Looking at the differences between
3.2.0.14 and earlier it looks like the use of EV
Can rather than CAR Can to enable battery
monitoring on the Leaf was the main trigger for
frequent event crashes (fixed 31/8 in commit
07f25e3).The
polling being turned on (when vehicle ON was
set) being more noticeable than at other times
as both buses have a lot of traffic - but
still not guaranteed to cause a crash as
reported by GlynHudson in issue #409. Based on
my own experience this issue could still occur
randomly but less frequently when the vehicle
was off/not polling.
The issue would still occur even if
enable canwrites were set to off as reported by
GlynHudson in #409, as canwrite logic was not
implemented in the Leaf code until 19/8 in
commit 41ac2f1.
A temporary workaround was also
implemented in the Leaf code to switch buses prior
to polling on 16/8 in commit
32db785.
The subsequent poller rework has
enabled the Leaf to direct polls to the correct
bus, allowing the PID polls originally defined
(but not working depending on the car state) to
function when the vehicle is on or off and
charging. Also the IncomingPollReply code was
tidied up and aligned with that used for the
Twizy. The vehicle poller extra frame request
delay was also originally fixed at 25ms
uneccessarily slowing down large replies such as
the battery cell voltage responses, now set to 0.
These changes mean the polling for the Leaf is now
working as it should.
The current edge version appears to be
stable and allows PID polling reliably to both
buses without causing crashes. I have been running
it for 24h with 6 start/stop events without
issue.
If others agree could 3.2.0.15 be
staged for EAP release to move Leaf users off
3.2.0.14? A note could be to 3.2.0.14 stating
there is a known stability issue on the Leaf and
to use 3.2.0.15 or similar.
GlynHudson and others on other Leaf
models can test to ensure we are back to a stable
release with the required polling functionality
working.
Kind regards
Derek