gnu@hoptoad.uucp (John Gilmore) (07/27/87)
I was checking the age of some of the news on my disk and noticed that
last month's OtherRealms (rec.mag.otherrealms) had a 1-month expiry
date, and that this month's 111K expires in November. I think this is an
abuse of Expires: and am not interested in keeping Chuq's "magazine"
on my "coffee table" for longer than the default expiration period.
Chuq sees no reason to fix it; this is our interaction:
Date: Wed, 22 Jul 87 16:36:34 PDT
From: gnu (John Gilmore)
To: sun!plaid!chuq
Subject: Expiration dates on OtherRealms
I was checking why /usr/spool was filling up and found that I have
back copies of OtherRealms from last month (due to an expiration date
late this month) and copies from this month which will not expire
until November!
Why should OtherRealms not expire like the rest of the netnews -- by
local option, based on how much disk space they have? Everything else
sticks around hoptoad for 9 days, but OtherRealms should be here for months?
I don't like to override expiration dates, because that blows away
things like the maps, news lists, and conference announcements that
have a legitimate reason to stick around. But this requires that
people don't abuse 'Expires:', which is why I'm asking you to fix it
rather than just blasting all old articles no matter what their expiration
date.
The first set of OtherRealms will expire in a day or two, but I think
you should send out a cancel on the other ones, since they have been out
there for 12 days now, which is roughly the right interval.
John
Date: Sun, 26 Jul 87 19:54:56 PDT
From: sun!chuq (Chuq Von Rospach)
To: hoptoad!gnu
Subject: Re: Expiration dates on OtherRealms
It's done for a couple of reasons:
1) because I'm constantly getting requests for back issues (or replacement
copies) because OtherRealms comes out much less often than the
default expiration, so the newsgroup is often empty. This implies
a lot of (needless) large e-mail files.
2) because the way Brian's demographics run, it is impossible to get a
reasonable number for OR's readers any other way, since it is likely
that the issue has expired before the arbitron is read.
3) because OtherRealms is a very low volume group when you think about it,
and can't be causing major disk problems. On a machine with little
readership turnover (like hoptoad) it probably isn't necessary, but
I really need some way of keeping OtherRealms around between issues to
make it available, and the Expires: is the only way -- unless I want
to republish it every few weeks, which I think is a real waste.
It doesn't take up much time (I'm surprised the previous issue hadn't
expired -- it should have from the date I put on it). Um, make that space.
If you want to override it, be my guest -- I don't believe that it can
really cause serious disk problems, and becuse of the ways that OR is
different from the rest of the net, it is justified.
chuq
---------
Gnu here again. I'd like to rebut Chuq's reasons here.
1) I'd like to put a long expiration date on everything I post so that
nobody would bother me about getting copies of it, too. In fact, let's
expire comp.sources in 1995 or so. As you can see, if everyone played
this game there would be no disk left. Wouldn't it be better to tell
folks to save away the issues when they arrive, if they really care enough
to email around the net looking for copies after they expire? Sounds
like education is in order, not long expiry dates.
2) This is an odd one! The claim is that because Brian Reid's readership
survey exists, Chuq's postings should stick around to be sure they are counted.
Why not time the release of the magazine to coincide with the surveys,
like the TV networks and such do? Or just ignore the survey?
3) I don't understand this one. Because OtherRealms is such a low
volume group, people don't read it when it comes out? Can't its
readers fight their way through all 111K before it expires? It seems
that a better solution would be to post it in pieces, say one per week,
so that there would usually be something there but the whole thing wouldn't
stick around for months.
I'd take the issue to Stupid Peoples' Court but it seems to be out
of session. So I'll take it to the news administrators instead.
If we assume that, like mod.mag.otherrealms, rec.mag.otherrealms
reaches 90% of the now ~6800 Usenet sites, that's ~6100 sites.
The current issue is therefore tying up ~677,100,000 bytes of disk around
the world and will continue to do so until November.
While burning disk space all over the world is a "solution" to Chuq's
problems, it bothers me. So far nobody else has refused to remove
Expires: lines or failed to cancel "never expires" type postings, once
I explained what Expires: was for, and how individual sites manage
their disk space. (Every once in a while I read the front of my
history file to see what old articles still linger there, and email the
folks who did it, educating them about how untutored use of Expires: is
a Bad Thing.) Chuq seems to see his group as more important than the
rules for Expires: lines. Do the rest of the news admins agree?
