swindon.ingr.com (Nik Simpson x4333) (03/19/91)
I'nm going for stupid question of the week award, before anybody says RTFM, I have and this is how I understood it, consider a site that doesn't wish to be fed the soc hierarchy, I thought a sys entry at the feed site like :- downstream:all,!soc:F:/usr/spool/news/batch/downstream should feed to the site "downstream" all articles except those in the soc hierarchy. Obviously articles cross-posted to soc will still get through if you receive any other matching groups. However if the newgroups line in the article only contains soc groups it shouldn't be sent to the site "downstream". The question is, have I misunderstood something, or should I be looking for a problem in my CNEWS installation. -- |--------------------------------------------------| | Nik Simpson Mail : nik@swindon.ingr.com | | Systems Consultant (UNIX). Intergraph UK Ltd. | |--------------------------------------------------|
henry@zoo.toronto.edu (Henry Spencer) (03/20/91)
In article <1991Mar18.170511.6535@st_nik!swindon.ingr.com> nik@swindon.ingr.com writes: >... I thought a sys entry at the feed site like :- > > downstream:all,!soc:F:/usr/spool/news/batch/downstream > > should feed to the site "downstream" all articles except those in the >soc hierarchy. Obviously articles cross-posted to soc will still get through >if you receive any other matching groups... I'd be inclined to write that as "all,!soc/all" to avoid potential troubles with distributions, but yes, that ought to work, apart from the cross-posting issue you mention. If it's not working I'd suspect problems elsewhere. -- "[Some people] positively *wish* to | Henry Spencer @ U of Toronto Zoology believe ill of the modern world."-R.Peto| henry@zoo.toronto.edu utzoo!henry
jc@skyking.UUCP (J.C. Webber III) (03/23/91)
>In article <1991Mar18.170511.6535@st_nik!swindon.ingr.com> nik@swindon.ingr.com writes: >I'd be inclined to write that as "all,!soc/all" to avoid potential troubles >with distributions, but yes, that ought to work, apart from the cross-posting >issue you mention. If it's not working I'd suspect problems elsewhere. Is this true for Bnews also? I am running Bnews and have the following entry in my sys file: ME:world,na,usa,all,!alt.sex.pictures.all,!sci.all,!soc.culture.all,!comp.binaries.all,!comp.windows.all,!comp.parallel.all,!comp.realtime.all,!comp.graphics.all,to.skyking:: Should I change my .all's to /all's? thx... -- J.C. Webber III jc@skyking.UUCP R&D Lab Manager ..!uunet!cadence!mips!skyking!jc Mips Computer Systems {ames,decwrl,pyramid,prls)!mips!skyking!jc (408)524-8260 CounterPoint Newsgroup: pointers@skyking -- J.C. Webber III jc@skyking.UUCP R&D Lab Manager ..!uunet!cadence!mips!skyking!jc Mips Computer Systems {ames,decwrl,pyramid,prls)!mips!skyking!jc (408)524-8260 CounterPoint Newsgroup: pointers@skyking
brad@looking.on.ca (Brad Templeton) (03/24/91)
One thing for sure, the "/all" convention that C news allows, while convenient, has destroyed the usefulness of the distribution header. Used to be that you could use the distribution header to restrict the distribution of an article posted to a wide-ranging group, ie. I could do a news.software.b posting just in Ontario. No longer. In recent times I have posted some rec.humor.funny messages with distributions like "uw" (local school) and "ont" and gotten replies from all over the globe. Yes, "all" is convenient, if you are feeding a true sub-leaf that is in every single domain you are in. But the convenience has caused sys files to be sloppily written everywhere. I strongly suggest that C news remove "all" as a valid distribution field. Force the people with the sloppy files to fix them if they upgrade. It will never become perfect again, due to those who don't upgrade, but it's a start. Better still, adopt the NewsClip convention of a hierarchy of distributions, each one with a numeric value. This allows people to specify all distributions greater or lesser than a specified one. Thus to another site in Ontario, you might feed "ont and above" which would include "can", "world" and the fake-distributions of "comp", "rec" etc. which shouldn't exist but some twizots use. You would have to hand code any unusual distributions, such as ones that don't belong in a hierarchy, or which you get but don't belong to. (ie. Buffalo people get "ont", we get "wny" (western new york). -- Brad Templeton, ClariNet Communications Corp. -- Waterloo, Ontario 519/884-7473
tale@rpi.edu (David C Lawrence) (03/24/91)
In <1991Mar23.193554.25944@looking.on.ca> brad@looking.on.ca (Brad Templeton):
Yes, "all" is convenient, if you are feeding a true sub-leaf that is in
every single domain you are in. But the convenience has caused sys files
to be sloppily written everywhere.
