dww@stl.stc.co.uk (David Wright) (08/05/89)
In article <1989Aug3.112451.21772@nc386.uucp> rhg@nc386.uucp (Rich Garrett) writes:
#This article generated by inews -C comp.lang.sigplan.
#--
#Richard H. Garrett rhg@ncoast.org
#uunet!hal.cwru.edu!ncoast!rhg ncoast!rhg@hal.cwru.edu
#NCoast Public Access UN*X - (216) 582-2460, 1200/2400 baud, login: makeuser
And there are more where that came from - they're just hitting our
news server as I type this. A likely consequence is that this and the
other affected news groups' moderators will get hundreds of copies of
articles posted as some parts of the net show the group moderated while
others don't.
I presume that C news is to blame (look at that great long C news style
Reference), rather than Richard himself deciding to disrupt the net.
It appears that doing what for B news is an innocent command to create a
news group locally will, with C news, send that newgroup control message to
the entire net. Of course there is a way in C news to specify a local
distribution, but it seems that people used to B news's more civilised
default control distribution of 'local' keep getting caught out by the
different behaviour.
WHY DOES C NEWS DO THIS?
PLEASE will the C news people issue an urgent patch or update to C news
(marked MANDATORY :-) ) to change its behaviour so that control articles
are by default sent to local distribution, as currently happens with B news.
I know that they believe C news doesn't have to be completely compatible
with B news, but in this respect IT REALLY DOES.
Regards, David Wright STL, London Road, Harlow, Essex CM17 9NA, UK
dww@stl.stc.co.uk <or> ...uunet!mcvax!ukc!stl!dww <or> PSI%234237100122::DWW
"In general, it is best to assume that the network is filled with malevolent
entities that will send in packets designed to have the worst possible effect"
From the draft "Requirements for Internet Hosts" RFC
henry@utzoo.uucp (Henry Spencer) (08/06/89)
In article <1631@stl.stc.co.uk> "David Wright" <dww@stl.stc.co.uk> writes: >It appears that doing what for B news is an innocent command to create a >news group locally will, with C news, send that newgroup control message to >the entire net... people used to B news's more civilised >default control distribution of 'local' keep getting caught out by the >different behaviour... >I know that they believe C news doesn't have to be completely compatible >with B news, but in this respect IT REALLY DOES. Unfortunately, B News disagrees with itself on this. 2.10 did indeed send to the entire net; the restriction to local posting came with 2.11. (No, 2.10 is not dead, there are quite a few of them around still.) B News does *not* have "more civilized default control distribution"; *some versions* of B News have it. There is *no* single behavior that is "the way B News did it", and C News *cannot* be completely compatible no matter what we do. The purpose of inews is to post news. It is irrational to have it do local system administration which does not involve posting, and indeed we have an "addgroup" command which does the job of purely-local creation. One function, one program, the way Unix does things (or did, back when its development was controlled by people with taste). Unfortunately, a lot of people can't read documentation, it seems. Geoff and I are considering what to do about this; our current feeling is that the right way out is simply to make "inews -C" print an error message (mentioning "addgroup") and terminate. This, or some variation if we think of a better approach, will be in the next patch. -- 1961-1969: 8 years of Apollo. | Henry Spencer at U of Toronto Zoology 1969-1989: 20 years of nothing.| uunet!attcan!utzoo!henry henry@zoo.toronto.edu
allbery@nc386.UUCP (Brandon S. Allbery) (08/11/89)
The recent batch of newgroups emanating from nc386 are the result of a program Rich runs every so often ("mvjunk") which apparently doesn't like C news's version of the active file. (Needless to say, he didn't warn me about this program until *after* the damage was done....) Yes, C news's propensity to post newgroup messages with a default distribution of "world" is a problem. A worse problem is that many programs expect to say "inews -C news.group moderated", but C news requires it to be "inews -C 'news.group moderated'" -- note the quotes, inews -C reads exactly one argument. (Yuck.) I hacked together an alternative that talks directly to relaynews and will alter inews if this isn't fixed in the next set of patches. I would, BTW, appreciate input from anyone who knows why "mvjunk" broke; it was non-obvious to me. I believe Rich got it off the net. I *did* go looking through our "control" newsgroup and issue new control messages reverting moderated groups to their original status. If I missed one or more groups, please let me know; although by now I daresay Greg or Spaf has dealt with any stragglers. ++Brandon (semi-official maintainer of news on nc386) -- Brandon S. Allbery, moderator of comp.sources.misc allbery@NCoast.ORG uunet!hal.cwru.edu!ncoast!allbery ncoast!allbery@hal.cwru.edu "Why do trans-atlantic transfers take so long?" "Electrons don't swim very fast." -john@minster.york.ac.uk and whh@PacBell.COM
henry@utzoo.uucp (Henry Spencer) (08/12/89)
In article <1989Aug10.231323.29146@nc386.uucp> allbery@nc386.UUCP (Brandon S. Allbery) writes: >... ("mvjunk") which apparently doesn't like C news's >version of the active file... This is a bit surprising, since the C News active file is the same as the B News one, with minor extensions. We too would be curious to know why it broke. >Yes, C news's propensity to post newgroup messages with a default distribution >of "world" is a problem. Fixed in next patch (Monday?). >A worse problem is that many programs expect to say >"inews -C news.group moderated" Uh, "many"? We aren't aware of any. (Undoubtedly there are a few, but we'd be curious to know about any programs that do "inews -C" -- they must post-date the initial 2.11 release, since 2.10 broadcast the creation message like C News.) > but C news requires it to be >"inews -C 'news.group moderated'" -- note the quotes, inews -C reads exactly >one argument. (Yuck.) Well, some people like to conform to things like syntax standards and some don't. We do. -- V7 /bin/mail source: 554 lines.| Henry Spencer at U of Toronto Zoology 1989 X.400 specs: 2200+ pages. | uunet!attcan!utzoo!henry henry@zoo.toronto.edu