[PATCH v2 weston] libinput: normalize WL_CALIBRATION before passing it to libinput

[Pekka Paalanen]

On Wed, 10 Sep 2014 08:25:13 +0200
Jonas Ådahl <jadahl at gmail.com> wrote:

> On Wed, Sep 10, 2014 at 09:34:52AM +1000, Peter Hutterer wrote:
> > On Tue, Sep 09, 2014 at 10:42:26PM +0200, Jonas Ådahl wrote:
> > > On a side note, there is still a problem with the current format
> > > of WL_CALIBRATION and that is that it's translation part is still
> > > in pixels, and outputs don't necessarily always have the same
> > > resolution, which would make the calibration off when not using
> > > the resolution used when calibrating.
> > 
> > yes. without breaking existing use-cases we can't really change that
> > though.
> 
> We'd eventually fix it by deprecating and finally removing
> WL_CALIBRATION though. Probably not for 1.6 though since it's a bit
> to close to release for making such changes, but maybe for 1.7?

Personally I'd be ok deprecating it in Weston 1.7, and removing
completely in 1.8 or maybe already removing in 1.7.

Doing deprecation in 1.6 is a little late indeed, but OTOH I understood
WL_CALIBRATION is now broken until libinput 0.6.0?

Therefore anyone using Weston and relying on WL_CALIBRATION needs to
disable libinput in Weston 1.6. I assume libinput becomes mandatory in
Weston 1.7, so maybe removing WL_CALIBRATION at the same time
would be ok.

Do you think libinput 0.6.0 would be released before this Friday?
If not, libinput + WL_CALIBRATION will be broken in Weston 1.6. Hmm,
should we add a deprecation warning at runtime already to Weston 1.6 for
WL_CALIBRATION? That might actually also be a warning that the
non-libinput path is going to go away.


Thanks,
pq