--
{dasys1,ncoast,well,sun,ihnp4}!hoptoad!gnu gnu@postgres.berkeley.edu
Alt.all: the alternative radio of the Usenet.chuq%plaid@Sun.COM (Chuq Von Rospach) (07/27/87)
>Gnu here again. I'd like to rebut Chuq's reasons here. And I'll re-rebut. Glad I was warned this was going public... >1) I'd like to put a long expiration date on everything I post so that >nobody would bother me about getting copies of it, too. In fact, let's >expire comp.sources in 1995 or so. As you can see, if everyone played >this game there would be no disk left. The question is, John, how often you get requests for copies of your articles? Before I upped the Expiration date, I was averaging four or five requests a week (just to UUCP -- I'm ignoring the ARPA and non-phone-bill sites here). That's 100K per copy, or about 2 megabytes a month flowing out to handle speciail requests. Since I implemented the Expires, I've had two requests, over two months and two issues. A significant drop in mail volume, which I think justifies the Expires. The disk is cheap, and any sysadmin dying for disk space can override it as they wish. >2) This is an odd one! The claim is that because Brian Reid's readership >survey exists, Chuq's postings should stick around to be sure they are counted. >Why not time the release of the magazine to coincide with the surveys, >like the TV networks and such do? Or just ignore the survey? Not odd at all. I can't time the release of OtherRealms to the surveys, because the surveys happen whenever a given site feels like doing them -- and it also depends on the vagaries of transmission delays, holding times, etc. such that there is no way for something that is distributed to the net can hit the survey reliably. I could possibly ignore it, but the size and demographics of OtherRealms happens to be of strong interest to me. >3) I don't understand this one. Because OtherRealms is such a low >volume group, people don't read it when it comes out? Can't its >readers fight their way through all 111K before it expires? You misunderstand completely -- it isn't the folks who are there when it comes out, it is the folks who come on-line in between issues that I'm trying to serve. They get told about OtherRealms by a co-worked, the group is empty, so the write and ask me for it. By giving it a long expiration, the new users can get it on the local system without having to have it mailed to them. As I said in my mail to you, on a site without a lot of users being added, the Expires won't really help. But on the other sites, it HAS helped, because OtherRealms is there when a new user goes looking for it, so I don't have to mail a fresh copy out (saving modem costs along the way...) >It seems >that a better solution would be to post it in pieces, say one per week, >so that there would usually be something there but the whole thing wouldn't >stick around for months. Not possible, because of what OtherRealms is. That may be a better solution to you, but you don't seem to understand the philosophy behind OtherRealms. >I'd take the issue to Stupid Peoples' Court but it seems to be out >of session. So I'll take it to the news administrators instead. >While burning disk space all over the world is a "solution" to Chuq's >problems, it bothers me. I don't seen any reason in what John says to remove the Expires. disk is cheap, and by using it I've significantly cut the special mailings I've had to make, significantly cutting the overall cost of OtherRealms on the net. I consider this a Good Thing. Any system administrator that disagrees with me is free to override me on their local system, which is perfectly fine with me as well. But I don't see anything in Johns arguments that convince me that what I'm doing is wrong from a new-wide point of view. If anyone wants to try to convince me otherwise, they're welcome to. If the size, distribution, etc of OtherRealms has gotten to the point where it is really a hassle to the net, then perhaps it is time to move OtherRealms off of USENET and make it strictly a printed publication. I consider the electronic version of OtherRealms to be a service I give to the net. If that service is really a problem in disguise, then perhaps it is time to re-think whether it really belongs here anymore. chuq (editor, otherrealms) Chuq Von Rospach chuq@sun.COM Delphi: CHUQ We live and learn, but not the wiser grow -- John Pomfret (1667-1703)
mcb@lll-tis.arpa (Michael C. Berch) (07/27/87)
I tend to agree with Chuq on this. He is using the "Expires:" field
for exactly the purpose for which it was intended: when the
author/poster of an article believes that it is of some more lasting
use or value as opposed to the normal flood of Usenet articles.
Otherrealms is a MAGAZINE. It's quite long, professionally edited, and
comes out quarterly (formerly monthly). Just like the way print
magazines live on the newstands, the previous issue should not expire
(e.g., be discarded) before the new one replaces it; there should
always be a "current issue" available for new readers. Compared with
the flow of a normal newsgroup, even a moderated one, its disk usage
should be statistically insignificant for administrative purposes.
But for those sites that don't want to keep OR for longer than the
default period, that is exactly what the override date options on
expire are for. Rec.mag.otherrealms can just be in an expire line by
itself with -e12 or whatever; overriding the date for it DOESN'T mean
you have to override the date for other things like the map postings.
In my opinion, Chuq is right to use the Expires: field the way he
does, and John Gilmore is right to expire rec.mag.otherrrealms
articles on his machine any way he sees fit. I don't see that there is
really a policy conflict.
(I do not claim to be unbiased; I am an enthusiastic reader of
OtherRealms and have contributed to it in the past. I'm also a news
administrator for a largish site and have to keep track of things --
like the fact that our /usr/spool overflowed this weekend...)
Michael C. Berch
ARPA: mcb@lll-tis.arpa
UUCP: {ames,ihnp4,lll-crg,lll-lcc,mordor}!lll-tis!mcbsean@ms.uky.edu (Sean Casey) (07/28/87)
In article <2525@hoptoad.uucp> gnu@hoptoad.uucp (John Gilmore) writes: >last month's OtherRealms (rec.mag.otherrealms) had a 1-month expiry >date, and that this month's 111K expires in November. I think this is an >abuse of Expires: and am not interested in keeping Chuq's "magazine" > [discussion of parlay back and forth between him and Chuq] That's why they have an "Expires:" line. Most articles need to be around about two weeks. Some need to be around more. Some need to be around less :-). I think Chuq's reasoning is valid. The newsgroup is extremely low volume. I doubt anyone is needlessly suffering for it. I believe the fact that no one else is complaining is an indication of this. John can: 1. Drop the newsgroup 2. Do a "find /usr/spool/news/mag/otherrealms -mtime +14 -exec rm {} \;" 3. Put a lot of pressure on Chuq. If John's site is having trouble, I'd recommend 1 or 2. Sean -- -- Sean Casey sean@ms.uky.edu, {uunet,cbosgd}!ukma!sean -- sean@ms.uky.csnet, sean@UKMA.BITNET -- We want... a shrubbery!
roy@phri.UUCP (Roy Smith) (07/28/87)
I agree with John! (*) Nuke the Expire: headers on OtherRealms. To
expound on how rec.mag.otherrealms is different/better/whatever than the
other groups, and thus deserves a longer expiry is pretty arogant (in my
opinion).
