chuqui@nsc.UUCP (Chuq) (08/30/84)
From here on the usenet side of Unix-wizards it's easy to say that 'unix
trivia' belongs in net.games.trivia. Unfortunately, life is a LOT more
complicated than that because of the different ways the newsgroups are
handled.
the ARPA people have the disadvantage that something posted here on usenet
to net.unix and net.unix-wizards will show up in their mailboxes twice,
once to INFO-UNIX, once to INFO-UNIX-WIZARDS. Usenet has the advantage that
the software will (should?) only show it once. When ARPA people post to
both, everyone sees it twice. And we also get to see all of those wonderful
messages about mailing lists that people forget to send to *-REQUEST. All
of this, is, of course, trivia in itself and completely beside the point.
What isn't beside the point is the question raised in all of this--
appropriateness of material on a given topic. This has been a subject of
growing controversy on the Usenet side as the volume of material increases.
There basic problem seems to be that there is no well known definition of
appropriate material. On any given subject we seem to be able to find at
least one person who will yell loudly that the material is bogus and
another will yell JUST as loudly that is isn't. I've noticed over the last
six months that we are starting to spend a lot more time yelling about
things and a lot less time discussing them. A lot of newsgroups are quickly
becoming useless because of the sheer volume in them and the fact that most
of that volume is basically useless material. The causes for that are many:
1) people who respond to things by following up to the group
instead of mailing. How many articles have you seen with the
definition of grep in it? Why weren't these mailed to the person
interested instead of plowing them through the net?
2) People who are new to the net asking questions that they don't
know the answer to, but touching on things that were discussed
before they got on the net.
3) people who are sloppy about their postings, putting things in
the wrong places, asking questions that don't belong on the net
(hom many places do you know of with a unix system that wouldn't
have ONE person who knows what GREP means?) and generally not
thinking about the consequences of their action.
the REAL question is: What can be done?
the real answer is, I don't know. When a mailing list gets large enough on
ARPA, someone takes over as moderator and does some cleanup to make it
tolerable again. This is a BIG advantage of ARPA over usenet. We've tried
in the past to experiment with moderators and every attempt has been yelled
down with screams of dictators and fascists. Usenet people will kill for
the freedom to post garbage, it seems, even if it eventually kills the net.
Because of the lack of coordinated control of the net, I doubt moderators
would work anyway because that assumes that all of the sites will set up
their software to work with moderators. We all know the success of getting
sites to do things like upgrade software and fix bugs, much less install
software to implement controvesial functions.
If we assume that tighter controls on the net are a hopeless wish, the only
other alternative we have is information. We need to teach people to use
the net properly. How? Good question. We've written and posted guides,
we've mailed comments and suggestions to transgressors, we've posted
suggested, screams, yells, and whimpers, and things keep getting worse. I
certainly don't know what will work anymore, nothing that has been tried
seems to.
Perhaps it is time to simply junk Usenet as a failed prototype and come up
with a better communication scheme. Perhaps things will miraculously fix
themselves. I doubt it, though. I've got lots of questions about things,
but no answers, and the future of Usenet as it stands now looks rather
bleak to me. Anyone else out there have some answers that will work?
--
From the depths of the Crystal Cavern: Chuq Von Rospach
{amd,decwrl,fortune,hplabs,ihnp4}!nsc!chuqui nsc!chuqui@decwrl.ARPA
Dreams, dreams, enchanter! Gone with the harp's echo when the strings fall
mute; with the flame's shadow when the fire dies. Be still, and listen.chip@t4test.UUCP (Chip Rosenthal) (08/31/84)
--- REFERENCED ARTICLE --------------------------------------------- >Subject: Re: Origin of the name grep >Newsgroups: net.unix-wizards,net.news >From: chuqui@nsc.UUCP (Chuq) >Date: Wed, 29-Aug-84 23:06:04 PDT > >The ARPA people have the disadvantage that something posted here on usenet >to net.unix and net.unix-wizards will show up in their mailboxes twice, >once to INFO-UNIX, once to INFO-UNIX-WIZARDS. Usenet has the advantage that >the software will (should?) only show it once. When ARPA people post to >both, everyone sees it twice. -------------------------------------------------------------------- I would like to sensetize people to one fact here. It has to do with sites which are running news under Eunice. Either these sites throw away multiple newsgroup postings, or the articles are physically duplicated into each of the newsgroups. So, if somebody posts an article to seven newsgroups, I read it seven times. The number of sites running news under Eunice is probably increasing. Yes, we are still the minority. But if folks can show consideration and minimize their postings to multiple newsgroups, it would be appreciated. Flames about using `real' Unix would not be appreciated. Those sites running Eunice have their reasons, just as the sites running PDP-11's have reasons for not running VAXen. Nope, Eunice is here, and will continue to be used as long so people develop software for the VMS environment. My general comment to the network community is to reconsider when you are going to post to multiple newsgroups. To software developers, please recognize that Eunice exists. If possible, get a Eunice system as part of your beta-site activities. I will certainly offer my services towards these ends. -- Chip Rosenthal, Intel/Santa Clara { idi|intelca|icalqa|kremvax|qubix|ucscc } ! t4test ! { chip|news }
gregbo@houxm.UUCP (Greg Skinner) (09/02/84)
