GNOME Bugzilla – Bug 670655
Wacom 'touch' devices are initialized to absolute mode by default
Last modified: 2012-04-17 13:59:36 UTC
Wacom touchpads (e.g. on the Bamboo 2FG) are being initialized to absolute mode by g-s-d. The xf86-input-wacom driver initializes them to relative mode, which is the prefered mode of operation by most users. A check should be made durring initialization to see if the device in question is a 'touch' device, and to use relative mode if no preference has been set by the user.
Created attachment 208238 [details] [review] wacom: Force touchpads to use relative mode
Does that fix it for you?
Does the trick, though g-c-c overrides the setting the first time its opened since it sets the 'is-absolute' property.
Created attachment 208297 [details] [review] wacom: Make touch devices insensitve to KEY_IS_ABSOLUTE changes Pen devices are typically used in absolute mode, while touch devices are used in relative mode. However, the tracking mode of *both* currently depend on the value of KEY_IS_ABSOLUTE. If a preference is set for absolute or relative it will be applied to both devices. This patch has touch devices ignore changes to KEY_IS_ABSOLUTE. Ideally it would be nice to have this key available independently for both pen and touch devices at some point in the future.
Pushed. Any changes to the UI would need to be a separate bug. Chris Bagwell mentioned that he would want the pad and touch interfaces to show up as 2 separate items in the control panel. We would at least need to remove the tracking mode from the UI for the touch part. Attachment 208297 [details] pushed as 389276e - wacom: Make touch devices insensitve to KEY_IS_ABSOLUTE changes
This change forces all my tablet _touchscreens_ (eg, Thinkpad X61 tablet) into relative mode such that even xsetwacom can't fix it. REQUESTING REOPEN: renders touchscreen tablet computers useless with no good workaround.
(In reply to comment #6) > This change forces all my tablet _touchscreens_ (eg, Thinkpad X61 tablet) into > relative mode such that even xsetwacom can't fix it. > > REQUESTING REOPEN: renders touchscreen tablet computers useless with no good > workaround. A good "work-around" is actually fixing the bug, isn't it. Which we already did: http://git.gnome.org/browse/gnome-settings-daemon/commit/?id=037073ddfebbfdfed8ed104f923f0ac1fc896196
Thanks for pointing out the newer patch, rawhide apparently doesn't have it yet. I did inspect the latest source package before commenting. The bug isn't fixed because "built-in" detection is also broken (looking at the libwacom source, it appears to only work for USB tablets; these are all serial). So, my Thinkpad tablets still don't work and there's still no way to force it to work except to remove all the related wacom modules from g-s-d and cc1 so they don't constantly override xsetwacom with a bogus setup. That's an unrelated bug except for the 'no mercy, no appeal' nature of the interaction. libwacom/g-s-d are forcing a default, incorrect config when they fail to read state properly as opposed to leaving things alone (perhaps that's in error). Does 'built-in' detection being non-functional have a BZ number, or should I report?
(In reply to comment #8) > Thanks for pointing out the newer patch, rawhide apparently doesn't have it > yet. I did inspect the latest source package before commenting. > > The bug isn't fixed because "built-in" detection is also broken (looking at the > libwacom source, it appears to only work for USB tablets; these are all > serial). So, my Thinkpad tablets still don't work and there's still no way to > force it to work except to remove all the related wacom modules from g-s-d and > cc1 so they don't constantly override xsetwacom with a bogus setup. > > That's an unrelated bug except for the 'no mercy, no appeal' nature of the > interaction. libwacom/g-s-d are forcing a default, incorrect config when they > fail to read state properly as opposed to leaving things alone (perhaps that's > in error). Fail to read state? No, we're forcing what we think is the correct configuration. I'd rather a bug report because we got the detection or application of a default wrong rather than trying to debug something when we let the user shoot themselves in the foot. > Does 'built-in' detection being non-functional have a BZ number, or should I > report? No. And it (unfortunately) wouldn't because libwacom doesn't have a bug tracker. Trying sending a mail to the linuxwacom mailing-list, we should be able to get this working by adding BuiltIn=true to the definition file for that type of serial device (can it even be non-builtin? if it can, then we'll need to add more code to libwacom).
And you can use gmane to send a mail to the list, if you don't want to subscribe. http://dir.gmane.org/gmane.linux.drivers.wacom.devel