Newsgroups: news.admin
Path: utzoo!henry
From: henry@utzoo.uucp (Henry Spencer)
Subject: Re: Question on non-dbm history files
Message-ID: <1990Feb28.172640.25896@utzoo.uucp>
Organization: U of Toronto Zoology
References: <253@uucs1.UUCP>
Date: Wed, 28 Feb 90 17:26:40 GMT

In article <253@uucs1.UUCP> gaf@uucs1.UUCP () writes:
>Regarding the 2.11 software...
>If I understand rightly, these ever-growing files in my admin/history.d
>directory are used to detect duplicate articles.  Okay, but is that all
>they're for?

The one other purpose that dbm (and the assorted substitutes for it)
is intended to fulfill is random article lookup by message-ID.  Most of
the news readers will try to do this in some circumstances, but the
circumstances in question are usually use of some obscure command that
nobody ever invokes in practice.

>Since our site is fed by only one other site, the other site would presumably
>filter out dups.  This means I could just make these unbounded files links
>to /dev/null and save a bunch of space and effort.

If you bear in mind the possibility that it might break future newsreaders,
yes, this should be workable.  I think.  It's been a long time since I
looked at the 2.11 dbm fake.

Alternatively, the dbz package (distributed several ways, including in the
"contrib" directory of C News) is a fairly fast and space-efficient dbm
lookalike for news, although it may not solve your problems with expire.
-- 
"The N in NFS stands for Not, |     Henry Spencer at U of Toronto Zoology
or Need, or perhaps Nightmare"| uunet!attcan!utzoo!henry henry@zoo.toronto.edu