I guess I should respond to this, as I have experienced both sides of the fence (or both sides of the gateway, if you prefer). Before I came to Bell Labs, I did work on the ARPAnet and read net.unix-wizards (actually unix-wizards@brl-tgr.ARPA, I think) and sent a few questions in occasionally. I won't post this to net.unix-wizards though, because I don't think we need to bother the ARPAnet with our internal problems. > From: chuqui@nsc.UUCP (Chuq) > From here on the usenet side of Unix-wizards it's easy to say that 'unix > trivia' belongs in net.games.trivia. Unfortunately, life is a LOT more > complicated than that because of the different ways the newsgroups are > handled. > the ARPA people have the disadvantage that something posted here on usenet > to net.unix and net.unix-wizards will show up in their mailboxes twice, > once to INFO-UNIX, once to INFO-UNIX-WIZARDS. Usenet has the advantage that > the software will (should?) only show it once. When ARPA people post to > both, everyone sees it twice. And we also get to see all of those wonderful > messages about mailing lists that people forget to send to *-REQUEST. All > of this, is, of course, trivia in itself and completely beside the point. For those of you who don't know, our postings show up as mail to ARPAnet people, and their mail shows up as postings. Through various hacks to mail and news software, at the ARPA/USENET gateways one format is turned into another. I can recall when I first read SF-LOVERS I wondered what all the weird addresses were with the ! signs in them, and when I sent replies a lot of them were rejected! Anyhow, back to the main point ... > What isn't beside the point is the question raised in all of this-- > appropriateness of material on a given topic. This has been a subject of > growing controversy on the Usenet side as the volume of material increases. > There basic problem seems to be that there is no well known definition of > appropriate material. On any given subject we seem to be able to find at > least one person who will yell loudly that the material is bogus and > another will yell JUST as loudly that is isn't. I've noticed over the last > six months that we are starting to spend a lot more time yelling about > things and a lot less time discussing them. A lot of newsgroups are quickly > becoming useless because of the sheer volume in them and the fact that most > of that volume is basically useless material. The causes for that are many: > 1) people who respond to things by following up to the group > instead of mailing. How many articles have you seen with the > definition of grep in it? Why weren't these mailed to the person > interested instead of plowing them through the net? Actually, I missed this discussion. There is a lot of "followup fever" on the net these days. I have noticed in a few newsgroups conversations taking place between posters which would have been better continued in private mail. Unfortunately, there is not much that can be done about this because this is the nature of netnews, as opposed to ARPAnet mail. As Chuq says later, to eliminate this problem would require a different implementation of netnews (not necessarily in software but some rules and regulations would have to be implemented, at least). > 2) People who are new to the net asking questions that they don't > know the answer to, but touching on things that were discussed > before they got on the net. Perhaps all system administrators should set up .newsrc's so that their users get net.announce.newusers or whatever that newsgroup is that contains emily-post and all the other introductory material. 3) people who are sloppy about their postings, putting things in the wrong places, asking questions that don't belong on the net (how many places do you know of with a unix system that wouldn't have ONE person who knows what GREP means?) and generally not thinking about the consequences of their action. Actually, I'm surprised when I saw the original, that the poster was from AT&T. I have to agree with Chuq on this one ... especially in the case of unix-wizards, one should be careful in what one posts their. Questions that cannot be answered by co-workers in adjoining offices to yours are probably good candidates for posting to net.unix-wizards, but not much else. > the REAL question is: What can be done? Run TCP/IP instead of UUCP. :-) > the real answer is, I don't know. When a mailing list gets large enough on > ARPA, someone takes over as moderator and does some cleanup to make it > tolerable again. This is a BIG advantage of ARPA over usenet. We've tried > in the past to experiment with moderators and every attempt has been yelled > down with screams of dictators and fascists. Usenet people will kill for > the freedom to post garbage, it seems, even if it eventually kills the net. > Because of the lack of coordinated control of the net, I doubt moderators > would work anyway because that assumes that all of the sites will set up > their software to work with moderators. We all know the success of getting > sites to do things like upgrade software and fix bugs, much less install > software to implement controvesial functions. In all honesty, I must say that it is the loose control and moderation over USENET that I prefer over the stricter controls of the ARPAnet. I prefer the freer exchange of ideas and the varied backgrounds of the participants. The reasons for this are that there is no centralized governing body of USENET people who could enforce rules and regulations for message content, etc. (where would we find time to do this, let alone money?). On the other hand, ARPA, being (at least up until the ARPA/MILNET split) being under control by the DoD, has a centralized governing body which can take action in situations which are deemed abusive to the ARPA population. However, since the split, the ARPA- net has become more like USENET in terms of message content. For example, we note that a lot of USENET groups make it into ARPA, and vice versa, so ARPA is getting exposed to our style of conversations, and I have noticed a general change in the style of ARPA messages over the past two years. Also, a number of high-flamage lists (which USENET doesn't get, I think) have sprung up on the ARPAnet, with subject matter similar to already existing USENET