mikel@teda.UUCP (Mikel Lechner) (02/15/91)
I get postings to completely bogus newsgroups regularly. Even ones
for which I've never even seen a "newgroup". I would expect that this
newsgroup ("misc.sheep") doesn't appear it the "sys" file at any
of these sites (see below). Why then are these sites passing on this message
since it contains NO valid newsgroup names?
Is this some new "feature" in C news (yes, I'm still running Bnews),
or is it common practive to put "all" in all your "sys" file entries.
Since my system does not list this newsgroup in its active file,
these messages end up in "junk" and go no further. It seems like a mistake
to pass on articles to bogus newsgroups.
This perpetuates problems like continued posting to "comp.sys.ibm.pc"
which was replaced months ago with "comp.sys.ibm.pc.misc".
>> Path: teda!vsi1!ames!decwrl!elroy.jpl.nasa.gov!sdd.hp.com!zaphod.mps.ohio-state.edu!julius.cs.uiuc.edu!psuvax1!psuvm!cunyvm!ndsuvm1!mtus5!kitbash
>> From: KITBASH@MTUS5.BITNET (Joe Ammond)
>> Newsgroups: misc.sheep
>> Subject: Re: Wyle E. Coyote
>> Message-ID: <91044.161408KITBASH@MTUS5.BITNET>
>> Date: 13 Feb 91 21:14:08 GMT
>>References: <IbLzou_00WBL03P5gj@andrew.cmu.edu> <obiAzWa00YU6Q9pI1Y@andrew.cmu.edu>
>> Organization: Computing Technology Services, Michigan Technological Univ.
>> Lines: 3
--
Mikel Lechner UUCP: teda!mikel
Teradyne EDA, Inc.
5155 Old Ironsides Drive
Santa Clara, Ca 95054tale@rpi.edu (David C Lawrence) (02/16/91)
In article <20752@teda.UUCP> mikel@teda.UUCP (Mikel Lechner) writes:
Is this some new "feature" in C news (yes, I'm still running Bnews),
or is it common practive to put "all" in all your "sys" file entries.
It is a feature of C News, one which I have mixed feelings about. It
has really nothing to do with putting "all" in the line though, at
least no more than putting "misc" in, for the case of misc.sheep which
you cite. Forwarding is determined based on the Newsgroups: of an
article, not where it is filed.
Since my system does not list this newsgroup in its active file,
these messages end up in "junk" and go no further. It seems like a
mistake to pass on articles to bogus newsgroups.
Yes and no. Henry and Geoff will maintain that it is a feature which
helps the propagation of valid new groups, and this is a reasonable point
which I appreciate. It also leads to greater traffic in groups carried by
very few people, but who manage to communicate (?) with each other through
the junk spools of intermediate sites, which is a questionable feature.
Additionally, it contributes to propagation of bogus groups by incipient
seepage; I believe that C News (as much as I love it) is the primary
reason that anyone other than Berkeley is carrying a newsgroup named
"comp.dcom.lans.novell". Again, however, some people would argue that
such is a feature. I can even side with _part_ of their argument, but not
with the way it leads to communication given current software.
This perpetuates problems like continued posting to "comp.sys.ibm.pc"
which was replaced months ago with "comp.sys.ibm.pc.misc".
I am inclined to agree. Another problem here is that even on C News
sites which refile such a group, it is still simple to post to the
group which is being refiled. I first became aware of this when one
of my users posted to comp.os.os2 despite the fact that I have been
refiling it to comp.os.os2.misc since the reorg. There are two
possible changes; a) rewrite the header in inews to substitute correct
names for the refiled groups or b) give an error as with the n or x
flag. The latter is much easier to implement.
This whole thing, and newsgroup naming itself, is something which will
should matter much, much less if the network news software developers have
their way with dissociating groups from their current implementation, and
foster a new concept of distribution. In this next generation of article
conferencing I would expect that very large sites like mine would still
have something like comp.dcom.lans.novell spontaneously generate, but it
would be less of a problem than with the way it happens now. Personally I
have yet to see a plan in which I have confidence, but I have a lot of
respect for the people who are working toward such things.
