henry@utzoo.UUCP (Henry Spencer) (07/07/87)
> Hell, from one standpoint, I'd call utzoo a broken cog. The only reasonable > argument I got against arbitrary newsgroup creation was that 16 bit > machines can't hack lots of groups... Actually, there is a more decisive argument: what good are lots of groups? Is there really a reason why we need comp.lang.c.null-pointers? All it does is clutter up an already-cluttered name space for a group that is seldom if ever used. (Incidentally, if you really mean "arbitrary" newsgroup creation, that way lies chaos -- people who cannot even get subject lines more or less right will not magically switch to carefully choosing appropriate names for their 700 brand-new newsgroups.) > In fact, the problem is really that > 16 bit "BACKBONE" machines can't hack it. Leaf sites are only constrained > by the groups they actually receive. You greatly underestimate the continuing presence of 16-bit machines on the net; there are many with full feeds, although they *are* definitely in the minority these days. > So isn't it time that utzoo gave up that hallowed position? Well, as a matter of fact utzoo is in the early stages of mutating into a larger machine. The argument stands, however. > In fact, isn't it time to start the > total phaseout of the backbone completely? Have you a substitute to suggest? I assure you that if there were an alternative that could provide equivalent service without involving any of the backbone admins, every last one of us would be *delighted* to give up our "net god" positions. I have (only half in jest) suggested that the backbone sites ought to all take a six-month holiday, and watch the chaos and madness that would ensue: if no substitute evolved, a lot of people would have learned from the experience, and if one *did* evolve, GREAT, we'd be off the hook! Being a backbone sysadmin is more hassle and less fun than a lot of people seem to think. > UUNET can provide a total feed for about $200/month. Several UUNET's > can supply the whole network with a new backbone. Yup. Is your site paying it? No? Then keep quiet about it, please. (Better yet, join up for it: UUNET needs more customers.) If UUNET can survive -- it's not out of the woods yet financially -- it is a good idea. If it succeeds big, undoubtedly there will be interest in doing the same thing elsewhere, although expanding UUNET itself to handle more customers might be better. Furthermore, I think every backbone admin would agree *wholeheartedly* that having a backbone which *charges* for its news feeds is a damn fine idea and long overdue. -- Mars must wait -- we have un- Henry Spencer @ U of Toronto Zoology finished business on the Moon. {allegra,ihnp4,decvax,pyramid}!utzoo!henry
brad@looking.UUCP (07/07/87)
In article <8261@utzoo.UUCP> henry@utzoo.UUCP (Henry Spencer) writes: >Actually, there is a more decisive argument: what good are lots of groups? >Is there really a reason why we need comp.lang.c.null-pointers? All it does >is clutter up an already-cluttered name space for a group that is seldom if >ever used. (Incidentally, if you really mean "arbitrary" newsgroup creation, >that way lies chaos -- people who cannot even get subject lines more or less >right will not magically switch to carefully choosing appropriate names for >their 700 brand-new newsgroups.) >-- >Mars must wait -- we have Henry Spencer @ U of Toronto Zoology I don't see a problem with comp.lang.c.null-pointers. If there is a hierarchy arranged by keyword creation moderators, there is no trouble in comprehending thousands and thousands of groups. If you are talking about "name-space" in a software sense, then you are agreeing with the poster who complained that old software was being used as an excuse not to create new groups. -- Brad Templeton, Looking Glass Software Ltd. - Waterloo, Ontario 519/884-7473
rick@pcrat.UUCP (Rick Richardson) (07/07/87)
The "> >" are from me, uunet!pcrat!rick (Rick Richardson) In article <8261@utzoo.UUCP>, henry@utzoo.UUCP (Henry Spencer) writes: > > Hell, from one standpoint, I'd call utzoo a broken cog. The only reasonable > > argument I got against arbitrary newsgroup creation was that 16 bit > > machines can't hack lots of groups... > > Actually, there is a more decisive argument: what good are lots of groups? No, I don't mean absolutly arbitrary creation of *.*.*.null-pointers (although Guy might want to post to that one :-) ). But from time to time new group requests get posted that are unique, and the category broad enough that the group shouldn't get shot down just because it may be lower volume than other groups. Rec.gambling comes to mind here. > You greatly underestimate the continuing presence of 16-bit machines on the > net; there are many with full feeds, although they *are* definitely in the > minority these days. This site is a 16 bit site. I don't ever intend to get or need a full feed. If a 16 bit machine starts having trouble with the number of newsgroups, then they can just pare down the feed. Else, C news could be made to work for more newsgroups on 16 bit machines (but what I waste of effort that would prove to be). > > In fact, isn't it time to start the > > total phaseout of the backbone completely? > > Have you a substitute to suggest? I did make the following suggestion: > > UUNET can provide a total feed for about $200/month. Several UUNET's > > can supply the whole network with a new backbone. > > Yup. Is your site paying it? No? Then keep quiet about it, please. > (Better yet, join up for it: UUNET needs more customers.) If you'd looked at the path I sent this by, you'd have seen that I AM already a UUNET customer. So I'll be noisy about it, thank you. BTW, my $200 figure is from taking Rick Adam's $175 figure, and adding in an approximation of the alt.* groups which were just recently added to UUNET. My bill was about $65 last month. I think it'l be around $120 this month, but I snarfed a bunch of stuff from the sources archives on UUNET. Most of the cost was the $3/hour of transmission time, at 1200 baud (Tymnet doesn't have 2400 baud access in Red Bank, my local node, sigh). Presumably, you could cut costs even further by using PC Pursuit to call down to Arlington (I think that's where UUNET is) and into UUNET. Rick Adams just said that the 1 UUNET machine could potentially handle 10% of all known USENET sites. 10 machines around the country wouldn't be bad from a reliability/logistics standpoint. But even 10 machines won't be needed at first, since only 1 machine at a site need be a subscriber to UUNET. So, how much are the backbones paying to carry USENET? Less than $200/month? More? If it's more, then why haven't you P.O. 'ed yourself a subscription to UUNET yet? If it's less, perhaps you'd like a little autonomy for a price, or direct access to source archives? -- Rick Richardson, President, PC Research, Inc. (201) 542-3734 (voice, nights) OR (201) 834-1378 (voice, days) seismo!uunet!pcrat!rick
geoff@utstat.UUCP (07/09/87)
In article <362@pcrat.UUCP> rick@pcrat.UUCP (Rick Richardson) writes: > Else, C news >could be made to work for more newsgroups on 16 bit machines (but what >I waste of effort that would prove to be). C news already runs on 16-bit machines: utzoo and utcsstat are PDP-11s. C rnews weakly prefers machines large enough to cache the active and sys files in memory, but it will merely run a bit more slowly on very small machines. Current sys and active files (with the possible exception of ucbvax's sys file :-) are small enough that they fit in memory on PDP-11s. -- Geoff Collyer utzoo!utstat!geoff, utstat.toronto.{edu,cdn}!geoff