"Perry E. Metzger" <perry@piermont.com> writes:
> Pardon, but why is a flag better than a kernel option in this case?
> Having it be a flag seems somewhat counterintuitive to me so I'm
> interested in your reasoning.

Uh, how about: because the behaviour is an optional behaviour in a
driver that's conditional on the state of a boolean flag?  (Many
things are kernel options that arguably should not be, and fall into
this category.)

Also, how about: because you may want to change the setting of the
flag on a per pciide-instance basis?

Also, how about: because then it's patchable?  (OK, you could do this
with an #ifdef and patchable kernel variable, too, but "see above.")


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