Au contraire, mon frere! (Whatever. My second language is German, not
French.) I do not use "all" simply because it is convenient, and I am
offended at the implication that I am a sloppy news administrator. I am
proud to be what I believe is a better than average news administrator,
and I'll thank you not to second guess why my sys file is configured the
way it is.
I strongly suggest that C news remove "all" as a valid distribution
field. Force the people with the sloppy files to fix them if they
upgrade. It will never become perfect again, due to those who don't
upgrade, but it's a start.
I then will personally hack it right back into the locally hacked
relaynews I run.
Why do I use all,!local,!rpi? Because Distribution: in every current
news system I have seen is broken. The concept is nice, but the
implementation is very weak. It is a philosophical problem with news.
I have seen many articles come through news with Distribution:s like
"everyone", "all", "net", &c. Distributions which clearly were intended
to propagate the article widely but which the policy you suggest would
prevent from happening. On the other hand, I see roughly the same
percentage, perhaps even a little less, of articles in mainstream
hierarchies which really should have been limited to the geographical
distribution which was intended.
Sometimes "the right thing" is quite tricky to see. As an example, what
about when someone announces a singles event in soc.singles with a
Distribution: of ba? I currently get ba here in New York because we
have displaced Californians and others who appreciate the hierarchy.
Asking us to come to a new singles volleyball event in San Jose on
Saturday is pretty dumb; the article probably shouldn't appear here for
logistical reasons. The same is not true of most of the traffic in most
of the regionals I have read, which is why it is not uncommon for a
regional to leave its physical vicinity. The same could also be said of
a great many (attempted) limited distribution articles in major
hierarchies.
I prefer in these cases, and the cases of bogus Distributions like
mentioned above, for the article to be propagated more widely than
intended rather than to have it silently fail to get to the audience
which was expected to receive it. I know of several good admins who
share this opinion. My "sloppy sys file" comes from a very sloppy
attempt at general network news distribution control in widely spread
hierarchies. It has been considered and reconsidered many times by me
and I do not as yet have a compelling reason to change how I forward
news.
Often we as administrators can see to what location the article should
reasonably go, but it is a grey area. Now how do we code it, and how do
we instruct the users about the best way to target the distribution?
What? They shouldn't have to worry about some complex system, and the
simpleness of Distribution: should be fine? Then how come so many of
them got it wrong before C News even showed up widely? C News is not
the major illness here.
... the fake-distributions of "comp", "rec" etc. which shouldn't
exist but some twizots use.
They exist according to RFC 1036. Supplement the RFC if this is truly a
wrong thing. I would agree that it is, but the Distribution: vs
hierarchy scheme has traditionally been very closely related. "Good luck."
