--kfjH4zxOES6UT95V
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

[Ccing back to port-macppc, Derek, as I'm only responding to stuff
you already posted there. :^>]

On Fri, 2 Nov 2001 17:37:19 -0800, Derek Peschel wrote:
> * 1.74.2.1 tries to call `usb-kbd-ihandle which doesn't exist on my syste=
m.
>   I have a USB keyboard (not an ADB keyboard -- AFAIK my computer doesn't
>   even support ADB).  So why am I still able to use my keyboard?
>=20
> * 1.86 tries to call `usb-kbd-ihandles.  What will happen if 1.86 is used
>   with an old OF that has only `usb-kbd-ihandle?

New person to try bugging, if I didn't say this already, would be
Lennart Augustsson <lennart@augustsson.net>, who filed the PR on
`usb-kbd-ihandle (14733). I wonder... can we just try both
`usb-kbd-ihandle and `usb-kbd-ihandles in succession without
breaking anything?

Lennart, you out there?

> * Can 1.86 be made to work with both method names?  Would it have to get
>   the OF version, or is there a way to ask OF directly if a method exists?
>   Note, ' is not what I had in mind since it prints an error if the method
>   doesn't exist.  I already see one error from OF (because of the call to
>   `usb-kbd-ihandle) and I would rather not see more.

A clean way to retrieve the OF version that works on all OF versions
would make a *lot* of our "works with this OF, not with this other"
issues go away. I don't think there is one. I find more and more
intricate ways in which to wish ill on the folks at Apple's family
trees the more and more I deal with OF.

> * When I run halt from the shell, I assume I'm supposed to be dumped back
>   to an OF prompt.  I do get the prompt but I can't type.  Does machdep.c
>   need to set standard input back to OF's on shutdown (just as it sets it
>   away from OF's on startup)?

Would this be PR 14732? (Also Lennart's.) Or, at least, related with
a similar fix? (He's talking about ddb, you're talking about a
halted system, but both need to reenable things properly for the
console.)

Looks like he's got a pretty reasonable solution for that one,
provided it doesn't also stray into "doesn't work with this OF
version" troubles...

> * I can imagine an error happening at an arbitrary point in the
>   initialization code.  The error might get me back to an OF prompt,
>   or it might bring up the kernel debugger.  Can machdep.c guarantee that
>   the input stream is correctly set for both possibilities (OF or ddb)
>   for all times during startup?

As near as I can tell, not right now. But fixing 14732 might fix
that (or be similar to the fix for that).

--=20
       ~ g r @ eclipsed.net

--kfjH4zxOES6UT95V
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (NetBSD)
Comment: For info see http://www.gnupg.org

iEYEARECAAYFAjwD6egACgkQ9ehacAz5CRqr/QCeNcus6+jiJgIdykyGz2MpxCt2
CRAAnipamxC5SVo53fAxX6eUDDxuAI9t
=nzts
-----END PGP SIGNATURE-----

--kfjH4zxOES6UT95V--