Subject: Re: sample utmpx implementation
To: firstname.lastname@example.org (Christos Zoulas), Andrew Brown <email@example.com>
From: John Nemeth <firstname.lastname@example.org>
Date: 12/05/2001 05:56:53
On Apr 13, 5:39am, Christos Zoulas wrote:
} On Nov 21, 2:55am, email@example.com (John Nemeth) wrote:
} | Do we really care about people doing the wrong thing. It is next
} | to impossible to protect the system from incompetent admins. I don't
} | think we should greatly complicate the system just on the off chance
} | that an admin might do something silly. Especially since making
} | lastlogx a db would greatly increase its size for the people that do
} | the right thing and leave it alone. I.e. this would actually penalise
} | people that do the right thing, which would be bad.
} Considering how sparse files work, a db file might be smaller (I
} think the way it works is that you need to allocate BLKSIZE on each
} block you write). These days it is not uncommon to have uids in
That sounds right.
} the 20 thousand range. Sparse files are still not commonly used in
UIDs on most systems normally only use a couple of ranges, so I
don't think there would bw much wastage.
} unix, and it could be mystifying to the uniniated to have a 40G
} file in a 1G disk. I agree that you cannot prevent a stupid person
Then, I guess they have an opportunity to learn something new :-)
} from doing stupid things. OTOH, db is part of libc, and the penalty
} for using it is minimal...
It is a heck of a lot more complicated then seek and read...
}-- End of excerpt from Christos Zoulas