New Plugin NmeaConvertor

[yollivie]

I have sent an MP to rooiedirk thank you

[rgleason]

I was wrong about the Frontend, it is using an old version of TestPlugin.
https://github.com/RooieDirk/NmeaConverter_pi


This could be updated to the current version here TestPlugin and you would probably have a working version once compiled.

[Jodel]

Hi,
I would like to install the NmeaConvertor plugin. I am running OpenCpn 5.6.2 which was installed by Openplotter on my Raspberry Pi 4.


NmeaConverter is not on the list of available plugins in OpenCpn plugin tab.
I have also tried apt-get install and it says that the package is not available.


Any suggestions as to how I might install it?
Jodel

[rgleason]

Jodel,
We have updated it for Opencpn v571. There should be one for rpi in options > PI

[Bella Donna]

The latest NmeaConverter plugin .pkg I have been able to find is NmeaConverter-Plugin_0.2.pkg

I ran the installer, but it's not showing up in the list in Options-->Plugins. I see this in the log:

10:21:33.322 MESSAGE pluginmanager.cpp:1168 Checking plugin candidate: /Applications/OpenCPN.app/Contents/Plugins/nmeaconverter_pi/libnmeaconverter_pi.dylib
10:21:33.322 MESSAGE pluginmanager.cpp:2271 Plugin is compatible: true
10:21:33.322 MESSAGE pluginmanager.cpp:2364 PlugInManager: Loading PlugIn: /Applications/OpenCPN.app/Contents/Plugins/nmeaconverter_pi/libnmeaconverter_pi.dylib
10:21:33.364 ERROR dlunix.cpp:120 Failed to load shared library '/Applications/OpenCPN.app/Contents/Plugins/nmeaconverter_pi/libnmeaconverter_pi.dylib': dlopen(/Applications/OpenCPN.app/Contents/Plugins/nmeaconverter_pi/libnmeaconverter_pi.dylib, 0x0002): symbol not found in flat namespace (__ZN12wxWindowBase19SetVirtualSizeHintsEiiii)
10:21:33.364 MESSAGE pluginmanager.cpp:2428 PlugInManager: Cannot load library: /Applications/OpenCPN.app/Contents/Plugins/nmeaconverter_pi/libnmeaconverter_pi.dylib
10:21:33.364 MESSAGE pluginmanager.cpp:2432 Jailing: /Applications/OpenCPN.app/Contents/Plugins/nmeaconverter_pi/libnmeaconverter_pi.dylib

What does "jailing" mean?

Is there a more recent plugin version? If so, what is the link?

MacOS 12.6.7

[Hakan]

See post https://www.cruisersforum.com/forums...ml#post3785445

[RhythmDoctor]

I have been trying to hunt down a bug with NMEA converter for over 5 years. I use the plugin to calculate true wind angle from my boat's apparent wind and SOG data. It usually works OK until I start sailing with the wind behind me. Each time I'd try to figure out the problem I'd give up and just stop calculating true wind angle.

I have finally characterized the bug by creating NMEA scripts with artificial numbers (to eliminate noise from the data). Whenever true wind angle goes over 100 degrees, the returned value suddenly drops to 0. In other words, when my equation should return 105 degrees, it returns 5 degrees; if the value is 106 degrees, it returns 6 degrees, etc.

I still have more work to do to fully describe the offending variable. Before I take the time to further diagnose it and write it up, is the plugin still actively supported and someone willing to work on it? Do I post my description in Flyspray, or is there a different location for plugin bug reports?

[bdbcat]

RD...
NMEA-Converter is alive and well.
You may post issues here:
https://github.com/rgleason/NmeaConverter_pi


Looking forward to your feedback.
Thanks
Dave

[rgleason]

Nmeaconverter 1.3.3 is available from within Opencpn options > plugins (pluginmanager or PIM).
Please use that version for any testing.

[RhythmDoctor]

Originally Posted by rgleason

Nmeaconverter 1.3.3 is available from within Opencpn options > plugins (pluginmanager or PIM).
Please use that version for any testing.

All my current testing uses that version. However I have had a similar problem with all prior versions.

I'm still having some reproducibility problems in my tests. I'll post details as soon as I have something that I'm confident that others can replicate.

