Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!decvax!decwrl!pyramid!pesnta!amd!amdcad!lll-crg!topaz!cord!ebh From: ebh@cord.UUCP Newsgroups: net.news.group,net.news.adm Subject: Re: If You Love This Usenet Message-ID: <262@cord.UUCP> Date: Tue, 27-May-86 16:34:24 EDT Article-I.D.: cord.262 Posted: Tue May 27 16:34:24 1986 Date-Received: Wed, 28-May-86 19:25:59 EDT References: <2140@cbosgd.UUCP> <252@ubc-cs.UUCP> <402@pyramid.UUCP> Reply-To: ebh@cord.UUCP (Ed Horch) Followup-To: net.news.adm Distribution: na Organization: AT&T Bell Laboratories, Liberty Corner Lines: 32 Keywords: expiration Xref: decwrl net.news.group:5105 net.news.adm:767 In article <402@pyramid.UUCP> csg@pyramid.UUCP (Carl S. Gutekunst) writes: >Hey, a lot of sites already have a much simpler mechanism: They let >/usr/spool/news overflow the disk once in a while.... >:-), sortof.... There seems to be one type of usenet site that nobody's addressed here. I'm talking about sites whose news connections cost nearly nothing in connect charges, because they're all local calls, or whatever. On these systems, disk space seems to be the limiting factor, even with scheduled expiration. What I propose for this type of problem is to have inews and arbitron interact: if arbitron knows that nobody reads net.X at this site, pass the articles to all downstream sites, the expire them immediately. This way, everybody gets all articles, but they don't hang around gathering dust (which is bad for disk drives :-). It may even be valid to check all .newsrc files, looking for newsgroups that nobody's read in some magic number of days. The question I don't have an answer to offhand is this: "After six months, I want to start reading net.Y again. How can I get back up to date with what's going on?" There are other ways to make expiration more intelligent as well, most of which could be implemented by hand: 2 weeks for net.A and net.B, 3 days for net.C, 10 minutes for net.jokes.. :-) Just a thought. -Ed Horch {ihnp4,cbosgd,allegra}!cord!ebh