[WIP] Multi-device support

[goretkin]

#13

[ggoretkin-bdai]

@jtsiomb Is this what you had in mind? It would be nice to avoid basically doubling the number of events for the multi-device mode on the libspnav side.

[jtsiomb]

It seems a bit more complicated than it has to be. First of all I wasn't thinking about adding more event types, but rather modifying how existing events work, when the client selects multi-device mode. Another thing is that most of these changes are about passing an extra argument to all the event handling functions, but the device id could just as well be a field in the event, instead of an extra argument. Also a minor point, but is_client_multidev feels like a dumb accessor which only ever returns a client field, so we might as well just access the field directly.

libspnav and the stability of its ABI needs some thought. Since event structures are not opaque, there is no acceptable way to add extra fields. The period field which i had in mind and could be co-opted for this is only present in motion events, and I certainly wouldn't want to add a bunch of more event types and structures just for this... at least not unless absolutely necessary. I'm leaning towards handling the device id information internally, somehow associating it with each event in a secondary data structure, and adding a function for the application to retrieve it. Complicated, but probably the only way to have the extra information without breaking compatibility.

[goretkin]

but rather modifying how existing events work, when the client selects multi-device mode.

For example, on the spacenavd side:

struct event_button {
	int type;
	int press;
	int bnum;
};

so do you think there should be an additional field in that event?

[jtsiomb]

no, in spacenavd and in the protocol, for a discussion of what libspnav needs to do, see the second paragraph of my last message.