How to read xcb_input_fp3232_t and xcb_input_fp1616_t

Question

I have implemented a xinput2 listener to my window with the following code:

...
else if(eID == XCB_GE_GENERIC){ // Generic events = XINPUT
        auto* gev = (xcb_ge_generic_event_t*)event;

        if(gev->extension != extensionXInput->major_opcode){
            continue;
        }

        auto inputEvent = gev->event_type;
        if(inputEvent == XCB_INPUT_MOTION){
            auto *ev = (xcb_input_motion_event_t *)event;
            std::cout << ev->event_x << " - " << ev->event_y << std::endl;
            //BUTTON_ON_MOTION(ev);
        }

But when I read the event_x and event_y fields I get very high numbers.

How can I properly read these numbers?

Answer 1

fp1616 means "fixed point 16.16". This means that the first 16 bits represent the integer part of the number and the second 16 bits represent the fractional part.

See e.g. Fixed point vs Floating point number and https://en.wikipedia.org/wiki/Fixed-point_arithmetic

For example, the number 1.0 is represented as 1 << 16. The number 1.5 is 1.5 * (1 << 16) which is equivalent to 3 << 15.

Your calculation is "almost correct". You are using 0xffff as the factor, but the correct number is 0x10000. This should result in a tiny, tiny difference.

Answer 2

While I have no idea how this works internally, and what the gobble returned represents, the following code will convert them to readable numbers.

    double fp3232val(xcb_input_fp3232_t* val)
    {
        return val->integral + val->frac / (double)UINT_MAX;
    }
    
    double fp1616val(xcb_input_fp1616_t val)
    {
        return val / (double)UINT16_MAX;
    }