[Gta04-owner] [Tinkerphones] Phosh/Librem5 system on a GTA04

Mark Müller markmueller86 at gmail.com
Sun Jan 6 21:02:17 CET 2019


Hi

On 2018/12/30 22:46, H. Nikolaus Schaller wrote:
> 
>> Am 30.12.2018 um 20:58 schrieb Andreas Kemnade <andreas at kemnade.info>:
>>
>> On Sun, 30 Dec 2018 20:06:34 +0100
>> "H. Nikolaus Schaller" <hns at goldelico.com> wrote:
>>
>>> Hi,
>>>
>>>> Am 30.12.2018 um 20:02 schrieb Andreas Kemnade <andreas at kemnade.info>:
>>>>
>>>> Hi,
>>>>
>>>> On Sun, 30 Dec 2018 19:20:33 +0100
>>>> Mark Müller <markmueller86 at gmail.com> wrote:
>>>>
>>>>> On 2018/12/30 0:07, Andreas Kemnade wrote:
>>>>>> adduser purism
>>>>>> is important.
>>>>>
>>>>> Thanks, that's right - I knew I forgot something... ;-)
>>>>>
>>>>>> I started a fresh debian buster using a patched
>>>>>> cdebootstrap/libinstaller
>>>>>> (see https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=904699)
>>>>>>
>>>>>> Now I am with a non-working touchscreen but screen output.
>>>>>
>>>>> Well, touchscreen is working, but there seems to be a bug in
>>>>> wlroots/rootston concerning how key presses are handled. I had a lengthy
>>>>> test yesterday and a phosh developer helped me a lot. So here's what we
>>>>> found so far:
>>>>>
>>>>> * "libinput debug-events" shows the touch events for the touch screen
>>>>> correctly
>>>>>
>>>>> * try starting /usr/lib/wlroots/touch - it's a stand alone touch demo
>>>>> for wlroots and needs a tty (f.ex. start it over ssh via "openvt -c 8 --
>>>>> /usr/lib/wlroots/touch" while rootston/phosh is not running)
>>>>>
>>>>> * you can start a GTK+3 app under X11 directly and you'll see touch
>>>>> input working; same app won't work correctly, if you start it directly
>>>>> with rootston
>>>>>
>>>>> * here's the issue i've found: the position of a touch event is read
>>>>> correctly, but only the first touch seems to be correctly recognized as
>>>>> a "pressed" event. Every following touch event seems to be categorized
>>>>> as a "mouse over".
>>>
>>> Well, this categorization is done on GUI level. The /dev/input events
>>> only know about mouse press/move/release.
>>>
>>> So it should not even be possible to generate mouse-over events, because
>>> there are no position reports between release and press.

You are right - there shouldn't be mouse-over events (not without any
clicked event at least). This issue is getting strager, the longer you
look at it. But perhaps that's just the case, because I've got no
experience in debugging touch related problems and not much experience
in Wayland as well ;-)

I looked a bit more into this issue by writing a small GTK+3 vala app
with three Gtk.Buttons and different signals connected to each button:
clicked, pressed, released, enter, leave.

When using Rootston as Wayland compositor, it seems there are signals
emitted for other GTK+3 UI objects than the one you touched. F.ex. you
tap button 2 and the following signals are emitted: button 1 entered,
button 1 pressed, button 1 left, button 1 released, button 2 left.
...and no, that's not a typo!

This doesn't happen if you use Weston as Wayland compositor. For people
interested in this issue, I filed a bug report upstream:
https://github.com/swaywm/wlroots/issues/1453

Best regards,
Mark


More information about the Gta04-owner mailing list