ArduPilot EU Dev Call 2025-02-05

Attendees (max) 8

UTC0701

Andrew Would be nice if the uint16 macros were replaced with the standard pointer macros.

  • Testing it on real hardware would be nice.

UTC0703

Peter Weirdly enough it costs a bit of flash to remove the unused variable.

Merged!


UTC0705

Merged!


UTC0706

Randy Approved.
P The trim autotest seems to be flaking, I should fix it.


UTC0707

P Should make working with mission and fence items in the memory easier.
A The msg reference argument wasn’t being used at all?
P In one place only.
A Again, weirdly, we remove arguments and the flash space occupance increases.

  • Perhaps you can have get_item return “invalid result”. Then the new item_exists() could be avoided. Perhaps.

UTC0717

P to_str() returns an object which will not be destroyed, unless it was scoped.
A Ah, OK. Nice to find out the reason.

  • Approved.

UTC0719

Sid I think Andy has approved. What’s pending is Andrew’s request on the IOMCU protocol.

  • I created a branch for that, which can become a PR once this one is merged.

A Some comments are not marked unresolved yet.

[ Discussion on the ./waf tricks required to trigger two different builds for CubeRed ]

[ Discussion that the secondary MCU on the CubeRed will act as IOMCU, not a fully secondary flight controller ]

P Why not implement the IOMCU protocol as a serial protocol, and flash a normal FCU firmware instead?
S It’s kind of a weird protocol. We eventually want to have a PPP link between the two processors, and that’s where we want to get to.
P Seems a bit unconventional to try to send IO expand commands, over a serial, encapsulated over a PPP.
S The alternative is going to be a very bleeding edge protocol that we need to develop.
A I’m concerned that the bootloader version checking is wrong, it might overwrite existign bootloaders.
S Perhaps the version number is wrong. But the idea is to avoid re-flashing the same bootloader.
A It would be nice to be able to discern if we’re trying to flash the old or the modern bootloader.

  • My principle concern is about breaking old boards.

A I’m surprized that the DMA changes were so small. That’s fortunate.

  • What testing has been done for SBUS and RCIN? And other functionality of teh IOMCU?

S I have checked with RCIN, flown with CubeRed. Did a size compare and IO firmware didn’t change size.
A I checked too, and some IO firmwares did change size. Something like 8byts, so not large. Perhaps it’s best to do a binary comparison, not just the size.

  • But this is getting close.

UTC0751

Hasn’t been looked at in the last couple of weeks. We can’t progress this one today.


UTC0754

A Approved!


Demo of the static analyzer by Peter.
Discussion on one example issue in lua_bindings.cpp.


Discussion on a question of Henry’s, regarding a SITL plane getting locked in a QAssist situation.
Turns out it was a corner case, in conjunction with the fact that we limit the TECS pitch to ±3 deg during VTOL.