Absolute coordinates for HID mouse #3020

awatterott · 2015-04-22T20:39:28Z

The define MOUSE_ABS_ENABLED enables absolute instead of relative coordinates for the HID mouse device. This can be used for a touch panel driver.

update

ffissore · 2015-04-23T06:51:13Z

@ArduinoBot build this please

ArduinoBot · 2015-04-23T07:41:18Z

Build successful. Please test this code using one of the following:
http://downloads.arduino.cc/javaide/pull_requests/arduino-PR-3020-BUILD-246-linux32.tar.xz
http://downloads.arduino.cc/javaide/pull_requests/arduino-PR-3020-BUILD-246-linux64.tar.xz
http://downloads.arduino.cc/javaide/pull_requests/arduino-PR-3020-BUILD-246-windows.zip
http://downloads.arduino.cc/javaide/pull_requests/arduino-PR-3020-BUILD-246-macosx.zip

awatterott · 2015-04-23T09:46:33Z

I have checked the test build and it is working.

damellis · 2015-04-23T14:51:12Z

hardware/arduino/avr/cores/arduino/USBAPI.h

@@ -110,11 +110,15 @@ class Mouse_
 	Mouse_(void);
 	void begin(void);
 	void end(void);
+#ifdef MOUSE_ABS_ENABLED
+	void moveAbs(unsigned int x, unsigned int y, signed char wheel, unsigned char buttons); // x and y have the range of 0 to 4095


For consistency with the Arduino API style guide (http://www.arduino.cc/en/Reference/APIStyleGuide), this should probably be "moveAbsolute()" not "moveAbs()".

Or moveTo() as Paul has named it.

PaulStoffregen · 2015-04-23T18:52:55Z

When I tried this on Teensy, a few unexpected issues came up.

The main one was absolute positioning is incompatible with nearly all games. Desktop usage works fine, but apparently most game software requires relative.

The other problem, still unresolved, was incompatibility with most Linux systems if both relative and absolute are in use. Windows and Mac seem to handle that fine, but there are problems with X11.

awatterott · 2015-04-23T19:20:13Z

We are using the mouse device with absolute positioning on our HDMI displays. It is a board with a TFP401A HDMI to RGB converter and ATmega32u4 as touch and backlight controller. The only problem we found is that some systems (e.g. OpenELEC) want the position data in screen coordinates and not in the range of 0...4095. On our display adapters we know the display resolution so it is no problem to send screen coordinates.
Maybe this is also the issue you have?
On a standard Raspbian system with LXDE desktop the mouse is working with the default coordinates 0...4095.

code cleanup

NicoHood · 2015-04-27T16:35:51Z

I can report the same issues. I also noticed that the absolute mouse doesnt work under windows server 2012.

And for some weird reasons it only worked if another consumer device is also in the same multireport. But I never found a reason. But I found a report combination that seemed to work fine on most systems.

@awatterott try this: https://github.com/NicoHood/HID

Edit: and as stated in many other usb related PRs: If you change the descriptors the device might not be recognized correct under Windows. And if you dont know why this happens and how this can be solved, people will be really confused. There are some solutions though. The whole USB core should be reworked, but noone is willing to do this sadly.

awatterott · 2015-04-27T17:05:55Z

@NicoHood Do you refer to my code or your code? Have you tried to test screen coordinates if it did not work?

NicoHood · 2015-04-27T17:08:56Z

I tested with my own core. USB can cause so many problems, it would be very critical to add this PR like this. They should think of different USB options the user can choose and keep the default setting to not kill the windows driver recognition.

awatterott · 2015-04-27T17:59:47Z

I found no real problems with the absolute mouse and coordinates in the range of 0 to 4095 (see my descriptor). This is also what Microchips AR1100 touch controllers use. I have tested it with the following Windows versions: XP, 7 and 8.1

NicoHood · 2015-04-27T18:59:59Z

@awatterott Andreas, the problem is, that if you change the USB HID Descriptor some things could be broken on several OS. I've tested a lot of USB stuff and Paul Stoffregen will aggree that very fast with a few changes things can go really wrong.

I could give you a lot of examples. But here is a simple one: I've added Media keys to a report. Everything seems to work pretty good. And someone else reported that it doesnt work at all under MAC. Even the keyboard was dead. Simply because a single line in the descriptor was wrong so the OS didnt accept it. Not sure if it was technically wrong because it worked under the other OS.

Well now you dont use Media keys. But still there is this driver problem. If you now switch back to the old core, Windows might throw errors. And also it might not work under all OS, maybe the 2nd Mouse breaks the whole function.

Its not that I dont support your idea. I am not even one from the Arduino team. I've created a lot of custom stuff with USB and the Arduino IDE if you've clicked the link above. Your idea is great and really useful! But simply adding this stuff to the Core HID files isnt a good idea. The user should be aware what he is doing, if he wants additional absolute mouse positioning. There are to many untested scenarios where bugs can occur with this addition. The Arduino team wont accept any USB PR I bet. They should redevelop the whole core and take some time to get it done better.

