In message <20000613084842.D691@dr-evil.z.zembu.com>, Jason R Thorpe writes:
>Sigh.
>
>One of the things that's missing in this discussion is where ACPI fits
>into the grand scheme of things.

On some level, this is less than completely relevent. The hard
part about ACPI is not the interface between the ACPI code and
the user -- it is the parsing of the virtual machien stuff.

>ACPI is *NOT* the correct way to implement MI power management.  Instead,
>we need a higher-level power management API.  ACPI would be just one
>back-end to that high-level API.

Sure. Ithink we all agree on this.

>It just seems silly to say "ACPI ACPI ACPI!" when so few of the supported
>NetBSD platforms will actually *have* it (and would thus have to emulate
>it, a pretty gnarly task).

It seems quite reasonable to say "We want ACPI support on these machines!"
because it is the only power management mechanism that is useful
on them. That doesn't mean that anyone is proposing emulating it
or making it the standard interface.

--jhawk