APM Planner (2.0.25-rc1-31-gd9bae3ff5-DEV) crashes when reading log files from Arducopter dev (3.6dev)

I am testing the most recent version of Arducopter off the master branch (3.6-dev) and noticed that my version of APM Planner crashes while reading the log files. Here are a few of the error messages before the crash:

[20180111 9:43:39.064 WARN ] - Corrupted log data found - float resolves to NAN - Graphing may not work as expected for data of type “POS”
[20180111 9:43:39.064 WARN ] - BinLogParser::parse - No values within data message
[20180111 9:43:39.065 WARN ] - Corrupted log data found - float resolves to NAN - Graphing may not work as expected for data of type “POS”
[20180111 9:43:39.065 WARN ] - BinLogParser::parse - No values within data message
[20180111 9:43:45.144 DEBUG] - BinLogParser::BinLogParser - DTOR
[20180111 9:43:45.145 DEBUG] - LogParserBase::LogParserBase - DTOR
[20180111 9:43:45.145 INFO ] - Plot Log loading took 10.193 seconds - 29097984 of 29097984 bytes used

Interesting–I don’t recognize the Git commit. Where did you get that version of APM Planner?

You built it for me here:

Ok, try this one instead (I know it says 32-bit but it is for 64-bit too.):

@rrr6399 it would be great if you could share your logfile as I need something to test. There is a known issue with the new logs regarding NaN in dataflash POS.RelOriginAlt. If you provide the log I will try to fix it.

1 Like

Here is one that caused even the most recent version posted above to crash:

Cool - Thank you :+1:
I tried to fix that issue in the last days and now I have logs to test whether the fix works.
So stay tuned :grinning:

Hi @Arne-W.

I’m trying to learn a little so I did some digging. Any chance this has to do with NaN in RelOrignAlt for POS?


Hi @carpy,
yes you got it right it was triggered by the NaNs which are used to show that the data is not valid. I created a fix for that is the last 2 days which will get in today.

Moreover I have enabled C++11 support for Linux and Windows (OSx was already C++11). I have tested that it works for Linux but I cannot build for Windows. it would be very cool if you could test if the windows build still works. I made a test branch in my repository which contains some C++11 elements and the adapted project file https://github.com/Arne-W/apm_planner/tree/test_c++11. If that works on windows I would push the adapted project file into the master.

@rrr6399 as you could read above we have a fix for this issue - the logs you provided can now be parsed without any problem. Thanks again for those logs.

1 Like

Will do tonight. Thanks Arne.

Perfect - thank you Dave :+1:

Looks great Arne–no compile issues. Also, the change tested good (not that you didn’t know that already).

I suppose we should think about an official release huh?

Thanks for all your help!


1 Like

Cool - was easier than I thought :smiley: thank you for testing! I will push the C++11 enhancement to master later. Regarding the next release I think we should wait until we have implemented the unit Information enhancement? I hope I can work on this in the next week.
Also we should talk to @billb to get access to the official place for storing new releases. You are doing all those builds and have to share them with dropbox - using an official place would be much better! Moreover when using the official place the users would get informed about a new release automatically.

1 Like