MAVLink support for the STorM32 Gimbal Controller: How to make it "compatible"?

[hamishwillee]

@olliw

as regards D, MAV_PROTOCOL_CAPABILITY_PARAM_FLOAT/UNION.
So it’s not just me who isn’t sure what the meaning is. Would be nice if it could be cleared up. For the moment, I guess I can take that as indication that I shouldn’t be concerned

I would! As long as you stick to INTS/Floats for PX4 and floats as ardupilot interprets them you should be fine for interacting with those two autopilots.

@olliw

as regards E
except of the somewhat unfortunate naming that it’s called MAV_CMD_REQUEST_AUTOPILOT_CAPABILITIES and not MAV_CMD_REQUEST_AUTOPILOT_VERSION, your answer makes fully sense, and is clarifying. I will do as you listed. Thx. It would be great if the docs could be updated.

I’ve got a PR in place following this discussion. Needs to be reviewed and approved, and I want to confirm with both ArduPilot and PX4 the same interpretation.

@olliw

Maybe you could unambiguously confirm/disconfirm with a simple yes/no, if possible: Does it mean that PROTOCOL_VERSION is no longer WIP/DONOTUSE, but in fact official now?

No, not until the PR goes in.
But I suspect that it is “official”, just that no one has removed the wip yet. It has been around for some years now.

@olliw

I’d like to point again to that ArduPilot has an AUTOPILOT_VERSION_REQUEST message (not command, message) in it’s dialect, which is in fact used by e.g. MissionPlanner to query for AUTOPILOT_VERSION, and MissionPlanner in fact kind of requires a component to respond to this request in order to work properly. As you claryfied, it now turns out that this message is tautologous to the official MAV_CMD_REQUEST_AUTOPILOT_CAPABILITIES, but since MissionPlanner relies on it, it’s quite a breach of the standard. I guess one should bring this to the attention of the ArduPilot/MissionPlanner folks.

We don’t know that their dialect message doesn’t predate the command. But yes, in either case it would be good for us all to use the same approach. I will add it for discussion.

@olliw

Also, if the behavior, that a GCS essentially requires a component to support AUTOPILOT_VERSION for proper functioning, is how it is supossed to be, then it should be said so in the specs.

Yes. I need to revisit the heartbeat protocol and document some more of this. If you create a bug against the page I will certainly do so. Even if you don’t I hope to do so, but a bug report helps reduce the chance things will fall off my list.