(*) I believe this is a first :-)
--
Roy Smith, {allegra,cmcl2,philabs}!phri!roy
System Administrator, Public Health Research Institute
455 First Avenue, New York, NY 10016roy@phri.UUCP (Roy Smith) (07/28/87)
In article <6957@g.ms.uky.edu> sean@ms.uky.csnet (Sean Casey) writes: > I doubt anyone is needlessly suffering for it. I believe the > fact that no one else is complaining is an indication of this. Actually, I think the reason nobody complained is because nobody noticed. Most of us don't have a lot of time to waste babysitting news and resent it when anybody does anything which makes it more difficult to run. True, I could special-case r.m.or and put a "find" in my /usr/lib/crontab to zap the old articles, but think what would happen if everybody started doing that? What happens when the good folks in unix-wizards decide they don't like answering the same questions over and over again so they start putting long Expire: headers on their answers? And what about the AIDS info postings; don't they deserve long Expire: lines? And the definitive compendiums of light bulb jokes? And..... As for Chuq's complaint that if he doesn't put Expire: headers in, the arbitron poll will miss his stuff, that's the kind of pathetic reasoning I expect out of New York City politicians. A couple of years ago, we had a case of a bicycle hitting a pedestrian. The pedestrian was wearing earphones so couldn't hear the bike coming. The City Council passed a law making it illegal for *bicyclists* to wear earphones! The answer is to fix arbitron, not cow-tow to its bugs by keeping articles around longer. -- Roy Smith, {allegra,cmcl2,philabs}!phri!roy System Administrator, Public Health Research Institute 455 First Avenue, New York, NY 10016
roy@phri.UUCP (Roy Smith) (07/28/87)
In article <24226@sun.uucp> chuq@sun.UUCP (Chuq Von Rospach) tries to
justify why he puts long Expire: headers on OtherRealms. From what I can
tell, it comes down to the fact that doing so cuts down on requests for
retransmission, which costs in modem time and phone bills. Besides, argues
Chuq, disk space is cheap.
Seems to me that if people want to get "reprints" of OR, they should pay
the phone bills themselves for a direct UUCP connection. Maybe you could
set up a public limited-access uucp account with a archive server as so
many other people have done? Or, if you want to be nice, you could pay the
phone bills yourself. Then it becomes a matter of comparing your phone
bills to my disk space and the equation becomes kind of lopsided.
--
Roy Smith, {allegra,cmcl2,philabs}!phri!roy
System Administrator, Public Health Research Institute
455 First Avenue, New York, NY 10016brad@looking.UUCP (07/28/87)
I agree 100% with Mr. Gilmore. mag.otherrealms should definitely not be posted with a special expiry date. Chuq's reasons (especially the arbitron one) are just plain silly. Because we won't be able to stop people from abusing expiry dates (even moderators) the answer may have to come from smarter expire software. Perhaps expire software should take a list of groups, and be given parameters for the group. (pattern matching, like "rec.all" would apply) Parameters like maxiumum size of the directory in bytes, whether to ignore the expiry date, default and MAXIMUM expiry periods, and perhaps something like a priority number (this group is boring, get rid of it quickly). Others might include maximum number of articles in the group etc. Chuq: News is for the readers, not for the posters. If your readers can't catch your fanzine during the default expiry for their machine, have them take it up with their sysadmin, not you. (You can have some groups expire more slowly, even with current expire.) You are correct that you shouldn't be mailing off lots of copies to people who miss an issue, but that doesn't suggest you userp the sysadmin's position of deciding how long news stays around. I always thought of expiry dates as being something of the reverse. You put a SHORTER expiry date on an article that you know won't be meaningful in more than a few days. This you do as a courtesy to the readers. News is so bad now that with my 70 megabyte filesystem and a feed that doesn't send me most of the big groups, I still have to expire in 2-3 days on most groups, and 1 day on several. Not trusting expire, I now just do a "find" that kills anything older than 4 days. -- Brad Templeton, Looking Glass Software Ltd. - Waterloo, Ontario 519/884-7473
jbuck@epimass.EPI.COM (Joe Buck) (07/28/87)
In article <2821@phri.UUCP> roy@phri.UUCP (Roy Smith) writes: > I agree with John! (*) Nuke the Expire: headers on OtherRealms. To >expound on how rec.mag.otherrealms is different/better/whatever than the >other groups, and thus deserves a longer expiry is pretty arogant (in my >opinion). To expound on how rec.mag.otherrealms is different/better/whatever than the material in any other rec, soc, or talk group is plain and simple fact. Arranging so that one issue will stay around by default is reasonable to me. -- - Joe Buck jbuck@epimass.epi.com {seismo,ucbvax,sun,decwrl,<smart-site>}!epimass.epi.com!jbuck Old arpa mailers: jbuck%epimass.epi.com@seismo.css.gov
vnend@engr.uky.edu (D. V. W. James) (07/28/87)
In article <2821@phri.UUCP> roy@phri.UUCP (Roy Smith) writes: > I agree with John! (*) Nuke the Expire: headers on OtherRealms. To >expound on how rec.mag.otherrealms is different/better/whatever than the >other groups, and thus deserves a longer expiry is pretty arogant (in my >opinion). >Roy Smith, {allegra,cmcl2,philabs}!phri!roy It may be 'arogant', but that does not change the value of the arguements presented. I am posting this from a small system, but even here the cost of maintaining 111k of disk space is trivial. Chuq has presented valid arguements for his use of the Expires: field. OtherRealms is posted to the net as a service, an electronic version of a printed magazine that has attracted national attention for its reviews of SF and Fantasy books. It is argueable that Chuq has, in posting OR to the net, lessened his paid subscriptions. It is also true that he has reached far more people via the Net than he would have otherwise. As his reasons for using the expires field seems valid, and in my personal case allows me to be certain of seeing issues that I could easily miss with a two week expiration, I agree with Chuq's use of the field. -- Later y'all, Vnend Ignorance is the Mother of Adventure. cbosgd!ukma!ukecc!vnend; vnend@engr.uky.edu; vnend%ukecc.uucp@ukma.BITNET Also: cn0001dj@ukcc.BITNET, Compuserve 73277,1513 and VNEND on GEnie Anyone going to Simple Day (in Indianapolis) let me know, I'll see you there!
