Re: [Linuxwacom-discuss] PLEASE TEST, udev rules for USB?Tablet PC

[Ron <ro...@de...>]

Hi,

Sorry for shouting in the subject, but if anyone who is able would like
to test these, please do, and let us know if you have any trouble.


On Tue, Oct 21, 2008 at 10:41:48AM -0700, Ping@LinuxWacom wrote:
> Did you receive my previous email?  I figured out the SUBSYSTEM usb and
> input issue.  For kernels 2.6.22 or later, SUBSYSTEM(input) is needed if it
> wasn't defined before (the wrong one worked because input was defined on my
> 2.6.26 system).  For other 2.6.x systems, SUBSYSTEM(usb) is needed.

I think we might be able to get away with not testing the subsystem at all
here.  In theory, if it passes the vendor id test, we know it's a usb tablet
of some flavour or another.  At least for the present.  So in theory that
test is redundant.

But logic in udevland seems to have it's own laws, so I can't be sure about
that until we've really seen it work.  There are at least two weird things
I cannot explain yet, except as a bug in udev, but I'll try and get some
more opinions on that too.  One is that a != comparison fails where the
inverse test using == (and a nasty GOTO hack) does succeed.  The other is
why I need to repeat the idVendor test to stop the penpartner (product 0000)
rule from matching on my graphire (as well as the correct graphire rule).

Even if these rules were applied more than once for different nodes in the
device path, as they appear to be, things still appear to be matching, that
by the documented rules of its logic (so far as I can fathom them), should
be entirely excluded.  So either I'm confused, it's broken, or some mix of
both is still in play here :)

Uh, what else ...  for the tabletpc devices, these rules should create just
input/tabletpc-stylus and input/tabletpc-touch for both 93 and 9a type devices.
If they won't ever be plugged together (and I imagine they shouldn't appear
to each other as tablets if they were?) then using these as the well known
names for all tablets will save having separate instructions required for all
of them.  If I'm wrong about that, we can divide them by type again, but this
seemed good.  I also went with stylus over pen, just because we already seem
to use the term stylus in a few places.  All that is still changeable right
now if other names are better, but I think that's about the last thing that
we need to finalise now, if it does indeed work...

Big Thanks to everyone who chimed in to help with this again!

  Ron