The check for UART port pointer is not sufficient
to know if the update() was possible or not. When
MarvelmindHedge construction failed, the hedge pointer
might be a nullptr and there are no checks to avoid
nullptr dereference.
The MarvelmindHedge structure had complex initialization
but was done in a C style, with intermixed initialization functions.
malloc() was changed to cleaner new operator. Given that the
file already contained new operator calls it didn't make
sense to have a mix. The files are cpp so C++ operators
are used.
Everybody defines the macro (or should be defining). Let smaller users
that don't have a filesystem to rather protect it with HAL_OS_POSIX_IO,
which also allow them to save more flash space.
/var/lib/<project> is more in line to where projects save their data.
Also, remove APM from dir name as it should had been done some time
ago.
This is only the default (and recommended for any new board). Existing
boards that want to adopt the default can remove their defines.
this logs critical messages while disarmed if we have logged any
messages while armed. This fixes issue #7010 where log files show the
incorrect mode if the log includes any portions where the user
disarmed. It makes analysing users logs very difficult. It also
affects parameters, so we don't always know the true parameter values
in logs from users.
fixes after peer review:
renamed disable to relaxed
make parameters private
default type to 1
enabled method checks backend created
servo range initialised in init method
contrain rate-desired during position control
use set-output-limit instead of set-safety-limit
release_length accepts rate
this fix ensures the output throttle is never in the opposite direction from the desired-speed
there is a possibility that this could lead to rougher throttle response when the vehicle is transitioning from forward to backwards motion because the throttle response will immediately go to zero when the desired speed cross over zero
we could call init on two devices with the same name, which caused
init_ok to be false. This could cause the SMBus battery to fail to
initialise
Thanks to Michael duBreuil for finding this!
