On Sun, 18 Mar 2001, Jason R Thorpe wrote:

> On Sun, Mar 18, 2001 at 04:32:05PM -0500, Joseph Mallett wrote:
> 
>  > I'm going to say this again, sure that most (all) people will oppose. I
>  > think DEVFS might do well to handle this. My 2p.
> 
> Whether or not you use DEFVS, you have to come up with a mapping
> *somehow*, if nothing else, a convention to use.
> 
> Plus, DEVFS has some problems that we won't bother going into on this
> list (they've been hashed over N times on tech-kern already, and besides,
> DEFVS is a little heavyweight for an embedded system).

I suppose you're right about the weight that would be added, I was just
thinking about say... /dev/maple/device[n], and n is based on what
slot/connector the device is attached, and nodes like /dev/maple/keyboard
being created on the fly as devices are recognised/configured. In any
case, thanks for the clear response, and I suppose the general view of
DEVFS (at least from the NetBSD project's perspective) isn't the most
favourable one.

Thanks for your time,
jm

> 
> -- 
>         -- Jason R. Thorpe <thorpej@zembu.com>
>