--
(setq mail '("tale@rpi.edu" "uupsi!rpi!tale" "tale@rpitsmts.bitnet"))brad@looking.on.ca (Brad Templeton) (03/24/91)
Ok, I admit that sloppy was a bit of a nasty word to use, and in checking myself, I see I have used all for internal gateway lines in the sys file and "all,!this,!that" for a few of my single group feeds. So lazy is the right word. As far as I am concerned I see no reason at all to propagate the articles that say, "Distribution: everyone" or some other such mistake. Perhaps the local inews program should check for this and warn the user, but to have any typo or mistake go to the entire world is not the way to do it in my opinion. Long ago I pushed that distribution and group name should be disjoint. This was part of the philosophy behind the great renaming, and the introduction of the Distribution header field in the first place. But we went half-way, because we still have these hierarcies that match distributions. There is a "ba" (bay area) hierararchy, as in ba.this, and a "ba" distribution. We should have split these. C news took a partially right step by splitting the newsgroup and distribution parts of the SYS file, but it caused further harm by letting people conveniently slip "all" in the latter field. Under B news, you almost never put "all" in the combined newsgroups&distributions field, unless you wanted a complete and utterly full feed. There is software that handles distributions completely distjointly (except, sadly for control messages) and that is dynafeed. In dynafeed you list the groups you want explicitly, or with regular expressions, and you list the distributions you want explicitly. (No regular expressions, although I could do this if sites kept distributions they carry in a standard place and format) I had to kludge it a bit. If you get the distribution "can" then a distribution of "can.general" on an article is accepted. And I had to kludge it for control messages, which in ordinary news are propagated because the name matches the name hierarchies listed. (It's difficult in B and C, I believe, to get an individual selection of comp groups and still get control messages for new comp groups.) There's no easy fix. I propose: a) Completely disassociate distributions and hierarchies. Rename one or the other -- probably easiest to rename the distributions. b) Define some new rules for the propagation of newgroup messages. It should be easy to say, "give me newgroups for groups that match the following patterns without feeding me all the groups that match that pattern. c) Define a standard file like the active file that contains the distributions and information about them, to be used by posting programs and others. This file should list distributions in a hierarchical fashion. Newsclip uses a file called distlist which does this roughly. It assigns a numeric value to each distribution that is an approximation of the number of people who read it, thus ordering them to some degree. In retrospect, this is not perfect, because it doesn't tell us how to combine conflicting hierarchies that a site exists within. UUNET would be a royal mess since it lives in every distribution it can get its hands on. -- Brad Templeton, ClariNet Communications Corp. -- Waterloo, Ontario 519/884-7473
henry@zoo.toronto.edu (Henry Spencer) (03/24/91)
In article <449@skyking.UUCP> jc@skyking.UUCP (J.C. Webber III) writes: >Is this true for Bnews also? I am running Bnews ... >Should I change my .all's to /all's? I *think* this feature is a C News exclusive :-) and isn't available under B News. -- "[Some people] positively *wish* to | Henry Spencer @ U of Toronto Zoology believe ill of the modern world."-R.Peto| henry@zoo.toronto.edu utzoo!henry
peter@taronga.hackercorp.com (Peter da Silva) (03/25/91)
brad@looking.on.ca (Brad Templeton) writes: > I strongly suggest that C news remove "all" as a valid distribution field. The only way to make that work would be to remove the C news separate distribution altogether, because of all the people who put "alt", or "comp", or even "rec.arts.sf-lovers" in their distribution lines. Or you could have it toss illegal distributions (ones that match a newsgroup name except for regionally named newsfroups). -- (peter@taronga.uucp.ferranti.com) `-_-' 'U`
igb@fulcrum.bt.co.uk (Ian G Batten) (03/26/91)
In article <AD}=J-%@rpi.edu> tale@rpi.edu (David C Lawrence) writes: > wrong thing. I would agree that it is, but the Distribution: vs > hierarchy scheme has traditionally been very closely related. "Good luck." Indeed. Part of the problem is the local distributions have usually had the same name as their associated set of groups. ian
chip@tct.uucp (Chip Salzenberg) (03/27/91)
According to tale@rpi.edu (David C Lawrence): >I have seen many articles come through news with Distribution:s like >"everyone", "all", "net", &c. Distributions which clearly were intended >to propagate the article widely but which the policy you suggest would >prevent from happening. So what's next, automatic correction of misspelled newsgroups? If the poster can't be bothered to learn the correct distribution keywords, then he deserves whatever limited propagation he gets. My sys file entry for my feed ("tscs") reflects this fact: tscs:fl,tba,to.tscs/world,na,usa,fl,tba:f: tscs:all,!fl,!tba,!to/world,na,usa,fl,tba:L1f: On the other hand, "/all" is useful in some contexts: ME:all/all So I like the feature, I simply disapprove of its wide use in sys file lines that describe outgoing feeds. -- Chip Salzenberg at Teltronics/TCT <chip@tct.uucp>, <uunet!pdn!tct!chip> "All this is conjecture of course, since I *only* post in the nude. Nothing comes between me and my t.b. Nothing." -- Bill Coderre
tale@rpi.edu (David C Lawrence) (03/27/91)
From: chip@tct.uucp (Chip Salzenberg)
Date: 26 Mar 91 17:00:22 GMT
So what's next, automatic correction of misspelled newsgroups?
In obvious cases, sure. I wish my inews rewrote =groups. I live with
it just rejecting the article, forcing the user to correct it. Either
is acceptable enough for me. Why do you insinuate that automatically
correcting something like a misspelling is such an unspeakable evil?
