From owner-freebsd-hackers  Tue Sep  5 13:49:30 1995
Return-Path: hackers-owner
Received: (from majordom@localhost)
          by freefall.freebsd.org (8.6.11/8.6.6) id NAA05227
          for hackers-outgoing; Tue, 5 Sep 1995 13:49:30 -0700
Received: from gvr.win.tue.nl (gvr.win.tue.nl [131.155.210.19])
          by freefall.freebsd.org (8.6.11/8.6.6) with ESMTP id NAA05219
          for <hackers@freebsd.org>; Tue, 5 Sep 1995 13:49:24 -0700
Received: by gvr.win.tue.nl (8.6.10/1.53)
    id WAA00730; Tue, 5 Sep 1995 22:05:53 +0200
From: guido@gvr.win.tue.nl (Guido van Rooij)
Message-Id: <199509052005.WAA00730@gvr.win.tue.nl>
Subject: passwword update times
To: hackers@freebsd.org
Date: Tue, 5 Sep 1995 22:05:52 +0200 (MET DST)
X-Mailer: ELM [version 2.4 PL24]
Content-Type: text
Content-Length: 733       
Sender: hackers-owner@freebsd.org
Precedence: bulk

I've been thinking about a possible solution to the large time
pwd_mkdb needs to update large password databases.

What about this possible solution:
For each of the programs passwd, chsh, chfn and chpass only one user per
invocation can be changed. This is unlike vipw where everything can be
done to the password file.
For the non-vipw ones, we could add an option to pwd_mkdb, let's say -u
that makes that pwd_mkdb, in staed of completely rebuilding the pwd.db
and spwd.db files, only updates (via put()) the entry for the specified user.
(so pwd_mkdb' operation remains the same as now, except it copies {s,}pwd.db
and does only one put()).
That would mean that only vipw has the performance problem.

What do you think?

-Guido