Re: [Linuxwacom-discuss] Wacom Intuos3 / Debian Stretch / Tilt values wrong in relative mode

SourceForge Headquarters 1320 Columbia Street Suite 310 San Diego, CA 92101 +1 (858) 422-6466

On Tue, Jan 09, 2018 at 09:16:54AM +0100, Hanns Holger Rutz wrote:
> I mean -- to elaborate: the relative mode works perfectly on macOS, with
> the assumed operation: relative coordinates, absolute pressure and tilt.
> If an application uses tilt and pressure, it would have to assume
> absolute values. So even if there is only a "global" mode, that mode can
> only be interpreted as relating to coordinates.

comparing X where we have to have backwards compat to 1994-ish to macOS
where there's only one toolkit and everyone has to follow Apple's "my way or
the highway" approach is comparing apples to oranges. Instead, let's just
assume that we're not intentionally trying to make things not work. Also fun
fact: this commit was introduced 7+ years ago and afaict this is the first
time someone's complained. For all I know there is 7 years worth of
applications that can make use of the data, but honestly, we just
don't know.

> On 09/01/18 09:10, Hanns Holger Rutz wrote:
> > Hello,
> > 
> > thanks for the answers. But as I said, you cannot do _anything_ useful
> > with relative/integrated tilt because the values are meaningless, so I
> > can't imagine any application using those values. Or am I missing some
> > use case?
> > 
> > Would it be possible to fix this issue?

First of all, there is no concept of relative axes in XI 1.x, only a concept
of relative *mode*. It's shoe-horned in from the beginging. So when the
device is in relative mode, we just add/subtract from the current axis
value. doing a pen down, move right, pen up, reset to original position,
repeat will eventually reach the x axis maximum because we never reset. Same
for y.

The pressure axis itself doesnt really do anything in relative mode, it
still looks like an absolute axis. Which makes sense, because you
effectively always reset to 0.

The tilt axes are the ones that look broken though because we don't reset.
So if I Bring the tool into proximity at an angle, the initial value is
whatever we exited proximity with and then we add to that. That's....
broken. I wonder if the best solution here is to take an example from the
pressure and just send an extra event on prox out that resets tilt to 0/0.
It's a bad hack, but seems more sensible than the current solution. Jason?
Ping?

Cheers,
   Peter
> > 
> > 
> > On 09/01/18 04:34, Peter Hutterer wrote:
> >> On Mon, Jan 08, 2018 at 12:00:19PM -0800, Jason Gerecke wrote:
> >>> My apologies for the delay in replying --
> >>>
> >>> When the tablet is switched to relative mode, it seems that all axes
> >>> (not just X and Y) begin emitting relative data. This appears to be
> >>> intentional, however: commit 49d7635e19 introduced this behavior,
> >>> noting that "the [axis value] builds up to the maximum and stays
> >>> there, even when the pen is lifted off the tablet." It links to
> >>> https://sourceforge.net/p/linuxwacom/bugs/156/ which suggests leaving
> >>> the pressure/tilt/etc. values absolute and only making the X and Y
> >>> axes relative. Why this suggestion was not acted on, I'm not sure...
> >>> Perhaps applications assume(d) that all axes were absolute or
> >>> relative, and so having a device "partially relative" caused problems?
> >>
> >> I think this was the issue. In the XI 1.x protocol you only have a device
> >> mode, not a per-axis mode (that's XI2, but well, backwards compatibility
> >> guarantees we can never use it). So you can't have one axis as relative and
> >> one as absolute, no way to tell the client that's the case. And the mode is
> >> device-global (with no event, iirc).
> >>
> >> Cheers,
> >>    Peter
> >>
> >>> I'm pinging the person who made that change to see if they can
> >>> remember anything, but its been this way for an awfully long time...
> >>>
> >>> Jason
> >>> ---
> >>> Now instead of four in the eights place /
> >>> you’ve got three, ‘Cause you added one  /
> >>> (That is to say, eight) to the two,     /
> >>> But you can’t take seven from three,    /
> >>> So you look at the sixty-fours....
> >>>
> >>>
> >>> On Mon, Jan 8, 2018 at 2:14 AM, Hanns Holger Rutz <co...@sc...> wrote:
> >>>> ping...
> >>>>
> >>>> On 21/11/17 13:40, Hanns Holger Rutz wrote:
> >>>>> hi,
> >>>>>
> >>>>> I don't know if this is the right place to ask (if not, please kindly
> >>>>> point me to a better forum) -- I'm having problems with a Wacom Intuos3
> >>>>> on Debian Stretch; whenever I use relative coordinate mode, the tilt
> >>>>> values are fluctuating more or less randomly and not correctly
> >>>>> initialised when I approach the tablet with the stylus; it seems there
> >>>>> is a bug that tries to apply some "relative integration" to the tilt
> >>>>> values as well, which of course is wrong.
> >>>>>
> >>>>> Is this a known issue?
> >>>>>
> >>>>> Best, .h.h.
> >>>>>
> >>>>
> >>>> ------------------------------------------------------------------------------
> >>>> Check out the vibrant tech community on one of the world's most
> >>>> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> >>>> _______________________________________________
> >>>> Linuxwacom-discuss mailing list
> >>>> Lin...@li...
> >>>> https://lists.sourceforge.net/lists/listinfo/linuxwacom-discuss
> >>>
> > 
> > 
> > 
> > ------------------------------------------------------------------------------
> > Check out the vibrant tech community on one of the world's most
> > engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> > 
> > 
> > 
> > _______________________________________________
> > Linuxwacom-discuss mailing list
> > Lin...@li...
> > https://lists.sourceforge.net/lists/listinfo/linuxwacom-discuss
> > 
> 