Re: [Linuxwacom-discuss] setting limits for devices that do not report correctly

[Ping C. <pin...@gm...>]

On Mon, Jan 28, 2013 at 7:46 AM, Kyle Evans <ke...@an...> wrote:

> On 01/25/2013 04:44 PM, Ping Cheng wrote:
>
>> On Fri, Jan 25, 2013 at 11:14 AM, Chris Bagwell <ch...@cn...
>> <mailto:ch...@cn...>> wrote:
>>     I think model # is shared by a lot of unrelated products and so I
>>     suspect we will have to live with 0-4096 and resulting mm issues and
>>     rely on calibration to recover from that.
>>
>>     Ping typically does the touchscreen driver work (I do not own a
>>     Wacom touchscreen) so maybe she can give some input into TPC93's and
>>     their HID report behavior.
>>
>>
>> X/Y min/max defined in the spec is the same as we read from HID report.
>> Calibration was a solution to correct minor offsets. That's why we added
>> calibration for touch in wacomcpl of linuxwacom package. Thanks to
>> technology improvement, newer touch pads work better out of box nowadays.
>>
>> Ping
>>
>
> That's just it, there is nothing minor about this. When you say offset, I
> think shift (e.g. the touch film was not aligned properly with the screen).
> This is a gross error. The scale is completely wrong and it is impossible
> to touch anything around the edge of the screen. Since most all operating
> systems have the important buttons around the edges, this is a fairly big
> out of box oops. It's not that the technology wasn't there, it's that the
> manufacturer cut corners.
>

You are right. It is a manufacturer issue. But, it is at least not this
manufacturer (I mean Wacom) that is responsible for the offset.

The reason I think it is a technology improvement is manufacturers have
improved/precised their assembly lines for the job. Don't you consider that
as  technology improvement?


> A user posted a video demonstrating the issue:
> http://qik.com/video/46638135
>
> In light of Favux's numbers, I see there needs to be a way to fine tune
> the calibration beyond having sensible limits. However, I do not think that
> calibration should be required to push the home button, or use the task bar.
>
> I've updated my patch to encompass Favux's sample values with the intent
> that there be no part of the screen that is untouchable by the user. I've
> tested it on my device and the error is less than the size of a finger, so
> I think it will be acceptable for most users. I can send it here or to the
> linux-input list if you'd like.


As discussed before, there are obviously more than one vendors using this
model. I doubt their assembly lines have the same offset. My concern is
something works for you may not work exactly for the others.

You can post your patch here or directly at linux-input. If linux-input
accepts your patch, we'll back port it here.

Thank you.

Ping