Thanks for the response, William. One thing I wanted to follow up on:
> > The enumeration call was kept separate from the attributes call in > anticipation of other possible calls in future Caiman releases. The > attributes call is the most expensive call, since it involves issuing > ioctls that could hang on unresponsive devices and is done in a > separate thread, so I didn't want this to be bound to the > enumeration. > What I am not understanding here is whether there's a usage model for the data which makes this separation truly useful. It seems to me that the callers are only interested in the data once all the device data that the attributes call provides is available, so the separation is only theoretically useful. Perhaps I'm wrong about that, but it appears to be the case in the GUI designs and I'm having trouble constructing a case where it would be used otherwise. Do you have one in mind? Dave
