[PATCH v7 wayland-protocols] Add the tablet protocol

[Peter Hutterer]

On Thu, Mar 31, 2016 at 02:17:02PM +0200, Carlos Garnacho wrote:
> Hey :),
> 
> On Thu, Mar 31, 2016 at 9:32 AM, Peter Hutterer
> <peter.hutterer at who-t.net> wrote:
> > sorry about the delay, this one slipped through
> >
> > On Fri, Mar 11, 2016 at 04:12:55PM -0800, Jason Gerecke wrote:
> >> On 03/08/2016 10:10 PM, Peter Hutterer wrote:
> >> > Signed-off-by: Peter Hutterer <peter.hutterer at who-t.net>
> >> > Reviewed-by: Daniel Stone <daniels at collabora.com>
> >> > Reviewed-by: Jonas Ådahl <jadahl at gmail.com>
> >> > ---
> >> > Changes to v6:
> >> > - a bunch of typos/grammar fixes
> >> > - clarified the "down" event on enter
> >> > - clarified the "up" event, specifically: the up event isn't sent when the
> >> >   tool leaves the input region but rather when the compositor deems it's up
> >> >
> >
> > [...]
> >
> >> > +
> >> > +   Clients should choose either value and avoid mixing degrees and
> >> > +   clicks. The compositor may accumulate values smaller than a logical
> >> > +   click and emulate click events when a certain threshold is met.
> >> > +   Thus, wl_tablet_tool.wheel events with non-zero clicks values may
> >> > +   have different degrees values.
> >> > +      </description>
> >> > +      <arg name="degrees" type="int" summary="The wheel delta in 0.01 of a degree"/>
> >> > +      <arg name="clicks" type="int" summary="The wheel delta in discrete clicks"/>
> >> > +    </event>
> >>
> >> I just noticed this while working on the Xwayland implementation -- is
> >> there a reason the angles (in tilt, rotation, and here in wheel) aren't
> >> just using the "fixed" type? If its a wart, it might be one to remove
> >> before too long now that v1 has made it upstream...
> >
> > hmm, good point, not sure why I didn't use wl_fixed other than "i didn't
> > think of it". It has finer granularity and provides the required range, so
> > it would be the better choice. This is something to fix in a new version
> > though :(
> >
> > Carlos, any yay/nay comments for the GTK side?
> 
> It indeed makes sense, I also missed this after we started using
> angles for those. I guess this just means v2 will get its own file to
> keep ahead with the incompatible change, so be it...

I'll do that, but after we merged the pad protocol, I think. That one is
backwards-compat.

Cheers,
   Peter