Buggy multi-touch layer

[baskerville]

While testing the diagonal spread gesture with Plato's upcoming Touch Events application on the Glo HD, I noticed that the recognition rate was low, and yet high on the individual swipe gestures. I then tried the diagonal spread gesture within Sketch and the underlying problem came to the surface (cf. first screenshot). My Aura ONE can perform this gesture without the jerky dance but has another multi-touch flaw that the Glo HD hasn't: if I hold the device in portrait mode, put one finger in the middle of the left half of the screen and swipe from the top to the bottom of the right half of the screen with another finger, then I get a strange unexpected output (cf. second screenshot). What happens in this particular case is that one of the touch points disappears when the angle between the two contact points reaches a certain angle. We then get coordinates for an imaginary contact point equidistant from the real ones. And then, when the angle gets out of the dangerous range, the lost touch point is brought back into existence.

I was wondering if similar bugs were present in the newer devices?

[NiLuJe]

Thankfully nothing as weird as that over here.

On the H2O, I can get a single jumpy line with the spread gesture, apparently stemming from my previous tap on "New" in Sketch's menu.

Otherwise, it works (despite the H2O's funky way of numbering slots).

No issues whatsoever on the Forma.

(The H2O uses a neonode zforce v2 IR grid, the Forma a cyttsp5 capacitive grid).

[NiLuJe]

(Assuming cyttsp, which I *think* holds for the Aura One)

Since I know you bypass Nickel, you might want to check dmesg after a fresh boot to see if the firmware has been loaded properly, I'm not *quite* sure who's responsible for that, but it's not entirely impossible some of it might be done by Nickel.

[baskerville]

Originally Posted by NiLuJe

(Assuming cyttsp, which I *think* holds for the Aura One)

Relevant ntx_hwconfig info:

TouchCtrl	ektf2132
TouchType	C-Type

Originally Posted by NiLuJe

Since I know you bypass Nickel, you might want to check dmesg after a fresh boot to see if the firmware has been loaded properly

What should I be looking for?

If I remember correctly, I had noticed the same jumpy lines using Nickel's Sketch Pad.

[baskerville]

I just tried within Sketch Pad: same results.

[NiLuJe]

Originally Posted by baskerville

Relevant ntx_hwconfig info:

TouchCtrl	ektf2132
TouchType	C-Type

Huh, no idea how that one behaves, sorry . I'm only vaguely familiar with the cyttsp ones because that's what was also used on (most, except the latest cheap-ass Basic, IIRC) Kindles, where the blob was also loaded as part of the boot process .

At least the different controller explains why there were so many issues with it compared to other devices ^^.