Sid: For legacy devices the old config will be run.
For new devices (that don’t support the old config) the new config will be run.
There’s a bitmask to force the new config for all devices that support it.
For 4.8 we can deprecate the option to use the legacy config.
Andrew: Will sending the unique ID of the GNSS RX be ok? It will be visible in the logs. Sid, can you make sure it’s ok for this key to not be secret? It is part of the licensing process after all. S: I think it’s used as a public, unique ID of your device. You still need to log into your account and link to it.
But I’ll remove it for now, I think this authentication process isn’t set in stone yet anyways.
UTC0732
New inertial sensor, ASM330.
Peter: Data looks consistent in the log. A: 3.5GHz data rate, that’s not too bad.
I think I see some delays in the data, sometimes up to 1ms.
Polling the temperature in every sample looks a bit excessie.
UTC0745
P: Do we want to return attitude wrt the autopilot mounting, or wrt the body frame? A: We don’t have a hard rule. Some parts of the code use one convention and others differ. P: Okay, then as far as I can tell, this PR now provide a non-rotated quaternion, whereas before it would provide a rotated (autopilot-frame) result.
UTC0820
Andy: The goal was to bring back boot to DFU. A: The issue is probably that the ST memory at boot has multiple parts. In some of them, the stack might be unusable during boot. The PX4 bootloader and the AP bootloader code might be different enough and use different parts of the memory. So there’s a chance we are trying to call C code (which needs a stack) very early and that would fail.
It may be safer to move that entry hook to after the initialization of the vectors and the stack pointers.
UTC0837
A: It’s probably cheaper to create atoff; it might incur less flash cost.
UTC0842
Andy: I would prefer to apply this only to Copter for now. It’s too much work to do it for all vehicles. Randy is OK with it. A: What happens if you get a GPS lock later? Andy: You set the origin once, so the origin won’t get updated. A: Why not set the origin upon arming for all flight modes then? It would make data visualization more consistent. P: We could at least put this behind an options bit.