sean@ms.uky.edu (Sean Casey) (07/28/87)
In article <2823@phri.UUCP> roy@phri.UUCP (Roy Smith) writes: > Actually, I think the reason nobody complained is because nobody >noticed. If you didn't notice, then you weren't suffering, were you? :-) >What happens when the good folks in unix-wizards decide they >don't like answering the same questions over and over again so they start >putting long Expire: headers on their answers? The fact is, they are NOT putting long expires. If that problems arises, the let's deal with it. In the meantime, Chuq's reasoning is completely valid. I have read the arguments against, and so far no one has said "Gee, rec.mag.otherrealms is causing me hardship by using all my disk space". THAT would be a legit argument. All this speculation and "arguing in principle" isn't really legit because it has nothing to do with reality. If you're insulted by long expire times, you can override them. Don't punish the people that benefit from them with your personal feelings. If you are having real problems because of them, then you should let Chuq know. Sean` -- -- Sean Casey sean@ms.uky.edu, {uunet,cbosgd}!ukma!sean -- sean@ms.uky.csnet, sean@UKMA.BITNET -- We want... a shrubbery!
rlong@felix.UUCP (Roger L. Long) (07/28/87)
In article <2525@hoptoad.uucp> gnu@hoptoad.uucp (John Gilmore) writes: >While burning disk space all over the world is a "solution" to Chuq's >problems, it bothers me. So far nobody else has refused to remove >Expires: lines or failed to cancel "never expires" type postings, once >I explained what Expires: was for, and how individual sites manage >their disk space. (Every once in a while I read the front of my >history file to see what old articles still linger there, and email the >folks who did it, educating them about how untutored use of Expires: is >a Bad Thing.) Chuq seems to see his group as more important than the >rules for Expires: lines. Do the rest of the news admins agree? I don't see that Chuq is breaking any rules with his use of Expires header lines. His reasons are perfectly valid, and as he's pointed out, anyone who feels the need to override the expiration date at a given site is welcome to do so. I really don't see what you see is the problem with this. -- Roger L. Long FileNet Corp {hplabs,trwrb}!felix!bytebug
warren@pluto.UUCP (Warren Burstein) (07/28/87)
In article <6957@g.ms.uky.edu> sean@ms.uky.csnet (Sean Casey) writes: :John can: :1. Drop the newsgroup :2. Do a "find /usr/spool/news/mag/otherrealms -mtime +14 -exec rm {} \;" :3. Put a lot of pressure on Chuq. No reason to do this, you can fix your problem easily enough. Better yet, 4. Use options so expire ignores expiration dates on rec.mag.otherrealms. See the manual for expire. -- /|/~\~~\ The entire world Warren Burstein |__/__/_/ is a very strange carrot. | But the farmer philabs!tg!pluto!warren / is not afraid at all. Why doesn't life come with subtitles?
