On Sun, 10 Nov 2013 16:09:21 +0200 Oskari Rauta <[email protected]> wrote:
> > Raspberry Pi version doesn't support currently udev properties, I once > modified evdev, so it accepted touchscreen calibration values from udev.. Really? I see the code reads ID_SEAT. How does that not work for ignoring a device? Granted, I have never had to play with seat designations. This also reminds me we still haven't ported the rpi backend to use the shared udev input code in udev-seat.c. Thanks, pq > But that wasn't helpful with my touchscreen;; So I didn't make that a patch.. > Complete ignoring of the device-- Well doing that through the udev > sounds good, but cursor-disabled.. That can't be in udev env vars.. I'll > think about what you said- Maybe I'll work on a patch later.. > > On 8.11.2013, at 11:41, Pekka Paalanen <[email protected]> wrote: > > > On Thu, 7 Nov 2013 17:46:34 +0200 > > Oskari Rauta <[email protected]> wrote: > > > >> Hi. > >> > >> I plan to use weston on Raspberry Pi with a touchscreen. I purchased > >> a touchscreen that sports a eGalax touch digitizer. It seems not to > >> be properly supported by Weston or.. Actually with nearly with any > >> software, except tslib. Although even tslib didn't support my > >> version(0x0100) before I patched it. EETI provides binary-only > >> drivers for x86(32&64), arm and mips, but I wasn't able to get the > >> driver to work - and when I checked out internet for reports, people > >> who had got those drivers to work, were un-satisfied.. > >> > >> I found a project opengalax, which I forked and modified to support > >> tslib instead of eGalax on serial. It works very good. As TSLIB > >> doesn't support multiple simultaneous touches, gestures, like > >> pinching - are not possible. Virtual input device created works great > >> (after tslib's calibration) with weston- but weston tryies to access > >> touchscreen through 2 devices, original and through virtual - I > >> wanted weston to ignore this real device, so I made a patch to > >> weston, that adds a new option to configuration: > >> > >> [input-method] > >> ignored=eGalax Inc. USB TouchController > >> > >> And if this is set, it compares it to device name, and skips it. If > >> nothing else, atleast there's one device not to monitor.. > > > > Hmm, maybe you could use udev properties somehow instead. Weston has a > > mechanism to get devices belonging to a seat, and if you mark the device > > to belong in a different seat in udev, Weston should ignore it. I can't > > give any specific instructions off hand, though. There might be also > > other udev features that the seat designation, perhaps. > > > >> Second thing.. As I am using a touchscreen.. I wanted to disable > >> mouse cursor. I patched weston some more to add option [shell] > >> cursor-disabled=true > >> > >> - if this is set, mouse cursor isn't drawn. > >> > >> My fork of opengalax is not yet publicly available, but patch to what > >> I described is included in this message. > > > > The mouse cursor sounds like your touchscreen is detected as a pointer > > device instead of a touchscreen. AFAIK, Weston does not draw cursors > > for touchscreens. You can check the detection result in the weston log. > > > > I suspect the virtual evdev device is not set up properly to act only > > as touch input. I guess it is (was?) common to have single-touch devices > > emulate a mouse instead of a touch device, so that you wouldn't need > > any changes in other software, so it might even be deliberate that your > > touch device is detected as a pointer device and gets a cursor. > > Needless to say, emulating a mouse like that is full of fail. So, > > checking how your virtual evdev device looks like is one thing. > > > > > > Thanks, > > pq > _______________________________________________ wayland-devel mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/wayland-devel
