Spec suggestions regarding IDE

[Artem Kachitchkine]

> If I understand then, the host/channel properties are purposefully not 
> describing the device absolutely, because one can infer what controller 
> a device is on by inferring the device tree up from the device. Makes 
> sense. I definitely think the specification should be updated to clarify 
> however. I could help if the effort's needed.

Many parts of the spec could use clarification. I suppose the usefulness 
of each depends on the consumers. I don't know of any consumers of the 
IDE-related properties, are there any?

There's also an interesting question of how complete HAL's device tree 
should be compared to the OS-specific view of hardware. For any given 
class of devices, the efforts associated with maintaining the extra code 
and keeping the two views in sync might not be worth it until there's at 
least one consumer. Even if that consumer is a human wanting to inspect 
the system, but that doesn't seem to be the requirement at the moment 
(lshal and hal-device-manager are developer tools).

-Artem.