csg@pyramid.pyramid.com (Carl S. Gutekunst) (07/29/87)
>>To expound on how rec.mag.otherrealms is different/better/whatever than the >>other groups, and thus deserves a longer expiry is pretty arogant.... I'm surprised by the number of people (observed over the past year) that have trouble with an author who thinks his meticulously edited magazine is "better" than the drek the engulfs even many of the moderated groups. While certainly some groups are a judgement call, I don't think that there can be any dispute that the edited magazines on USENET -- OtherRealms, alt.gourmand, Risks, Arms Digest, and a few others -- are of vastly superior quality, and the authors are indeed providing us with a service. And if the author asks (even demands) special handling for his superior product, I am perfectly willing to comply. And we put our money where our mouth is: we're paying to ship alt.gourmand around the world. (Literally.) Anyway, the level of Chuq's arrogance is not the issue here; it is whether the nature of rec.mag.otherrealms warrants long expiration dates. And I vote yes. I'd claim that *any* newsgroup that is periodical in nature -- including comp.mail.maps and news.announce.newusers -- should be set to expire for at least as long a period as it is posted. I'm interested in minimal user confusion, with minimal intervention on my part. Empty newsgroups, especially on popular topics, definitely cause confusion. The author knows when his next publishing date is, and I do not; so it's easier for me when he indicates this in the header (instead of my having to run a special expire to keep the group around longer). When disk space is a problem, there is always find(1); but OtherRealms is TINY relative to just two day's worth of control messages or a week's worth of talk.bizarre, so it is not very high on my list of places to go salvaging space. It is a problem of meeting the needs of the lowest common denominator. Small and minimally managed sites will be the least confused by keeping the latest articles in the group at all times. Those of us who run a tighter ship and are most concerned about disk space can exercise our rm privileges. <csg>
grr@cbmvax.UUCP (George Robbins) (07/29/87)
In article <4040@pyramid.pyramid.com> csg@pyramid.UUCP (Carl S. Gutekunst) writes: > >>To expound on how rec.mag.otherrealms is different/better/whatever than the > >>other groups, and thus deserves a longer expiry is pretty arogant.... > ... > Anyway, the level of Chuq's arrogance is not the issue here; it is whether the > nature of rec.mag.otherrealms warrants long expiration dates. Look, I like otherrealms, and think it's worth keeping around. To accomodate this I put a few characters in my expire script that say !mod.mag.otherrealms. To force expiration despite explict expiration dates without screwing up the rest of the groups I like to keep around, I'd have to run a separate expire run or do some off the wall diddling. It is arrogant to assume that some particular material is "so good" that it should override the recipient's decisions about news expirations. Why do so? What would you think if the magzaines you recieve in the mail said "warning: do not discard before ...."? -- George Robbins - now working for, uucp: {ihnp4|seismo|rutgers}!cbmvax!grr but no way officially representing arpa: cbmvax!grr@seismo.css.GOV Commodore, Engineering Department fone: 215-431-9255 (only by moonlite)
barnett@vdsvax.steinmetz.UUCP (Bruce G Barnett) (07/29/87)
In article <2824@phri.UUCP> roy@phri.UUCP (Roy Smith) writes: |Seems to me that if people want to get "reprints" of OR, they should pay |the phone bills themselves for a direct UUCP connection. This is fine if you are the system administrator. But if you were, you wouldn't be asking because you would have archived it. Most, if not all, of the requests come from people who don't manage the UUCP connection - I would bet. Perhaps Chuq did make a mistake when the Expires: of the last set overlapped the new issue by 13 days. Perhaps he should have waited 13 days before posting the current issue. (1/2 :-) Maybe I'm dense and am missing something. I would assume every site has an expire script, detailing which articles to archive, and which to expire. If you are tight for space, you change the script! If you use agef, you can see EXACTLY which newsgroups are taking up room. You then change your script to suit your needs. If fact, the first step is to (probably) expire some of the talk.* groups - overriding the default expiration date. This is exactly what you would do if otherrealms causes you the same problem. Right? I don't hear anyone complaining about news.announce.conferences, which on my system has 45k of articles older than 45 days. We have wasted more time and disk space discussing this `issue' than issue 16 of otherrealms (91K) overlapping issue 17 (114K) for 13 days. On July 23, the `problem' went away (assuming you expire daily). I believe the `cure' is becoming more of a problem than the `disease'. -- Bruce G. Barnett barnett@ge-crd.ARPA, barnett@steinmetz.UUCP, uunet!steinmetz!barnett
michael@stb.UUCP (Michael) (07/29/87)
In article <2525@hoptoad.uucp> gnu@hoptoad.uucp (John Gilmore) writes: >I don't like to override expiration dates, because that blows away >things like the maps, news lists, and conference announcements that >have a legitimate reason to stick around. But this requires that Gee, and I'd like maps, news lists, and conference anouncements to disapear when they should. Here's a bigger question: Is there ANY legitimate reason for an expires: line to lengthen the time until expiration? Maps and conference announcements are no more legit than otherrealms (in my opinion). -- : Michael Gersten seismo!scgvaxd!stb!michael : Copy protection? Just say Pirate! (if its worth pirating)
booter@well.UUCP (Elaine Richards) (07/29/87)
I agree with Gilmore. Think of this, friends. Our tax dollars pay for a lot of UNIX sites. My tax dollars are currently padding Chuq's ego. John's phone bill and disk storage costs are higher due to this behavior. My system, the WELL is often clogged with stuff people never read. I get very tired of slogging throught the WELL's disks with a shovel and pick- axe extricating postings and files generated by people who thing their words should be carved in stone. Don't abuse the system. Period. E *****
mcb@lll-tis.arpa (Michael C. Berch) (07/29/87)
In article <2164@cbmvax.UUCP> grr@cbmvax.UUCP (George Robbins) writes: > [...] > It is arrogant to assume that some particular material is "so good" that it > should override the recipient's decisions about news expirations. Why do so? > What would you think if the magzaines you recieve in the mail said "warning: > do not discard before ...."? But it simply doesn't work that way. The site admin ALWAYS has the last say; the poster can't "override" anything. All an "Expires:" header says is, "I [the poster] believe this material is of more lasting value than the norm. If you [the site admin] disagree, override it using the tools provided by expire (-e, -E)." What's the hassle about treating one group separately? We do two expire runs, one for archived groups (sources, bugs) and one for everything else. Adding a third would be no problem at all; I'm thinking of expiring the talk and soc groups in 14 days instead of 21 to save a little additional space and provide a bigger cushion. Jeez, it's not like you have to sit there and key it in every night or anything. Michael C. Berch ARPA: mcb@lll-tis.arpa UUCP: {ames,ihnp4,lll-crg,lll-lcc,mordor}!lll-tis!mcb
jbuck@epimass.EPI.COM (Joe Buck) (07/30/87)
In article <3646@well.UUCP> booter@well.UUCP (Elaine Richards) writes: >Think of this, friends. Our tax dollars pay for a lot of UNIX sites. >My tax dollars are currently padding Chuq's ego. John's phone bill >and disk storage costs are higher due to this behavior. John's phone bill is exactly the same, except for the tiny number of extra bytes for the Expires: header. The articles are sent to him only once. His disk costs are exactly the same, because if he has room to store the otherrealms issue when it first arrives, he has room to store it until the next issue arrives. The Expires: line is designed to keep one and only one issue around at a time. Chuq's behavior isn't costing John (or my company) a dime. We've already generated a year's worth of otherrealms traffic just arguing about the damned thing. It is John's anger at Chuq's behavior that's costing him (and us) money. -- - Joe Buck jbuck@epimass.epi.com {seismo,ucbvax,sun,decwrl,<smart-site>}!epimass.epi.com!jbuck Old arpa mailers: jbuck%epimass.epi.com@seismo.css.gov
dave@lsuc.UUCP (07/30/87)
Whether or not one believes Chuq should be using long expiry
dates (I personally see no problem, having read his explanations),
this discussion has revealed one point: the fact that the long
expiries weren't being brought to anyone's attention until
John posted his article.
