> 
> sparta@imsa.edu wrote:
> 
> > I think the best way to build up a better installer and isntlal process is to
> > build a more robust kernel (as has already been suggested).  Either put
> > together an EVERYTHING kernel for the distribution or wait til the persons
> > writing things like ADBTEST and intvidtest are ready to put in mainline
> > NetBSD.
> 
> This is the goal. In fact, most of the ADB code could be merged into
> the official source tree right now. This actually was the plan except
> the 1.2 release unfortunately coinsided with my move across the country,
> so I was not able to work on it.
> 
> The problem is that certain machines work only with the official 
> MRG code. Other machines work only with the ADBTEST code. We don't 
> want to make the ADBTEST code part of the official source until 
> we know it won't make someone's machine (specifically ADB) stop working.
> I haven't had time to sit down and verify that everything works in
> ADBTEST that works with MRG.

One thing I've wondered is why not have merged the code into the kernel, and
set the two different adb accesses up as compile options? We have two
different SCSI drivers, why not two different ADB drivers. Thuough it
sounds like you're close, so the point might be moot.

Take care,

Bill