Darren Reed <darrenr@reed.wattle.id.au> writes:
> This seems messy to me.

It's certainly interesting.  8-)

> Would it be worthwhile adding a buffer,
> somewhere, which various parts could add information to and was
> printed out once the `configure' was complete ?

It might, but it's kinda hairy.  I've not really thought about what it
would take to do that.

and, of course, that wouldn't solve the problem for any of the _other_
areas that use the calls like this...  8-)



> (I'm guessing,
> without RTFS'ing, that the function which prints out "sd0" also
> prints out the "\n").

oh, you wish it were that.  8-)


consider:

	foo0 at bar0 loc X loc Y: some device info\n
        AAAAAAAAAAAAPPPPPPPPPPPPDDDDDDDDDDDDDDDDDDDD

A: printed by generic autoconfig code.
P: printed by parent device 'print' function if any.
D: printed by child device attach function.

then consider the messages you get when you have an unconfigured
device on a direct config bus:

	usefuldevicename at bar0 loc X loc Y not configured\n
        PPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPAAAAAAAAAAAAAAAAA

same meanings as above (and of course the " not configured\n") bit
won't get printed if there's no parent device print function.



cgd
-- 
Chris Demetriou - cgd@netbsd.org - http://www.netbsd.org/People/Pages/cgd.html
Disclaimer: Not speaking for NetBSD, just expressing my own opinion.