Obviously, any sysadmin who wishes to override a group's
expiries can easily do so. Perhaps what's needed is a utility,
available for all sysadmins to have run once a week or
whenever, that would provide details of currently pending
extended expiries. Number of articles per group and total
size involved should be provided. Sample output
Group >14 days Total size
rec.mag.otherrealms 5 132K
comp.unix.wizards 2 6K
etc., with a total figure at the bottom. This could be
written easily enough in sh+awk or C; if anyone cares
to write it, I'm sure lots of news admins will use it
to monitor any potential abuse of the Expires: header.
In its most primitive form, this program is merely
find /usr/spool/news -mtime +15 -type f -exec ls -s {} ';'
David Sherman
Toronto
--
{ uunet!mnetor pyramid!utai decvax!utcsri ihnp4!utzoo } !lsuc!daveane@hal.UUCP (Aydin "Bif" Edguer) (07/30/87)
Intelligent Peoples Court, July 30, 1987.
I sure can't preside here but I will make a statement.
John, Chuq, I am sorry, you both are right. A consequence of an imperfect
universe.
In USENET news 2.11 there exists the option to change the default expire
date for an article. Though normally fourteen days, it can be changed to
almost anything. For reasons I leave as an exercise to the reader, this
option should be used sparingly.
For these reasons, I agree with you John that before ever using the option
of extending the expiration period, alot of thought should go into the
real need to do so.
Current examples of places where this option is in use is
news.announce.conferences (moderated)
news.announce.newusers (moderated)
news.config (15990@gatech.gatech.edu)
news.stargate (43389@beno.seismo.CSS.GOV)
What all these have in common is that they have been posted by responsible
people and have been given reasonable expiration dates.
What makes the expiration dates reasonable?
In my opinion it is reasonable to leave conference announcements up until
the conference. It is reasonable to leave the Call-for-Papers up until the
deadline for submission. It is reasonable to leave the information, we
expect all new news users to read, up until a new version can be posted.
These are examples of important information that have an explicit real world
expiration.
I agree with Chuq that it is reasonable to leave Otherrealms up until a
new issue comes out.
What is the difference between Otherrealms and say comp.sources.unix?
Comp.sources.unix is certainly run by a responsible person and contains
valuable information. The difference is in the SIZE (by almost two orders
of magnitude) and the fact there is NO reasonable expiration date.
The source will always be valuable and thus has no explicit time when it
should be expired and therefore MUST fall back to the default.
We can argue whether the default is too long or too short but we currently
leave that up to the individual site. And besides...that is a different
topic of discussion.
Just for reference:
Newsgroup Current Size Oldest Furthest Away
news.announce.conferences 90554 bytes Apr 19 Jun 20 1988
rec.mag.otherrealms 112180 bytes Jul 10 Nov 7 1987
Aydin Edguer
USENET : !{cbosgd,decvax,sun}!cwruecmp!hal!ane
INTERNET: edguer@cwru.EDU
Biometry Computer Lab, Case Western Reserve Universityeric@hippo.UUCP (Eric Bergan) (07/30/87)
In article <4160@felix.UUCP>, rlong@felix.UUCP (Roger L. Long) writes: > In article <2525@hoptoad.uucp> gnu@hoptoad.uucp (John Gilmore) writes: > >While burning disk space all over the world is a "solution" to Chuq's > >problems, it bothers me. > > ... > >Chuq seems to see his group as more important than the > >rules for Expires: lines. Do the rest of the news admins agree? > > I don't see that Chuq is breaking any rules with his use of Expires header > lines. His reasons are perfectly valid, and as he's pointed out, anyone > who feels the need to override the expiration date at a given site is > welcome to do so. It seems to me that Chuq is making a reasonable use of the Expires header. What John seems to argue is that we should not support the "Expires" header at all, or only for those articles which are explicitly "approved" to have a long expire date. But I don't know of any group of people that determines that an abnormally long "Expire" line is approved. Nor do I know any way of ensuring that all sites obey "the rules". As another poster pointed out, we are arguing about 112,000 bytes here. On our system, this translates into a one time cost of $2.22 - pretty cheap for a magazine subscription. You can argue that you are being forced to pay this, regardless of whether you want the subscription or not, and that the only way for you not to spend the money is by taking additional measures to "hand-expire" the articles. But I think this is true of a lot of the news. (Does everyone rely on the default 14 day expiration for all groups? We certainly haven't at all the sites that I have been associated with.) (It is true that I read and enjoy otherrealms. If this means that you now disregard all of my arguments as being biased - so be it. But I'm pretty sure that I would argue the same way, even if I hated otherrealms.) -- eric ...!ptsfa!hippo!eric
billw@wolf.UUCP (Bill Wisner) (07/31/87)
In article <3646@well.UUCP> booter@well.UUCP (Elaine Richards) writes: >Think of this, friends. Our tax dollars pay for a lot of UNIX sites. >My tax dollars are currently padding Chuq's ego. I challenge you to list fifty USENET sites that are paid for with our tax dollars. There's seismo -- but not for long. There's hao (?) and various sites at the Jet Propulsion Lab. Even if you peek at the map, I doubt you will find many. If you find and list over one hundred I will personally type the entire text of Tolstoy's WAR AND PEACE as penance, verifiable by my system administrator. Most -- nay, nearly all -- sites on USENET are paid for by private corporations, individuals, or universities. Choose one. These folks are NOT funded by tax dollars. Therefore, your reasoning has one helluva big hole in it. Perhaps next time you should reason your argument a bit more thoroughly before posting. -- Copyright (C) 1987, Bill Wisner. Redistribution prohibited if redistributor does not permit further distribution. UUCP: ..{ihnp4,sdcsvax}!jack!wolf!billw Unison: WABE
kyle@xanth.UUCP (Kyle Jones) (08/01/87)
> My tax dollars are currently padding Chuq's ego.
