Absolute mouse (touchscreen) HID support

[ladyada]

Good one to do when we do multimedia key support (i believe we just need to craft the report API)

[dhalbert]

I did a little research on this. It may be an issue finding report descriptors that will work on all the major platforms. Paul Stoffregen has implemented touchscreen HID support for Teensy. The report descriptor he developed I think appears to work on Windows and Mac without external drivers; it may not work on Linux (or at least Raspbian). Some background here and in other forum posts: https://forum.pjrc.com/threads/32331-USB-HID-Touchscreen-support-needed

[dhalbert]

Latent support in #776, but not turned on. Needs more testing: it appears touchscreen and gamepad on the same HID endpoints might conflict on at least some platforms (Linux). Might need to wait for dynamic HID descriptors. Also current report descriptor (emulating a digitizer with stylus), does not necessarily seem to do the right thing with existing built-in drivers on all platforms.

[dhalbert]

Now available via user-defined HID descriptors.

[ladyada]

woohoo 3 year old issue :D

[ATMakersBill]

While I understand how this solves the problem in theory, is there an example or doc that has the descriptor and hid code to process it?

[dhalbert]

Digitizer descriptor that worked on at least some systems:

https://github.com/adafruit/circuitpython/blob/6.3.x/tools/hid_report_descriptors.py#L341-L406

I don't know what Python library I was using with this, if any. I may have just seen whether the OS recognized it, as opposed to trying to send a report.

[Red-M]

For extra compatibility, these are the USB descriptors that QEMU/KVM use, https://github.com/qemu/qemu/blob/master/hw/usb/dev-hid.c#L486-L524

[bitboy85]

If anyone ends up here by google: I've created an example for absolute mouse here: https://gist.github.com/bitboy85/cdcd0e7e04082db414b5f1d23ab09005

Edit: Additional Info: I've testet on Linuxmint 20 and a dual screen setup. Coordinates 0,0 is the upper left corner of the left screen while 32767,32767 is the bottom right corner of the right screen. I've read somwhere that this behaviour on windows is different but i'm unable to test.
Note: There might be a small bug in the Linux USB handling for absolute mouse. I noticed the following: Set the pointer to lets say 1000, 1000. Works fine. Now move the pointer away with your normal mouse. Lets try to set the pointer again to 1000, 1000 and its not working. So it seems that two identical reports from an absolute mouse are not handled, even if the pointer is moved due to a different mechanic. Using different coordinates works fine again.