Lord Isildur drunkenly mumbled...
> 
>  we already have one! the kernel has been fine for years! We use the driver
> as the differentiating factor for different device 'types' and therefore
> device names. if some drifoo configures some device, then that device is 
> foo's device, since foo gets to control it. so, since it if the driver 
> foo who makes the interface thru the kernel to , ultimately a process, we 
> name the device foosomething. This is simple., consistent, and it makes sense
> considering that , well, the driver _is_ what drives the device! 

i don't have a problem with that.  my main concern, and this is also true in
Solaris, so this isn't a BSD specific thing, is how am i supposed to be 
expected to maintain a static configuration file of dynamic items?

the kernel has been fine for years, but does that make it right? maybe. i don't
think so, but that's just my opinion.

> what was religious rhetoric about what i just said? i said the same thing 

to me you implied, "This is how it was always done, so this is how it will 
continue to be done."  i would consider that religious rhetoric.

> a day or two ago! i think this is a very smart organization/convention for
> device names, and i find it nice that *BSD uses this convention, just one 
> more think i like about it, and i dont think we shoudl change this. I 
> find the soalris style names to be really awful. Some will agree with me 
> and some wont, certainly. 

there are a lot of things i don't like about Solaris.  the disk naming is not
one of those things.  i think it makes sense to do it this way.  you see widom
in the way BSD does things.

the only solution to this is you need to explain to me how floating device 
names for something like disks is a good idea.  i just cant figure out why.
maybe you haven't owned enough sun hardware. ;-) (i can't think of anyone else
who goes SCSI ID 3 and then 1 since most people start at 0 and work their way
up)

-brian