Re: [Linuxwacom-devel] [PATCH 00/12] serial id fixes

[Peter H. <pet...@wh...>]

On Mon, May 09, 2011 at 08:36:31PM -0500, Chris Bagwell wrote:
> On Thu, May 5, 2011 at 9:07 PM, Peter Hutterer <pet...@wh...> wrote:
> >
> > This patch series fixes up the serial number property to contain useful
> > data. The four read-only values this property now contains (and updates) are
> > <tablet id>, <prev serial>, <prev type>, <current serial>
> >
> > if the last one is 0, no tool is in proximity. Prev type is the device type
> > (some magic hardware ID), not a type like stylus, eraser, etc.
> > I think adding curent type would be useful too, not sure why we only have
> > prev type. comments?
> >
> 
> I had noticed that and wondered why as well. But I have to guess a lot
> of behavior since I don't have tablet with serial #'s.
> 
> > If the tool goes out of proximity, the last value is updated to 0. To avoid
> > race condition (we can't update immediately when getting a new serial, see
> > 05/12), Alexia's patch 04/12 updates the values before retrieving the
> > property.
> >
> > 11/12 adds a new property to force-bind the serial number on this device. so
> > xsetwacom set <device> BindToSerial 234223 is now equivalent to Option
> > "Serial" "234223" in the config.
> 
> I like this part... but I guess I'd like to finish the thread about
> intended user interface.  This property may should auto-hotplug
> another device.  Or maybe its not needed at all if we have a property
> that auto-hotplugs a new device and assigns a serial # to it.

I included this part because I think it won't change. The behaviour behind
it may changed (the auto-hotplugging) but the property itself should remain
such that it indicates which serial the device is bound to.

I'm still very hesitant about specifying behaviour that goes beyond the
immediate expectation of what the property does.

> > Not sure about 12/12, but the casting from unsigned to int and back annoyed
> > me. I think this should work, at least I haven't seen any adverse effects.
> >
> 
> hmmm, now that you've posted it, I have to go see why it was working.
> Must have been casting things just right all over the place.

It should just work as long as you're careful about value modification.
Since the serial comes in from the HW and we never change it, we're pretty
much free to cast it to anything. The only negative side effects I can see
here is if we had things like (if serial < 0) for uints, but the compiler
should have warned about that.

> For whole series:
> 
> Reviewed-by: Chris Bagwell <ch...@cn...>
 
thanks.

Cheers,
  Peter

> Its your judgement if we should commit patch #11 or not.  I'm OK with
> it since with are not 1.0 yet and could help get wider testing.
> 
> Chris
> 
> > available for testing from
> > git://people.freedesktop.org/~whot/xf86-input-wacom.git serial-id-fixes
> >
> > Cheers,
> >  Peter
> >
> > ------------------------------------------------------------------------------
> > WhatsUp Gold - Download Free Network Management Software
> > The most intuitive, comprehensive, and cost-effective network
> > management toolset available today.  Delivers lowest initial
> > acquisition cost and overall TCO of any competing solution.
> > http://p.sf.net/sfu/whatsupgold-sd
> > _______________________________________________
> > Linuxwacom-devel mailing list
> > Lin...@li...
> > https://lists.sourceforge.net/lists/listinfo/linuxwacom-devel
> >
>