Review and advice for Palm handheld rules

[Kevin R. Page]

On Thu, 2008-02-14 at 14:55 +0100, Frederic Crozat wrote:
> > 1) Flag matching USB devices as "pda"s, then have a single policy rule
> > matching on pda with "palm" as its pda.platform to set access_control
> >   - I found the discussion on this list about the pda namespace from a 
> >   couple of years back, but this doesn't seem to have been merged into
> >   the HAL spec, although it's used in 10-usb-pda.fdi?
> >   - the matching of USB IDs as pdas should be pushed upstream to HAL and
> >   merged into the current 10-usb-pda.fdi?
> >   - the VID/PIDs that HAL matches are those which pilot-link will 
> >   recognise when using libusb
> > 
> > 2) Match USB VID/PID directly in the access_control rules - don't set a
> > generic pda capability
> > 
> > I'm attaching the file for the first solution; the second solution and
> > more discussion can be found at:
> > https://bugzilla.redhat.com/show_bug.cgi?id=280251
> > (going from the end back to about comment 92)
> 
> For the record, independently, I wrote a FDI file for PalmOS devices which is available at http://svn.mandriva.com/cgi-bin/viewvc.cgi/packages/cooker/pilot-link/current/SOURCES/19-palm-acl-management.fdi?view=log
> 
> Could you merge the differences (if any) in your file ? (I think I added more comments about the various Clié models).

Yes, I'll can merge in the differences - thanks.

You're very much implementing as I described in (2) above. Do you
believe this to be better than (1)? If so, do you think the upstream
10-usb-pda.fdi is redundant now?

Obviously I could submit a patch to 10-usb-pda.fdi with the merged USB
IDs in, but this wouldn't be much use with method (2). Is PolicyKit
developed enough to start pushing access_control rules for specific
classes of devices upstream, or are these best left for individual
distributions?

For info, differences are:
- you have more commenting ;)
- I'm missing Clie 0x054c/0x000a
- I incorrectly have 0x0c88/0xa226 as 0x0c88/0x0226
- you're missing a dozen or so IDs I pulled from known_devices[] in:
  http://cvs.pilot-link.org/libpisock/usb.c?revision=1.56&view=markup

With these corrections (thanks!) my ID list exactly matches those in the
pilot-link source - I've double checked this time ;)

Regards,

kev.