--
(setq mail '("tale@cs.rpi.edu" "tale@ai.mit.edu" "tale@rpitsmts.bitnet"))henry@zoo.toronto.edu (Henry Spencer) (02/20/91)
In article <M_-&W`^@rpi.edu> tale@rpi.edu (David C Lawrence) writes: >Yes and no. Henry and Geoff will maintain that it is a feature which >helps the propagation of valid new groups, and this is a reasonable point >which I appreciate... This one is clearly a two-edged sword, to be sure. However, to be realistic one *must* acknowledge that things like group-creation messages often do not get the complete propagation they should. The net is a much messier place than one would think on theoretical grounds, and robustness in the face of such things is important. >... There are two >possible changes; a) rewrite the header in inews to substitute correct >names for the refiled groups or b) give an error as with the n or x >flag. The latter is much easier to implement. There is a good chance that the latter will appear in the official C News distribution in the near future; forbidding posting to refiled groups seems like a reasonable idea. -- "Read the OSI protocol specifications? | Henry Spencer @ U of Toronto Zoology I can't even *lift* them!" | henry@zoo.toronto.edu utzoo!henry
mikel@teda.UUCP (Mikel Lechner) (02/21/91)
henry@zoo.toronto.edu (Henry Spencer) writes: [On C news allowing users to post to non-existent, but re-filed newsgroups.] >>... There are two >>possible changes; a) rewrite the header in inews to substitute correct >>names for the refiled groups or b) give an error as with the n or x >>flag. The latter is much easier to implement. >There is a good chance that the latter will appear in the official C News >distribution in the near future; forbidding posting to refiled groups seems >like a reasonable idea. Yes, just giving the poster an error message would eliminate 90% of the problem. Most sites keep up-to-date active files (at least for the mainstream newsgroups). The current silent refiling lets them keep making the mistake indefinately. Your proposed solution would solve the problem to my satisfaction. -- Mikel Lechner UUCP: teda!mikel Teradyne EDA, Inc. Phone: (408) 980-5200 5155 Old Ironsides Drive Santa Clara, Ca 95054
garvey@johnny5.uucp (Joe Garvey) (02/21/91)
In article <M_-&W`^@rpi.edu>, tale@rpi.edu (David C Lawrence) writes: > In article <20752@teda.UUCP> mikel@teda.UUCP (Mikel Lechner) writes: > > Since my system does not list this newsgroup in its active file, > these messages end up in "junk" and go no further. It seems like a > mistake to pass on articles to bogus newsgroups. > > Yes and no. Henry and Geoff will maintain that it is a feature which > helps the propagation of valid new groups, and this is a reasonable point > which I appreciate. It also leads to greater traffic in groups carried by > very few people, but who manage to communicate (?) with each other through > the junk spools of intermediate sites, which is a questionable feature. > This perpetuates problems like continued posting to "comp.sys.ibm.pc" > which was replaced months ago with "comp.sys.ibm.pc.misc". Since almost all mail passes thru the major sites, why not have the sites filter "junk groups". In those cases where the newsgroup header is wrong (comp.sys.ibm.pc), the header could simply be modified, and that article propigated correctly. I've recently examined the local flow of articles into junk... we're not talking about an overwhelming number. When enough major sites started to do this, many of these articles would be damped out before getting too far. This of course assumes these sites keep up with changes to the newsgroup hierarchy. :-) In the cases of "dumb" (my point of view) news groups like alt.stupid.putz, I don't know what the net can do. Obviously in alt, nothing... it's the way alt works. In the case of "psuedo real" groups like comp...novell, I think you have to let them propigate... as long as there's interest. I'm having exactly that problem now. Johnny5 is a relatively new site. My neighbors (on the net) have been less than meticulous (sp?) keeping their active files current. I created my active file from theirs. Now, I find some junk groups... that clearly aren't junk. Posting queries on what are valid groups draws marginally useful responses. My solution has been to examine the articles... if they look good, then I do a local new group. This only aggrevates the problem, because anyone who gets a feed from me will use my active file as a starting point... further propigating possibly bogus (groups not having gone through the normal creation proceedure) news groups. But what else can I do? No one I know reads junk. But if there are useful postings, I want them seen. The only other method would seem to be a "newgroup" moderator... which I actually wouldn't mind... but I can see upsetting many people. [BTW, I heard Gene Spafford has stopped posting his list of newsgroups, is this true?] -- Joe Garvey uucp: sumax!ole!johnny5!garvey J5 Research map entries are wrong for johnny5. They're Bothell, Wa. being fixed. Please use address above.
