jerry@oliveb.UUCP (Jerry Aguirre) (03/09/84)
The documentation that comes with the news software talks about having
an optional history.pag and history.dir file. These are hashed
versions of the history file using the dbm(3) database routines. The
code seems to use these to check whether an article has already been
received. This sounds like it would be much faster than scanning the
entire history file to find the ID. At our site it seems to take
longer to process a news article than to receive it at 1200 baud. We
use dbm with delivermail for our rather large alias file and it works
well.
Does anybody have experience using the dbm library with news? Also does
anybody have software to scan the articles (or history file) and build
the database? We are running 2.8bsd and version B news.
Jerry Aguirre
{hplabs|fortune|ios|tolerant|allegra|tymix}!oliveb!jerry