(And no, I'm not talking about doing it to incoming news --- just
articles being newly posted.)
If the poster can't be bothered to learn the correct distribution
keywords, then he deserves whatever limited propagation he gets.
I disagree strenuously. Until I take the time to hack parsing and
rejection of bogus Distribution:s into inews then I will liberally err
on the favourable side of communication rather than silently fail to
inform him of problems.
On the other hand, "/all" is useful in some contexts:
ME:all/all
Wrong. The distribution subfield is not relevant in ME.
--
(setq mail '("tale@rpi.edu" "uupsi!rpi!tale" "tale@rpitsmts.bitnet"))brad@looking.on.ca (Brad Templeton) (03/27/91)
Somebody else mailed me something that suggested the right solution to this. A) Scrap "all" from the distribution field, except perhaps the "ME" field. B) Maintain the sys file in a macro language (M4, cpp, sed scripts, whatever) and edit the source. In your source define distribution macros like, "thistown", "thiscompany" etc. Use those macros in your distribution list, and even in newsgroup lists. -------------- At least if you still want to feed with sys files. -- Brad Templeton, ClariNet Communications Corp. -- Waterloo, Ontario 519/884-7473
woods@eci386.uucp (Greg A. Woods) (03/28/91)
In article <1991Mar23.220933.27563@looking.on.ca> brad@looking.on.ca (Brad Templeton) writes: > As far as I am concerned I see no reason at all to propagate the > articles that say, "Distribution: everyone" or some other such mistake. > Perhaps the local inews program should check for this and warn the user, > but to have any typo or mistake go to the entire world is not the way > to do it in my opinion. Agreed. The primary problem with Distribution: is a lot of broken news posting programmes. > But we went half-way, because we still have these hierarcies that > match distributions. There is a "ba" (bay area) hierararchy, as in > ba.this, and a "ba" distribution. Yeah, so what's wrong with that? > a) Completely disassociate distributions and hierarchies. Rename one or > the other -- probably easiest to rename the distributions. Why? Why can't the name-space be disassociated, thus allowing distributions of 'can' and hierarchies of 'can.all'? Distributions should be geographical or organisational, and heierarchies should be content related. > c) Define a standard file like the active file that contains the > distributions and information about them, to be used by posting programs > and others. This file should list distributions in a hierarchical fashion. Some parts of current news systems have such a definition. In rn's Pnews this seems to live entirely in the script, though there is internal mention to a /usr/lib/news(/rn)/distributions file, which I have occasionally seen used. If I remember, it looked just like a newsgroups file with 2 fields -- name<tab>description. I would like to see not a "standard file", but rather a standard set of distributions which could be locally supplemented in ways similar to newsgroups, with perhaps a checkdistrib message to update and verify one's distributions descriptions. Then, we need a change to the RFC to make Distribution: a required header, and new versions of news posting programmes that validate this header from the standard (plus local) definitions, perhaps providing a nifty menu interface for helping the user choose an allowable selection (which would also be used for newsgroups selection!). I remain unconvinced that we really need heirarchical distributions. -- Greg A. Woods woods@{eci386,gate,robohack,ontmoh,tmsoft}.UUCP ECI and UniForum Canada +1-416-443-1734 [h] +1-416-595-5425 [w] VE3TCP Toronto, Ontario CANADA Political speech and writing are largely the defense of the indefensible-ORWELL
chip@tct.uucp (Chip Salzenberg) (03/29/91)
According to tale@rpi.edu (David C Lawrence): > From: chip@tct.uucp (Chip Salzenberg) > > So what's next, automatic correction of misspelled newsgroups? > >In obvious cases, sure. ... Why do you insinuate that automatically >correcting something like a misspelling is such an unspeakable evil? See the Jargon file under hairy: `Do What I Mean is incredibly hairy.' I assert that it can't really be done right, so it shouldn't be done at all. >I disagree strenuously. Until I take the time to hack parsing and >rejection of bogus Distribution:s into inews then I will liberally err >on the favourable side of communication rather than silently fail to >inform him of problems. Hacking inews to generate bogus-distribution warnings is, IMHO, a Good Thing. I encourage you to post such hacks; I'd use them. On the other hand, eviscerating your sys file distribution fields to `compensate' for user mistakes is a Bad Thing. It compromises the behavior of the news transport in a misguided attempt to cover over a problem that isn't its fault. >The distribution subfield is not relevant in ME. (Learn something new every day...) Why is it not used in ME? -- Chip Salzenberg at Teltronics/TCT <chip@tct.uucp>, <uunet!pdn!tct!chip> "All this is conjecture of course, since I *only* post in the nude. Nothing comes between me and my t.b. Nothing." -- Bill Coderre