henry@zoo.toronto.edu (Henry Spencer) (02/22/91)
In article <1991Feb21.051023.8661@johnny5.uucp> garvey@johnny5.uucp (Joe Garvey) writes: >Since almost all mail passes thru the major sites, why not have the sites >filter "junk groups". In those cases where the newsgroup header is wrong >(comp.sys.ibm.pc), the header could simply be modified, and that article >propigated correctly... Three problems: (1) The flow of news is not nearly so simple as you seem to think; articles can get a long way out into the net without ever passing through a "major site". (2) Header rewriting is a sin, because it's very difficult to be *sure* that you are doing it right. (3) Automatic maintenance of the rewrite-control file is full of security problems, and manual maintenance is sufficient hassle that few people want to do it. -- "Read the OSI protocol specifications? | Henry Spencer @ U of Toronto Zoology I can't even *lift* them!" | henry@zoo.toronto.edu utzoo!henry
tale@rpi.edu (David C Lawrence) (02/25/91)
In <1991Feb21.051023.8661@johnny5.uucp> garvey@johnny5.uucp (Joe Garvey):
I've recently examined the local flow of articles into junk...
we're not talking about an overwhelming number.
My personal feeling is that an article ending up in junk means
something is wrong, either at my site or elsewhere. While I would not
call the numbers "overwhelming", any is too many. In the case of most
alternative hierarchies (except alt.*) and USENET, I take the time to
send a message to the person posting the article and to his news admin
if I think the problem was at their end, to help the person better use
the network for the communication that was being attempted. It
generally goes like this:
The group you posted to does not exist in the net at large; it should
either be removed at your site or the posting software should be fixed to
not allow posting to non-existent groups. Try posting to [[whatever group
looked appropriate for the topic of the article]].
In the cases of "dumb" (my point of view) news groups like alt.stupid.putz,
I don't know what the net can do. Obviously in alt, nothing... it's the way
alt works.
The "way alt works" is the most enigmatic of all hierarchies which get
very widespread distribution.
In the case of "psuedo real" groups like comp...novell, I think
you have to let them propigate... as long as there's interest.
I disagree. The propagation of groups like this is not a service; as long
as people in only very few sites think that it exists, their attempts to
communicate down that channel could very easily be frustrating. In the
case of this particular group, to people who actually had it created at
their site then it looked to them to be as valid as comp.dcom.lans.cisco,
though probably getting about 1/100th the propagation (in terms of sites
carrying a group by that name, not in terms of sites carrying the
articles). The net at large had no idea, also, whether the gateway was
bidirectional such that any new articles propagating back into Berkeley
for that group would have indeed been gatewayed back to the target
audience which was creating all of the traffic to begin with. This all
happened after there was already a novell group created in USENET, too.
My neighbors (on the net) have been less than meticulous (sp?) keeping
their active files current. I created my active file from theirs. Now,
I find some junk groups... that clearly aren't junk. Posting queries on
what are valid groups draws marginally useful responses. My solution
has been to examine the articles... if they look good, then I do a
local new group. This only aggrevates the problem, because anyone who
gets a feed from me will use my active file as a starting point...
further propigating possibly bogus (groups not having gone through the
normal creation proceedure) news groups. But what else can I do?
I read junk, for the very reason of tracking down these problems. You can
check out the news.lists group lists and then follow the groups which are
responsible for starting out new groups in hierarchies you get. The
traffic is normally light on them, news.groups being an exception (but you
can certainly make do just by following news.announce.newgroups) and
alt.config following behind but still not nearly as high on traffic.
Watch for newgroup message in control, and ignore those that look forged
--- just kill all of the "cancel" traffic in control and reading the rest
makes it a simple group to follow.
[BTW, I heard Gene Spafford has stopped posting his list of newsgroups,
is this true?]
No. They are probably also the best source of information of what is
valid and what is not; you can build your own checkgroups out of them to
not have such a haphazard active file.
--
(setq mail '("tale@rpi.edu" "uupsi!rpi!tale" "tale@rpitsmts.bitnet"))