>It sucks, I admit it.  I just don't agree with the blanket statement
>of "there is no way to install on troublesome machine X without a
>sysctl hack for PCMCIA", when in reality an install is possibly, if
>not terribly practical.

Have you installed NetBSD on such a machine, _including_ the kernel
sources?  I don't think I own enough floppies to do it (I know, I could
rotate them ... but that's even more painful).  We _do_ want
people to run NetBSD, right?  If "install via floppy" is the answer,
then we lose even more potential users (the only reason my friend
is running NetBSD now was because I was able to frob things enough
to get it going).

>NetBSD has often "Done The Right Thing", as
>opposed to the practical thing.  IMHO the one off kludge to "get it
>to work", often ends up the final solution.  "If it isn't broken, don't
>fix it" comes to mind.

Well, as Jason has already pointed out, this doesn't fix the larger
problem.  At the very least, we'll have a hack in place that will be
ugly; at the worst, we'll be able to install NetBSD on a series of
laptops that we couldn't before.  Sure, it doesn't solve the larger
problem; people with SCSI interface cards still lose, but they lost
before.

Now, like I said ... I'll at least _try_ to look at these things
and see if I can make them better _and_ "Do the right thing".
Can't promise anything, of course (I still need to do Kerberos
first), but at least I'll try to understand it and figure out if
it's possible for me to do.

--Ken