groups. (This is partly true because the ARPAnet contributors are USENET contributors also.) > If we assume that tighter controls on the net are a hopeless wish, the only > other alternative we have is information. We need to teach people to use > the net properly. How? Good question. We've written and posted guides, > we've mailed comments and suggestions to transgressors, we've posted > suggested, screams, yells, and whimpers, and things keep getting worse. I > certainly don't know what will work anymore, nothing that has been tried > seems to. Again, without a centralized governing body to enforce order, these things can- not be accomplished. I am reminded (regarding the difficulty of getting bug fixes and standardized software on all net machines) of what happened in January 1983 when the switchover from NCP to TCP took place. A message was sent out to all ARPA Internet sites warning them that on 2/1/83 NCP would no longer exist and TCP would be the standard communications protocol. Any site which did not run TCP would simply be unable to communcate with the ARPAnet anymore. Well, 2/1/83 rolled around and sure enough, a lot of sites mysteriously vanished (I remember quite a number of mailing lists were re- turning bad addresses for quite some time after that, so USENET is not the only net with internal problems). The point here is that since ARPA had a govern- ing body, it could enforce the rules required for subscribers to make required changes, or lose their ability to access the net. I have not been to a USENIX conference myself, but I thought that such issues might be resolved there. However, USENET is growing and sooner (or later) if we want to keep the net around we will have to have a legislative body, because there will be just *too much* policy to be administrated in a distributed manner. > Perhaps it is time to simply junk Usenet as a failed prototype and come up > with a better communication scheme. Perhaps things will miraculously fix > themselves. I doubt it, though. I've got lots of questions about things, > but no answers, and the future of Usenet as it stands now looks rather > bleak to me. Anyone else out there have some answers that will work? Like I said, there's always TCP/IP. :-) Thanks for your article, Chuq. It gave me an opportunity to voice some opin- ions and observations I have had about USENET and the ARPAnet for quite some time. I hope that others of you will voice your opinions as well. In par- ticular, I'd like to hear from some of you who have seen both sides of the fence like I have -- those of you who had access to the ARPAnet and USENET and had an opportunity to observe their respective cultures. -- Hug me till you drug me, honey! Greg Skinner (gregbo) {allegra,cbosgd,ihnp4}!houxm!gregbo
chip@t4test.UUCP (Chip Rosenthal) (09/03/84)
--- REFERENCED ARTICLE --------------------------------------------- >From: chip@t4test.UUCP (Chip Rosenthal) >Summary: comment on multiple newsgroups >Date: Fri, 31-Aug-84 09:06:22 PDT > >I would like to sensitize people to one fact here. It has to do with >sites which are running news under Eunice. Either these sites throw >away multiple newsgroup postings, or the articles are physically >duplicated into each of the newsgroups. -------------------------------------------------------------------- I received some mail questioning why not just fix news for Eunice. Unfortunately, I then proceeded to blast away the mail before I could answer. The question is certainly a fair one, so I will send my answer out over these here airwaves--with appologies. The problem is that multiple newsgroups entries are implemented with links. Eunice, does not support links. Although a link procedure is available, the link lasts only during the execution of that program. When the program terminates, the links disappear, and the file sits on the disk with the name used in the last linking operation attached to it. You can visualize the link as acting as a link while the program is running, and a rename upon process completion. (It's not a feature, it's a bug.) The 'inews' hack I have seen for Eunice is to perform a copy of the article. That is, rather establishing a link from 'net/foobar/123' to 'net/whoopee/432', you physically create a copy of the former, named as the latter. If this problem is to be fixed, there are two apparent solutions: either make links work, or devise a multiple newsgroup algorithm for Eunice. Obviously, the first is out of my hands. If TWG wants to implement links, wonderful! (However, I would like to see the 'vi' bug discussed in 'net.eunice' a while back fixed more than this.) The second one is more feasible. And I promise everybody that as soon as I find myself with several days with nothing more constructive to do, I will do it. (Translation: don't hold your breath.) So, I reitterate my original two points. First, think again before you post to seven newsgroups. Not only will it make the Eunice sites happier, but it will have the wonderful side effect of keeping news articles catagorized better. Secondly, software developers should realize that Eunice exists, and it don't work exactly the same as ATT & BSD stuff. (Kudos to Larry Wall, who has at least two Eunice sites trying out his 'rn' news reader program. The program works, and it is great.) -- Chip Rosenthal, Intel/Santa Clara { idi|intelca|icalqa|kremvax|qubix|ucscc } ! t4test ! { chip|news }
john@genrad.UUCP (John Nelson) (09/03/84)
Well, I know of no solution to Eunice keeping multiple copies of articles around (If you havn't got links, you haven't got them!). However, vnews has an article selection algorithm (easily incorporated into standard "readnews") which does not involve links or inodes or such, but which checks the newsgroups line against the groups posted to and those to which you subscribe. It will show you the article in the FIRST newsgroup to which it was posted to which you also subscribe. The solution is to FIX THE BROKEN SOFTWARE! Don't say "dont post to multiple newsgroups", because that is sometimes the only way to reach the desired people. What really burns me is the people who post to several newsgroups SEPERATELY rather than in a single posting to multiple groups.