brad@looking.on.ca (Brad Templeton) (03/29/91)
There is no need to force the bay area hierarchy and the bay area distribution to have a different name. I merely suggest we do it to force home the concept that the two are different. Giving distributions and hierarchies the same name has caused a lot of trouble. Many people outside an area are interested in its distribution. Alumni of Stanford read Stanford groups at MIT. However, they are not likely to be interested in "used car for sale" postings limited to the Stanford geographical area. This is all an unfortunate byproduct of B news, made worse when C news properly split them out but didn't go all the way in splitting them out. In many areas of the net, the use of a Distribution header to limit distribution of an article in a netwide group simply doesn't work properly at all any more. Am I the only one who considers this a serious trend that should be averted now? In NewsClip, I defined synonyms for the distributions, so that you could refer to the "site", "company", "city", "state/province", "country" etc. distributions and a local mapping would put these on to the right thing. This may make sense in other places, too. -- Brad Templeton, ClariNet Communications Corp. -- Waterloo, Ontario 519/884-7473
schmitz@scd.hp.com (John Schmitz) (03/29/91)
In article <1991Mar28.194737.5337@looking.on.ca> brad@looking.on.ca (Brad Templeton) writes: > In many areas of the net, the use of a Distribution header to limit > distribution of an article in a netwide group simply doesn't work properly > at all any more. Am I the only one who considers this a serious trend that > should be averted now? I agree this is a serious problem and I know that some news systems completely ignore distributions (no flames about that here, please). But if distributions were only geographic (or bounded) in nature, it would be easy to stop the propogation of unwanted distributions no matter how the articles arrived. Now, however, it is impossible to tell what is and isn't a valid distribution, so you are forced to pass everything. Just my two cents, not hp's. -- John Schmitz schmitz@scd.hp.com HP Santa Clara Division hplabs!hpscdc!schmitz For mx dumb mailers: schmitz%scd.hp.com@relay.hp.com
fitz@wang.com (Tom Fitzgerald) (03/29/91)
brad@looking.on.ca (Brad Templeton) writes: > But we went half-way, because we still have these hierarcies that > match distributions. There is a "ba" (bay area) hierararchy, as in > ba.this, and a "ba" distribution. What's the point of making the distributions and hierarchies have different names, if you're still going to have regional hierarchies? At least C-news separated the group list from the distribution list, so it's much easier to get things in the "ne" distribution without getting ne.* newsgroups. Since that's possible, I really don't see what you can do by renaming one of them that you can't already do. > Newsclip uses a file called distlist which does this roughly. It assigns > a numeric value to each distribution that is an approximation of the > number of people who read it, thus ordering them to some degree. In > retrospect, this is not perfect, because it doesn't tell us how to > combine conflicting hierarchies that a site exists within. UUNET would > be a royal mess since it lives in every distribution it can get its > hands on. I think there are more sites that take overlapping distributions than there are sites that live in a pure hierarchy. We get "dc.*" and "sco.*" here, and used to get more of the left-coast groups. A guy in Michigan recently posted something to ne.jobs. There's really nothing wrong with this. Any proposed change that requires kludging non-hierarchical distributions is seriously flawed. It might be real good to enforce a little more discipline on posting, for sure. Nn by default uses a Distribution: from the top-level hierarchy of the group being posted to, and that's a plain mistake. Inews could certainly complain about bogus distributions, at least malformed ones. But personally, I'm going to keep "all,!whatever" in the sys file for the sites I feed. It's just too convenient. Right now, of all the news on this system, 32% of the articles with Distribution: headers are incorrect. They're malformed, or the name of a news hierarchy or plain newsgroup, or something like that. I don't want to block a third of the news from passing through here, just to enforce my opinion of an ideal world. When enough posting agents get changed to bring the percent of bad distributions down to 1% or 2% then it might be worth changing the relaying code too, but I wouldn't be in any hurry. --- Tom Fitzgerald Wang Labs fitz@wang.com 1-508-967-5278 Lowell MA, USA ...!uunet!wang!fitz