What you could do for your shield users are two things:
Provide a new Hardware folder for the sketchbook with your new core
Link to the HID Project that I've made which also has this function. Also feel free to test it.

awatterott · 2015-04-29T09:22:38Z

I see no problems, why the changes should not be included in the official core. The code is tested and the absolute mouse will only be activated, if the user does it via the define statement.

NicoHood · 2015-04-29T09:27:52Z

Your HID report contains Buttons and a scroll wheel. You can do 3 things:

Use them with the API
Remove them and use the normal mouse along with the abs mouse. The normal mouse does the buttons and optional the normal variable positioning. The abs mouse is then only used for X+Y.
Remove them from the report only so you have no click functions.

Also I see no reason why to use abs mouse without buttons (right/left click)?

awatterott · 2015-04-29T12:21:16Z

We use the mouse for a touchpanel and there is only one button... I have changed it, so that the absolute mouse API is compatible with the normal mouse API.

NicoHood · 2015-04-29T13:38:08Z

hardware/arduino/avr/cores/arduino/HID.cpp

+	m[3] = LSB(y);
+	m[4] = MSB(y);
+	m[5] = wheel;
+	HID_SendReport(1,m,6);


use sizeof(m) here

facchinm · 2015-07-10T10:14:19Z

@awatterott , this PR can be an extremely useful test case for PR #3304 .
I'd be glad if you could test the modular HID framework within your environment, trying both a modified version of Mouse library and the CompleteHID library

facchinm · 2015-09-21T12:43:55Z

Closed with #1803

awatterott · 2015-09-21T18:01:58Z

@facchinm I have tested the new USB core with our absolute mouse changes and everything is working. The patching without changing core files is really good.

NicoHood · 2015-09-22T09:45:10Z

Absolute Mouse is also available here:
https://github.com/NicoHood/HID/tree/dev_2_4

awatterott and others added 3 commits April 18, 2015 22:09

Merge pull request #1 from arduino/master

b08dcab

update

Merge pull request #2 from arduino/master

eadaa54

update

Added absolute coordinates for HID mouse

16f3d71

ffissore assigned cmaglie Apr 23, 2015

ffissore added the Component: Core Related to the code for the standard Arduino API label Apr 23, 2015

damellis reviewed Apr 23, 2015
View reviewed changes

awatterott added 2 commits April 23, 2015 21:32

renamed moveAbs() to moveAbsolute()

345cb09

code cleanup

Merge remote-tracking branch 'upstream/master'

16b1c5a

awatterott added 2 commits April 29, 2015 13:50

Merge remote-tracking branch 'upstream/master'

2ad5b84

absolute mouse API now compatible with normal relative mouse API

da96414

NicoHood reviewed Apr 29, 2015
View reviewed changes

removed hardcoded size value

5decc56

cmaglie added Component: USB Device Opposed to USB Host. Related to the USB subsystem (SerialUSB, HID, ...) feature request A request to make an enhancement (not a bug fix) Architecture: AVR Applies only to the AVR microcontrollers (Uno, etc.) labels Apr 30, 2015

PaulStoffregen force-pushed the master branch from 57dcf07 to 5a68e34 Compare May 19, 2015 17:51

facchinm closed this Sep 21, 2015

ffissore added this to the Release 1.6.6 milestone Sep 23, 2015

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Absolute coordinates for HID mouse #3020

Absolute coordinates for HID mouse #3020

awatterott commented Apr 22, 2015

ffissore commented Apr 23, 2015

ArduinoBot commented Apr 23, 2015

awatterott commented Apr 23, 2015

damellis Apr 23, 2015

NicoHood Apr 27, 2015

PaulStoffregen commented Apr 23, 2015

awatterott commented Apr 23, 2015

NicoHood commented Apr 27, 2015

awatterott commented Apr 27, 2015

NicoHood commented Apr 27, 2015

awatterott commented Apr 27, 2015

NicoHood commented Apr 27, 2015

awatterott commented Apr 29, 2015

NicoHood commented Apr 29, 2015

awatterott commented Apr 29, 2015

NicoHood Apr 29, 2015

facchinm commented Jul 10, 2015

facchinm commented Sep 21, 2015

awatterott commented Sep 21, 2015

NicoHood commented Sep 22, 2015

Absolute coordinates for HID mouse #3020

Absolute coordinates for HID mouse #3020

Conversation

awatterott commented Apr 22, 2015

ffissore commented Apr 23, 2015

ArduinoBot commented Apr 23, 2015

awatterott commented Apr 23, 2015

damellis Apr 23, 2015

Choose a reason for hiding this comment

NicoHood Apr 27, 2015

Choose a reason for hiding this comment

PaulStoffregen commented Apr 23, 2015

awatterott commented Apr 23, 2015

NicoHood commented Apr 27, 2015

awatterott commented Apr 27, 2015

NicoHood commented Apr 27, 2015

awatterott commented Apr 27, 2015

NicoHood commented Apr 27, 2015

awatterott commented Apr 29, 2015

NicoHood commented Apr 29, 2015

awatterott commented Apr 29, 2015

NicoHood Apr 29, 2015

Choose a reason for hiding this comment

facchinm commented Jul 10, 2015

facchinm commented Sep 21, 2015

awatterott commented Sep 21, 2015

NicoHood commented Sep 22, 2015