Subject: Re: Another changer, another changer problem
To: Curt Sampson , Robert Elz <kre@munnari.OZ.AU>
From: John Nemeth <>
List: current-users
Date: 10/10/1998 17:12:16
On Oct 5, 12:28am, Curt Sampson wrote:
} On Mon, 5 Oct 1998, Robert Elz wrote:
} >     No. All that means is you're stuffing things the kernel does not
} >     need to know about into the kernel,
} > 
} > Huh?   You've just said that the kernel already knows magic about the 'c'
} > partition ..
} Knowing a few lines of magic that is common across every NetBSD
} port is somewhat different from having specialised partition-reading
} magic that it is only used on one or a few ports. Of course, if
} you ever want to be able to read that disk on another port, since
} there's no userland utility available, you actually do need to put
} that code in all ports. This brings us to the point where the Sun/3
} kernel has code in it to read DOS partition tables, and more code
} to read Amiga partition tables, and more yet to read whatever it
} is Macs do....

     It could be defopt'ed.  Or, we can come up with some kind of
userland method.  There have been some suggestions along this line.
Either way, we're looking at complex changes that are going to need
some studying.

} > Aside from compatability with existing /dev/ entries, and labels, which is
} > no small issue....
} Indeed. To my mind, if we're going to make this sort of change,
} changing to a system where we would still have to worry about moving
} this aside so we could have one more partition is foolish. We should
} start by changing to a system that gives us so many partitions that
} using one for the NetBSD portion, one for the whole disk, and a
} dozen others for accessing other OS's filesystems still leaves us
} with more partitions than we're likely to use for our own work.

     I don't like this idea, because it means keeping the same
information in multiple places.  This information must be manually
kept in sync.  If it isn't then there is the potential for a major
disaster.  The only thing going for this method is that it is the
simplest to implement, and doesn't require much thought.  We could
certainly start with this method (it would be really nice to get more
NetBSD partitions); but, I would like to see something better

}-- End of excerpt from Curt Sampson