What does Chuq's ego have to do with this? While Chuq is the editor
for the OtherRealms magazine and he does have a column in it, he certainly
isn't the only person who writes for the magazine.
Perhaps Chuq (as a moderator) should have posted an article to news.admin
stating that he was going to use longer than normal expirations on his
OtherRealms postings. News administrators could then make the decision
whether to expire rec.mag.otherrealms separately (depending on whether they
agreed with Chuq's reasoning).
However, I can understand why Chuq didn't see the need to do so. 100K or so
posted every three months isn't much compared to the 2-4 Meg of news we get
daily.
Now that the long expirartion dates on OtherRealms postings are known, there
should be no problems. The news software is certainly flexible enough to
allow overriding the Expires: lines of rec.mag.otherrealms articles, without
expiring the maps, conference announcements, etc.allbery@ncoast.UUCP (Brandon Allbery) (08/02/87)
As quoted from <21616@lll-tis.arpa> by mcb@lll-tis.arpa (Michael C. Berch): +--------------- | I tend to agree with Chuq on this. He is using the "Expires:" field | for exactly the purpose for which it was intended: when the | author/poster of an article believes that it is of some more lasting | use or value as opposed to the normal flood of Usenet articles. +--------------- I second this, especially since I'm doing the same thing in comp.binaries.- ibm.pc now. Specifically: the constant requests for PKXARC and UUDECODE, which eat up a LOT of net bandwidth in both news and mail, plus the responses to them which eat up even more of both. So I post once a month (for the benefit of new sites) and the stuff I post has an expire date 5 weeks in the future. Once a month is better than 3 times a day by any scale of measurement. -- Brandon S. Allbery, moderator of comp.sources.misc and comp.binaries.ibm.pc {{harvard,mit-eddie}!necntc,well!hoptoad,sun!cwruecmp!hal}!ncoast!allbery ARPA: necntc!ncoast!allbery@harvard.harvard.edu Fido: 157/502 MCI: BALLBERY <<ncoast Public Access UNIX: +1 216 781 6201 24hrs. 300/1200/2400 baud>>
paul%tut.cis.ohio-state.edu@osu-eddie.UUCP (Paul Placeway) (08/02/87)
In article <1371@epimass.EPI.COM> jbuck@epimass.EPI.COM (Joe Buck) writes: >To expound on how rec.mag.otherrealms is different/better/whatever >than the material in any other rec, soc, or talk group is plain and >simple fact. Arranging so that one issue will stay around by default >is reasonable to me. I also agree with Chuq. If one is starved for spool space, then one can hand-expire things. If not, then it is reasonable to keep things like OR, comp.sources.*, comp.binaries.* (Yes, we (the CIS dept.) DO like the binary groups, and would probably go out of our way to get comp.binaries.macintosh even if we had to call Cleveland or UUNET) etc. -- Paul Placeway Department of Computer and Information Science SNail: The Ohio State University 2036 Neil Ave. Columbus OH USA 43210-1277 ARPA: paul@ohio-state.{arpa,csnet} (soon): paul@tut.cis.ohio-state.edu UUCP: ...!cbosgd!osu-eddie!paul (soon): ...!cbosgd!osu-cis!paul -=- -- Paul Placeway Department of Computer and Information Science SNail: The Ohio State University 2036 Neil Ave. Columbus OH USA 43210-1277 ARPA: paul@ohio-state.{arpa,csnet} (soon): paul@tut.cis.ohio-state.edu UUCP: ...!cbosgd!osu-eddie!paul (soon): ...!cbosgd!osu-cis!paul
dww@stl.stc.co.uk (David Wright) (08/02/87)
I usually resist the temptation to enter discussions which already have
lots of postings, but I feel strongly on this one.
As news admin. here, I try to keep news around for as long as I can,
within the limits of the disk space we can spare. When an expiry date is
set for good reasons - as I believe the Otherrealms and conferences ones
are - I consider this helpful, as I can set a reasonable expiry date
without removing the few things that should stay on line longer.
