Subject: Re: Time to fix a 25 year old misdesign
To: NetBSD Kernel Technical Discussion List <tech-kern@netbsd.org>
From: Greg A. Woods <woods@weird.com>
List: tech-kern
Date: 10/15/2000 14:21:51
[ On Sunday, October 15, 2000 at 10:38:08 (-0700), Matthew Jacob wrote: ]
> Subject: Re: Time to fix a 25 year old misdesign
>
> > The close semantics are correct for fully shareable devices like disks. They
> > are helpful in that you have different functions you're trying to accomodate
>      ^
>      UN
> 
> > w/o having a separate device.

Are not all properly designed device interfaces sharable (to multiple
processes) in at least certain circumstances?  Why should a design
that's correct and most elegant for almost all situations have to be
bent to accomodate the 0.00001% of situations where the driver designer
"forgot" to logically separate functionality out to separate minor
device numbers (or even entirely separate drivers)?

I.e. why would anyone want to accomodate unrelated device functions with
the same /dev file in the first place?  This is exactly the kind of
thing that leads to trouble -- avoid it early on and you're much better
off in the long run.  If something's worth doing then it's worth doing
right the very first time.  Anything that's done wrong the first time
will have to be done again and again until it's finally done right.

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>