>    > I asked a very specific question, relating to your model of probing
>    > based on I/O addresses, and the fact that at least some devices don't
>    > *have* an I/O address, and you have never explained how you made that
>    > case work.
> 
>    what does "probing based on I/O addresses" mean?  _certainly_ i never
>    said that, since i have no idea what you mean by it.
> 
>    There's definitely no requirement in my code that devices have I/O
>    addresses.  I dunno what hat you pulled _that_ out of.
> 
> I pulled nothing `out of a hat', and I resent the insinuation that I'm
> lying.  You specifically told me that you were probing for all devices
> listed at a specific I/O range, and then choosing the one with the
> highest priority.

This is almost hilarious.  I've never implemented that, nor have i
thought about actually implementing it.  I _certainly_ did not
"specifically tell" you that i was doing that, because i never have
done it!  Because you say that I did, i will say flat out that either
ou are lying, or are making statements up to suit your notion of what
i said.

I believe that i have suggested at several times that such a thing
might be good to try to implement, to get back the original "match"
semantics.  However, i've not really thought about how i'd implement
it, let alone actually given it a shot.  At this point, for various
reasons, i don't think that's a reasonable way to probe ISA devices.
I came to this conclusion not because of your point about needing an
I/O port to do it (which i do _not_ recall you ever making), but
rather because of the nasty things that badly-written ISA probes can
do the devices. 


> Further on, you even said that you had talked with other people about
> the design of your code, and specifically not talked to me about it.
> This lack of communication is poor on your part.

see the previous comment about nitpicking a design that was
specifically described as 'rough.'

If it's highly unlikely that i'll receive constructive criticism from
you (and my interactions with you over the last 2+ years have led me
to believe that), then why am i going to waste my time?


>    Of course, if you know it's necessary for NetBSD/Alpha, and you don't
>    bother to look at the NetBSD/Alpha sources, then you still are missing
>    something.
> 
> The code *was not available to me*.  I've said that several times.

The code in question has been available as long ago as June 8, 1995
(which is when i made my initial source and binary distribution that
supported PCI alphas).

Unless i'm mistaken, my original discussion with you happened after
that.  If not, the code has been available for just under 6 months
now, so you certainly can't give the excuse that you couldn't look at
it _NOW_...  (and, as noted, if you look at it now: it's still a work
in progress, and there are a large number of things about it that i
_know_ i want to redo, so...)


cgd