On Mar 16, 2005, at 4:33 AM, Michael wrote:

> -current doesn't need load-base and real-base changed anymore ( at 
> least not on my boxes - a UMAX S900 with OF 1.0.5 and an iBook G4 ). 
> This leaves the fiddling with device paths - I think the way to go 
> here is to extend bootvars, it could...

Well, the key thing would be for NetBSD (and in particular sysinst) to 
be able to read and write to OF.  This way, sysinst could figure 
everything out and write the correct OF variables, reboot, and voila!  
Of course, sysinst would also come with a "quirk table" of sorts where 
it asks if you have a particular model (guessing from available 
hardware) to avoid or explicitly specify certain settings.  Things like 
beige G3 not being able to boot from slave IDE HD, warning about PB 
3400 ethernet not working, etc...

Don't forget a pause for HDs to spin up, some means of breaking out of 
the boot process, and (too much to ask for) some means of selecting 
which OS to boot and a program to run from that other OS which sets up 
OF to boot NetBSD.

On Mar 16, 2005, at 6:44 AM, Bill Squier wrote:

> I don't like the merging idea if it means the subsequent demise of the 
> pocket
> guide.  There's a difference between goals for the sake of goals and 
> goals
> that accomplish something.

Well, in the past many people have written their own "my guide to 
installing NetBSD on foo" and they've languished.  I'd say the "pocket 
guide" should be of limited scope (pre-burnt official NetBSD CD-R on OF 
3 machine), part of the "make release" process, and output into text, 
html, and PDF.  That said, the src/distrib/notes framework doesn't make 
PDF, but could handle the rest.  Especially since it'd be in the same 
place as the full install notes and (in theory) they could share source 
code.  Personally I hate writing in ROFF, but it's fairly well 
supported, is part of the build process, and will end up in all 
distributions of NetBSD.

> Nobody wants to be presented with a 50 billion page document that they 
> have
> to sift through to determine how to install their Mac Mini.  
> Especially if
> this document ridiculously self-referentially cautions that it SUCKS
> (http://tinyurl.com/4kha8).

Yeah, I'm well aware of how much the macppc INSTALL notes are.  It 
seems that only recently user interest in OF 3 machines has exceeded 
older models.  Cost, of course, is the primary issue.

   -- MW