New to HAL and confused.

[Kay Sievers]

On 1/13/07, Richard Hughes <hughsient at gmail.com> wrote:
> On Sun, 2007-01-14 at 08:43 +0000, Alan Chandler wrote:
> >
> > > However, not all references keys are actually in the table.  For
> > > instance - in several places in the text of the document there is a
> > > reference to block.storage_device key - but under the table which
> > > lists the allowable keys in the block namespace THAT key doesn't
> > > exist
>
> Yes, sometimes the spec wasn't always updated as well as it should have
> been. We really need some sort of auto-checking feature that checks if
> keys that are trying to be got or set are valid in the spec. - Maybe we
> could xlst the xml spec to key the keys, and then add a --check-keys
> option to hald that checks the keys against the list.
>
> > > Also the  info.bus key has a list of values it is allowed to be, but
> > > usb_device isn't one of them. Yet the introduction to the usb_device
> > > namespace the paragraph specifically says that info.bus could take
> > > the value usb_device.
> >
> > Nobody every replied to this point I made over a week ago.
>
> Sorry, been away from UK.
>
> > Is this a bug I uncovered in the spec or a misunderstanding on my
> > part?
>
> That part I don't know. Kay, what does block.storage_device do?

It's the pointer to the "device" that holds the data of the
block-device. Yes, it should go in to the list of allowed keys.

For the entire "bus"-namespace, that thing must be removed from HAL
soon. It just doesn't make sense to name something "bus", devices just
belong to a "subsystem", nothing else.

The subsystem "usb_device" is definitely not a "bus", even when the
current logic suggests and prints it like this. It's always the same
problem that shows up, when something tries to distinguish between
"physical" and "logical" devices - that can never be correct in todays
infrastructure. We should get rid of all "bus" and "physical"
properties in HAL sooner than later.

Thanks,
Kay