scs@foxvax1.UUCP (S.C. Schwarm ) (09/05/84)
Eunice could support links like UNIX does, because one can have multiple directory entries for a single file (as long as all the directories are on the same device. In fact, there used to be a switch on PIP (/EN I think) which performed the operation. The only problem is that if the file is deleted by any of the name it is removed from the system. It is much harder if not imposible to implement the remove command for these links. Steve Schwarm
smb@ulysses.UUCP (Steven Bellovin) (09/05/84)
I hate to say it, but the ability to post to multiple newsgroups was a major design goal for the very first version of netnews. And it was implemented as links in that version, which antedated A-netnews. All of this happened long before there was a Eunice. So -- I'm sympathetic to the problems of people who are using non-UNIX operating systems to run netnews. But it's a bit chutzpahdic to ask people to forgo using a useful feature of netnews because a very few sites can't handle it. If anyone wants to "fix" netnews to run on Eunice, I'll be glad to discuss alternative implementations. I'm not going to do the work, though... --Steve Bellovin
dmmartindale@watcgl.UUCP (Dave Martindale) (09/15/84)
An example of a fairly simple way to let Eunice-based B news handle postings to multiple newsgroups: When an article comes in, store the body of it in the appropriate place for one of the newsgroups, and for each of the other groups it was posted to create a file containing a "magic token" followed by the pathname of the file containing the body of the article. Since the first byte of a stored news article is always a capital letter (because it is the first character of a header line of some sort), any special character would do as the "magic token". Readnews then need only look at the first byte of an article file to determine if it is a real article, or a file containing a pathname which should be read in and then opened to obtain the real article. This should involve only minor changes to inews and readnews. Implementing this should require only slightly more effort than went into the letters that we've seen so far explaining why stock B news doesn't work properly on Eunice and why we thus shouldn't post to multiple newsgroups. This scheme won't work right in situations where different newsgroups are expired in differing lengths of time, since you lose UNIX's "all links have equal status" property. But then, it only required about 30 seconds to think of - you should be able to do better with a bit more thought.
guy@rlgvax.UUCP (09/27/84)
> This scheme won't work right in situations where different newsgroups > are expired in differing lengths of time, since you lose UNIX's "all links > have equal status" property. But then, it only required about 30 seconds > to think of - you should be able to do better with a bit more thought. A suggestion, from a VMS non-wizard: Use links (i.e., multiple directory entries pointing to the same file), but replace the "unlink" routine by one that just removes the link and doesn't delete the file (or "mark the file for deletion" or whatever). Keep a reference count in the history file, and delete the file when the last link disappears. Will this work? Can a user-mode Eunice program do all this, or do you have to be running in supervisor or executive mode or not be built from Eunice to create/delete directory entries and mark files for deletion? Guy Harris {seismo,ihnp4,allegra}!rlgvax!guy
gordon@uw-june (Gordon Davisson) (10/06/84)
Guy Harris suggests a way around Eunice's problems with links: >Use links (i.e., multiple directory entries pointing to the same file), but >replace the "unlink" routine by one that just removes the link and doesn't >delete the file (or "mark the file for deletion" or whatever). Keep a >reference count in the history file, and delete the file when the last link >disappears. > >Will this work? Can a user-mode Eunice program do all this, or do you have to >be running in supervisor or executive mode or not be built from Eunice to >create/delete directory entries and mark files for deletion? I don't know much about Eunice, but if it lets you use the VMS system services (and I don't see how it could avoid doing this), you can do it with RMS calls: sys$enter and sys$remove. You can also manipulate links with the undocumented DCL commands "set file <oldname>/enter=<newname>" and "set file <name>/remove". -- Human: Gordon Davisson USnail: 5008 12th NE, Seattle, WA, 98105 UUCP: {ihnp4,decvax,tektronix}!uw-beaver!uw-june!gordon