Recently the disks were re-arranged on this machine, with the result that
now there is more space for news, but in the interim I had to reduce news
space. It was no trouble to add the -i flag to over-ride the expiry dates.
If a local news admin does not want to keep Otherrealms until November (that
does seem a long time if you don't realise that the group really is a
periodical magazine) it should be no trouble to add (e.g.)
expire -e28 -i or even more specifically
expire -e28 -i rec.mag.otherrealms to the expire scripts.
P.S. Conflict of interests statement: I am an Otherrealms reader.
--
Regards,
David Wright STL, London Road, Harlow, Essex CM17 9NA, U.K.
dww@stl.stc.co.uk <or> ...seismo!mcvax!ukc!stl!dww <or> PSI%234237100122::DWWlear@aramis.rutgers.edu (eliot lear) (08/03/87)
Cc: lear@aramis.rutgers.edu In article <496@wolf.UUCP> billw@wolf.UUCP (Bill Wisner) writes: > > Most -- nay, nearly all -- sites on USENET are paid for by private > corporations, individuals, or universities. Choose one. These folks are NOT > funded by tax dollars. Therefore, your reasoning has one helluva big hole in > it. The above is a perfect example of misleading information. A huge cost of USENET is the communication of news. The federal government subsidizes most of the cost of such transfers by supplying (underhandedly) networks such as the ARPAnet. Take a good look at the backbone. Most of the sites are on a network like the ARPAnet which is mostly paid for by the federal government. -- Eliot Lear Rutgers University Department of Mathematics [lear@rutgers.edu]
chuq%plaid@Sun.COM (Chuq Von Rospach) (08/05/87)
In article <1794@xanth.UUCP> kyle@xanth.UUCP (Kyle Jones) writes: >> My tax dollars are currently padding Chuq's ego. Actually, your tax dollars are supporting a service I give to the net. I get my ego padded quite enough by the real version of OtherRealms, which is published and read (and enjoyed, and referenced [see the latest Zelazny paperback, for instance]) by the folks in the industry I'm writing for. I put it on the net as well so that the folks who have access to the net can enjoy it as well. Ego-boost? Being on the cover of Melissa Scott's new book is ego-boost. Being isn't even in the running -- I do that because I believe in putting something back into the net that has given me so much over the years. I don't believe in parasitism, but symbiosis. >Perhaps Chuq (as a moderator) should have posted an article to news.admin >stating that he was going to use longer than normal expirations on his >OtherRealms postings. News administrators could then make the decision >whether to expire rec.mag.otherrealms separately (depending on whether they >agreed with Chuq's reasoning). I said this once before, but evidently not loudly or clearly enough. The expiration date change was being tested over a couple of issues to see if it helped. John Gilmore was nice enough to take it public before the test was over, so by default the test is over and it's been hashed out. It would have been taken public when I was ready to defend it (or else it would have been killed silently if I didn't like the results). chuq Chuq Von Rospach chuq@sun.COM Delphi: CHUQ We live and learn, but not the wiser grow -- John Pomfret (1667-1703)
rob@array.UUCP (Rob Marchand) (08/10/87)
In article <1958@lsuc.UUCP> dave@lsuc.UUCP (David Sherman) writes: > >Obviously, any sysadmin who wishes to override a group's >expiries can easily do so. Perhaps what's needed is a utility, >available for all sysadmins to have run once a week or >whenever, that would provide details of currently pending >extended expiries. Number of articles per group and total >size involved should be provided. Sample output > >Group >14 days Total size >rec.mag.otherrealms 5 132K >comp.unix.wizards 2 6K > A program came around a month or two ago; it's called agef, and provides output of the form... 3 days 6 days 9 days 12 days 15+ days Total Name ---------- ---------- ---------- ---------- ---------- ------------ ---- 26 66k 6 13k 32 79k can 2556 7331k 594 1813k 14 33k 3164 9177k comp 4 4k 4 4k control 0 0k general . . . . . . . . . 417 889k 1 3k 418 892k rec 485 1219k 3 8k 488 1227k sci 94 224k 5 16k 99 240k soc 120 308k 2 3k 122 311k talk 1 1k 1 1k to 34 75k 48 116k 82 191k tor 4212 11230k 677 2033k 68 297k 4957 13560k Grand Total If anybody is interested, I can mail it out, or post it. (I forget who wrote it...It's likely in the source somewhere ... :-) Send me mail if you're interested. -- Rob Marchand UUCP: {mnetor,utzoo}!lsuc!array!rob Array Systems Computing ARPA: mnetor!lsuc!array!rob@seismo.css.gov 200-5000 Dufferin Street Phone : +1(416)736-0900 Fax: (416) 736-4715 Downsview, Ont CANADA M3H 5T5 Telex : 063666 (CNCP EOS TOR) .TO 21:ARY001
barnett@vdsvax.steinmetz.UUCP (Bruce G Barnett) (08/12/87)
In article <517@array.UUCP> rob@array.UUCP (Rob Marchand) writes: >A program came around a month or two ago; it's called agef, and >provides output of the form... I have the following in my daily cron script: agef `find /usr/spool/news -type d -print` >/tmp/news-size This gives you the size of each directory and newsgroup. >If anybody is interested, I can mail it out, or post it. It was just posted to comp.sources.unix - see v10i100 -- Bruce G. Barnett <barnett@ge-crd.ARPA> <barnett@steinmetz.UUCP>