're allowed to give your machine? If I pick a name the "authority" doesn't like, how is it going to stop me and my uucp neighbors from using it? For purposes such as automatic mail routing it is convenient to think of the uucp network as having unique absolute names but in fact the naming convention is strictly relative. The fact that there are two distinct sites named iuvax makes this obvious. Stephen C North
mark@cbosgd.UUCP (Mark Horton) (01/28/84)
How can there possibly be an "authority" that decides what uucp name you're allowed to give your machine? If I pick a name the "authority" doesn't like, how is it going to stop me and my uucp neighbors from using it? For purposes such as automatic mail routing it is convenient to think of the uucp network as having unique absolute names but in fact the naming convention is strictly relative. The fact that there are two distinct sites named iuvax makes this obvious. Stephen C North I'm sure you understand that the original UUCP software was designed under the assumption that each site has a unique name. The possibility of two sites having the same name is not considered to be a "feature" by more than a handful of way-out anarchists, as far as I know. Most of us view it as a problem that will prevent mail from being delivered properly. (In fact, already mail sent to the VORTEX machine on DECNET is being misdelivered to the vortex machine on UUCP.) It seems obvious to me that the only way to prevent the problem from reaching epidemic proportions is to create a central clearinghouse that will prevent duplicate names from being assigned. If you people at Princeton were to rename your university MIT, you could communicate internally just fine using the name MIT. You might even get a few places nearby to call you MIT as well. The rest of the world would continue to call the university in Cambridge MIT and ignore you (and send all your mail to Cambridge.) The same logic applies to UUCP. Mark Horton
north@down.UUCP (Professor X) (01/30/84)
How will the centralized clearinghouse prevent such duplicate name problems from arising? What authority will it have? The fact of its existence? Hey, I'd like it to work as much as you would, but experience shows that not everyone is smart enough or cooperative enough to abide by "the rules." Stephen C North
piet@mcvax.UUCP (Piet Beertema) (01/30/84)
>How can there possibly be an "authority" that decides what >uucp name you're allowed to give your machine? Since it's anarchy that rules the net, it's common sense that has to prevent it from getting screwed up. Thus a site joining in later and choosing a name already in use has a choice between anarchy and keeping that name or common sense and changing it. -- Piet Beertema CWI (Center for Math. & Comp. Science), Amsterdam ...{decvax,philabs}!mcvax!piet
piet@mcvax.UUCP (Piet Beertema) (02/07/84)
>A more obvious problem: how is the new site to know of the coordinator >until he's been on the net (and flamed at ...) ? The site he hooks up to should provide him with the latest net.news.map contents and/or the routing map. By checking that against the chosen sitename you can be reasonably sure about the uniqueness of that name. -- Piet Beertema CWI (Center for Math. & Comp. Science), Amsterdam ...{decvax,philabs}!mcvax!piet
jsq@ut-sally.UUCP (John Quarterman) (02/12/84)
> >A more obvious problem: how is the new site to know of the coordinator > >until he's been on the net (and flamed at ...) ? > The site he hooks up to should provide him with the latest net.news.map > contents and/or the routing map. By checking that against the chosen > sitename you can be reasonably sure about the uniqueness of that name. > -- > Piet Beertema > CWI (Center for Math. & Comp. Science), Amsterdam > ...{decvax,philabs}!mcvax!piet The site providing the connection should check the new name for uniqueness against net.news.map and whatever mail maps are available *before* allowing the new site to hook up. If a new site does get connected with the same name as one that connected earlier, other sites should refuse to use the name and it should not be listed in the news and mail maps; public opinion is the solution as in any good anarchy. -- John Quarterman, CS Dept., University of Texas, Austin, Texas jsq@ut-sally.ARPA, jsq@ut-sally.UUCP, {ihnp4,seismo,ctvax}!ut-sally!jsq
caf@cdi.UUCP (caf) (02/16/84)
A more obvious problem: how is the new site to know of the coordinator until he's been on the net (and flamed at ...) ? -- Chuck Forsberg WA7KGX CDI Portland OR (503)-646-1599 cdi!caf