Pete Carah pete at
Mon Sep 27 22:19:55 PDT 2004

On Mon, Sep 27, 2004 at 02:53:26PM -0700, Matthew Dillon wrote:
> :Fixed the mixup with pidfile that was preventing kerberos5 from being
> :built after pidfile() was added to libutil.  This could probably be done
> :right by redoing configure with the later libutil; configure does look
> :at this.  However, I don't know the canonical method, other than just by 
> :hand, of getting from autoconf contrib packages to the bmake system...
> :
> :Does noone else use kerberos?  I've seen no complaints.
> :
> :-- Pete
>     Maybe we should rename pidfile() in libutil to something else
>     instead.  It's a pretty aweful name for libutil to choose to use 
>     anyway!  Maybe something like make_pidfile() ?  Comments?  From
>     what I can tell there isn't much in our tree that actually uses it.
> 					-Matt

Well, it was apparently a direct import from netbsd, done only last week.
I don't know who committed it or why (looks like joerg).  Kerberos correctly 
identifies it if you run configure, so they expect it there...  Our compile 
only fell over because of the BMAKE stuff where we use pre-done "configure" 
steps.  OTOH I agree that it is a lousy name for a function.  (I see 
write_pidfile() and variants lots of other places; I'd think that would be 
better...)  My intuition says that "pidfile" ought to be either a string 
variable or a FILE *...

I know nothing of this functions's history.

-- Pete
/*      $NetBSD: pidfile.c,v 1.6 2001/10/20 09:20:28 taca Exp $ */
/*      $DragonFly: src/lib/libutil/pidfile.c,v 1.1 2004/09/22 05:06:57 joerg Exp $ */

 * Copyright (c) 1999 The NetBSD Foundation, Inc.
 * All rights reserved.
 * This code is derived from software contributed to The NetBSD Foundation
 * by Jason R. Thorpe and Matthias Scheler.

More information about the Kernel mailing list