On 10/18/21 3:52 PM, Rémi Bernon wrote:
On 10/18/21 3:04 PM, Gerald Pfeifer wrote:
On Mon, 18 Oct 2021, Rémi Bernon wrote:
Thanks, I was wondering whether we should define this constant instead, or use HAS_PROPER_INPUT_HEADER to guard it too. Then I believe I found some references in some FreeBSD port files, but I probably missed that it needed a specific include.
Yes, includes can be ... painful ... at times. :-) (Ask me. :-) )
Is this header file also providing other defines like SYN_DROPPED, EVIOCGBIT, EV_ABS or BTN_PINKIE? Would it be better if we do as with SYN_DROPPED and define it if it's not?
I'll have to defer to the pros on the latter question.
Regarding the other #defines, it seems that header covers them as well (except for EVIOCGBIT which is in yet another one):
% egrep -r 'SYN_DROPPED|EVIOCGBIT|EV_ABS|BTN_PINKIE' /usr/include/ /usr/include/dev/evdev/input.h:#define EVIOCGBIT(ev,len) _IOC(IOC_OUT, EVDEV_IOC_MAGIC, 0x20 + (ev), len) /* get event bits */ /usr/include/dev/evdev/input-event-codes.h:#define EV_ABS 0x03 /usr/include/dev/evdev/input-event-codes.h:#define SYN_DROPPED 3 /usr/include/dev/evdev/input-event-codes.h:#define BTN_PINKIE 0x125
(I've not been using FreeBSD on the desktop or Wine on FreeBSD for a while, but happy to test build a patch or two or giving a try at answering questions.)
Gerald
I would be happy if we could remove the HAS_PROPER_INPUT_HEADER ifdefs, but then I'm also not completely sure if that was meant for some other platform than FreeBSD, which would have a partial linux/input.h header.
Anyone knows anything about it?
I sent a different patch to define BUS_BLUETOOTH if it's not, as we're using it for the hidraw backend too, even if linux/input.h is not detected.
I'm still interested if the evdev code can be made more portable, although I don't know FreeBSD at all and I've got no idea if it would work there.