Richard L. Hamilton writes:
> [...]
> > Why is this different from the PCI-related tools?
>
> As someone who might use this, I see some differences about USB:
>
> * USB is _routinely_ hot-pluggable, typically by end users (although the
> ability to
> lock it down would certainly be desirable in some settings)
PCI also has hot-plug support, but that's not really what I was
referring to.
What I was referring to there was the mapping from ID numbers to
manufacturer and device information. That same sort of database is
kept by the PCI-related tools (such as scanpci), and is being used by
this new tool.
It doesn't make sense to me to have two closely-related features like
this (both USB and PCI are just bus technologies for I/O devices)
implemented separately and in radically different ways.
> * it doesn't necessarily take a kernel driver to support a new device, they
> could
> be handled using libusb (or its successor) instead
The same is roughly true of PCI. It's possible to poke the device
from user space as well (though perhaps not as useful to do so).
> However, I think you have a point, in that if there were libusb and generic
> driver equivalents
> for FireWire and (eventually) Bluetooth, similar notions would apply to them
> as well. Parallel
> tools for them, such that they could all be (in a GUI context) submenus of a
> menu of
> hot-pluggable devices with potentially online support for additional {info,
> drivers, ...}, would
> probably be ideal.
Right; that's the point exactly.
> The goals of looking a lot like something else and of doing things in the
> most sound manner
> possible are mutually exclusive _unless_ the former has a path to evolve into
> the latter,
> with the engagement of enough of a wider community that they too will start
> working more
> in terms of a bigger picture rather than their own individual project
> fiefdoms and glory.
Yes, that's the sort of issue that has me (and perhaps a few others)
concerned. Linux familiarity is nice, and obviously a good thing to
have, but what if there are places where we're already doing something
_better_? Does regression come as a non-negotiable part of the
package?
--
James Carlson, Solaris Networking <james.d.carlson at sun.com>
Sun Microsystems / 35 Network Drive 71.232W Vox +1 781 442 2084
MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677
