Does the Evtouch driver support the usb Elo 2216 resistive touch screen?

[Nate Moseman]

On 10/28/07, David Sharp <whereami at gmail.com> wrote:
>
> On 10/28/07, Nate Moseman <natemoseman at gmail.com> wrote:
> > On 10/27/07, Nate Moseman <natemoseman at gmail.com> wrote:
> > > I've been banging my head against this one for a while now, but I
> can't
> > figure out what I am suppose to be doing.
> > >
> > > I am using Debian Lenny/Sid with their Xorg packages. I've tried the
> > Elographics driver, but my thing is not serial; it's usb so I figured
> that
> > it's not suppose to be used with my device. I've tried hacking a
> 'elousb'
> > driver and was able to get it to compile and sorta get loaded in this
> > release... but I have absolutely know clue what I am doing. I can get
> the
> > module to load but in it still complains that the driver is
> non-existant.
> > >
> > > I got it from:
> > http://www.softcoded.net/eduard/elousb.html
> > >
> > > So I figured the evtouch seems to be the right driver.. but people
> only
> > talk about Elo Intellitouch devices work which are accuostic devices,
> mine
> > is a accutouch and is resistive so I figure it's different and I don't
> know
> > if it's supported. I can get it to respond to key presses and I can
> > sometimes get the cursor to wiggle back and forth at the bottom of the
> > screen or the top of the screen and once sort a middle of the screen.
> > > But I can't get it to move up and down any.
> > >
> > > When trying to run the calibration tool it'll respond to clicks, but
> it
> > doesn't seem to know that the cursor is moving around any. The minimal
> stays
> > at '0' and the maximum stays at '2000' always for both X and Y no matter
> how
> > much I move things around.
> > >
> > >
> >
> >
> >
> >
> >
> >
> > Ok.  PROBLEM SOLVED... Almost.
> >
> > I am still curious about the evtouch stuff. But in fact I got it to work
> > passably well through the JOYSTICK DRIVER INTERFACE?! Who in the hell
> would
> > think that would work. The only clue I had was the screwy
> *-event-joystick
> > symbolic link that was created. Otherwise I would of never guessed to
> try
> > the joystick driver.
>
> This makes sense if you think about how events are reported to the
> Linux input subsystem. Mice report relative motion events, while
> joysticks and touchpads report absolute position events. in addition
> to the available buttons, this is all the input subsystem knows about
> the device drivers that report events to it. Then it creates various
> event handlers (evdev, keybdev, mousedev, joydev) based partially on
> the type of axes the device drivers report. relative axes? mousedev.
> absolute axes? joydev. Everything gets an evdev handler as well.
>
> You can read all about this in the kernel documentation. see
> Documentation/input/input.txt



Cool. That makes it make a  bit more sense. I assumed that it was something
Elo did (like emulating a joystick to make driver development simple once
they moved away from a serial device).

Still things are a bit goofy from a usability standpoint. For example when I
log out of Gnome and GDM takes over it seems that it grabs the mouse pointer
and places it into the center of the screen. Then when you touch the screen
the mouse jumps away to some seemingly random point of the screen. Then you
can sort of 'recenter' it by touching the far edges of the screen (far left,
far right, very top, very bottom) and it's fine again.

Then you can also imagine that with things like Compiz 'spinning cube' were
you drag windows to the edge of the screen and the cube 'flips' you have a
similar issue.

Now I assume that evtouch would have a somewhat better way of dealing with
these situations, but I have no real idea.

At work we are making a few thousands of these machines. Maybe tens
thousands if people like them. Think of a sort of 'single app, touch screen
kiosk' sort of setup. Right now they use a nasty sort of proprietary Elo
driver that I don't know much about that has it's own issues.

For my purposes the joystick driver works good, but if it needs to be
'recalibrated' (which is what they'd call it) all the time then it's not
going to be good enough for them.

Any further information would be very kick-ass. Thanks. I expect that
input.txt is going to be helpful.

Oh, and for the record I figured out that the 3rd axis is generating mouse
button events... or something like that. It makes scrolling windows very
jerky, for example. Probably my fault on the jscal calibration, but the fix
I used is:

option "MapAxis3" "mode=none"

Maybe it would be usefull to disable buttons and use the axis? I donno. I
have to end up 'triple clicking' double-click things on the nautilus desktop
to open things up. (Not a big deal.)

again, Thanks.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.x.org/archives/xorg/attachments/20071029/7d45d842/attachment.html>