Re: DiskDevice API v2.4 remarks

> * BPartition* BDiskDeviceJob::Partition(): There is no BPartition
> object, since we only have te partition ID. We would need to allocate 
> a
> complete BDiskDevice hierarchy to be able to return a BPartition. I'd
> prefer to just return the partition ID (perhaps rename the method to
> PartitionID()?). If the caller wants an object, there's
> BDiskDeviceRoster::GetPartitionWithID().

Okay.

> * A BDiskDeviceJob type is missing: B_DISK_DEVICE_JOB_SET_PARAMETERS.

Okay.

> * Would watching disk systems make sense? We'll probably have to do
> that in the kernel anyway, so it would take little to implement it.

Sure, why not. Something like this okay?:

B_DEVICE_REQUEST_DISK_SYSTEM = 0x0100,	// disk system addition/removal

B_DEVICE_DISK_SYSTEM_ADDED,			// disk system added
B_DEVICE_DISK_SYSTEM_REMOVED,		// disk system removed

> * BDiskDevice::CommitModifications(): I know, I know, I proposed it,
> but I don't think the template message is that good an idea. It would
> mean passing a flattened version of it to the kernel and then further
> to the registrar (if we indeed implement delivery of kernel
> notification messages there). For rather little benefit, I believe.

:-) Fine by me if we get rid of it.

> * BPartition::VisitEachDescendent(): My dictionary tells me it should
> read VisitEachDescendant. :-)

Yeah, I guess you're right. I have the hardest time with those words... 
:-)

-Tyler