[RhythmDoctor]

I can't figure out where to put my bug report on Github (plus I don't have an account yet), so I'll try to post my first bug report example here. Please try it and see if your results are the same as mine.

To keep things simple for this first example I have cherry-picked some data at different wind angles. In all cases, I've set SOG=0, so that true wind speed = apparent wind speed. However, the way SOG is formatted causes the incorrect results.

First, set your dashboard plugin to show the "App & True Wind Angle" gauge.

Then, paste in the following sentence to NmeaConverter plugin to calculate true wind angle and true wind speed. The vector arithmetic is quite straightforward, but I can explain it in another post if you want to challenge me on it. Do not check the "calculate using degrees" box, since that feature has never worked properly for me:

$XXVWT,90-deg(atan(($??VWR3*cos(rad($??VWR1))-$??RMC7)/($??VWR3*sin(rad($??VWR1))))),$??VWR2,sqrt(sqr(($? ?VWR3*sin(rad($??VWR1))))+sqr(($??VWR3*cos(rad($?? VWR1))-$??RMC7))),$??VWR4,,,,

Now launch the VDR player (at the slowest speed to start out) using the attached file "Rotating wind direction SOG=000.0.txt". Note that my vYacht Seatalk-NMEA converter reports SOG in the xxx.x floating point format, which is what causes the problem. During playback, you can see the needle for the true wind angle following the needle for apparent wind angle as expected, until the needles reach 100 degrees. At that point, the true wind angle needle falls back by 100 degrees. This is the bizarre behavior that I have been sporadically trying to figure out for the past five years. (I gave up every time until now.)

However, as an academic exercise to help characterize the bug, I also created files where SOG is formatted as 00.0, 0.0, and 0, and they get different results. Formatting SOG as 0 or 0.0 works fine, but formatting it as 00.0 fails in all cases. You can open the associated files in VDR player and see for yourself.

Note that I have no control over how vYacht reports my SOG. It is always reported as xxx.x (or for my sailboat that never exceeds 9.9 knots, 00x.x).

Please delete the .pdf extension from these files when downloading. These massive 2.5 kb files were too large for CruisersForum to handle as .txt files.

There clearly seems to be something that causes NmeaConverter to blow up when it sees a floating point format of xxx.x or xx.x. It would be nice if NmeaConverter could be modified to successfully deal with the variable in any of these formats.

Note: In playing back these VDR scripts, your OpenCPN will go to my boat's home marina in the Chesapeake Bay. Come say hello sometime!

[RhythmDoctor]

Has anyone been able to replicate this problem? I am still seeing it, both with simulated data and with real-time data that I observed on my boat over the weekend. Works fine until true wind angle hits 100 degrees, then the calculator outputs TWA-100. This happens with boat sitting (RMC7=000.0) or moving (RMC7=00x.y).

[bdbcat]

RD....
Trying to reproduce, I have no direct experience with this plugin.
Following your steps on Win10, and looking at the NMEA debug log, I see unparsable garbage for the XXVWT sentence. Apparently I have not entered the formula correctly.
Pix attached. Please advise.


Thanks
Dave

[RhythmDoctor]

Hi Dave, thanks for checking.

First, your formula for true wind angle is returning values, but you took your screenshot too soon. The true wind angle had increased gradually up to 95.5 degrees, but the bug doesn't appear until its value goes over 100. If you had waited for 5 seconds longer I'd be able to tell whether you're experiencing the bug that I reported.

The "unparsable garbage for the XXVWT sentence" is in the True Wind Speed calculation. It's reporting the formula because it sees a defective variable. Your screenshot does not show me the entire formula, but I can see a typo where "$??VWR3" is incorrectly written as "$? ?VWR3". Eliminate that space between the two wildcard characters and the formula should return a value of 4.5 knots (unless there's another typo that you didn't capture in your screenshot).

So please delete the space, run it for a longer time, and hopefully you'll see the blue needle on the dashboard go back to zero when the orange needle hits 100, then both needles will continue moving with a 100 degree delta between them.

[bdbcat]

RD...
OK, I can reproduce. 2 random copy/paste errors in the formula.
Next time, better to place <code> tags around multi-lline formulae in posts. Simple copy/paste injected some errant spaces.


Digging in....


Thanks
Dave