Beta Test / Technical

[alsobrsp]

Originally Posted by alsobrsp

I am not sure what to call this, bug or user error, but I have seen this behavior for some time. At first I thought it was an issue with my GPS, it is old and I didn't buy it. Now I think is is something with OCPN, again bug or user error.

Happy Thursday,

I think I am going to call this a bug. I can trace the issue to when I started using a network connection to get the NMEA stream.

The gps is connected to a server on our boat. I share the NMEA stream via kplex. OCPN is getting the NMEA via TCP data source. Both the VDR and the track are recorded from this stream.

Syd

[nohal]

Syd...
What is that NMEA.gpx file? NMEA feed converted to a track using GPSBabel? Could we see the original NMEA feed recording instead?

Pavel

[alsobrsp]

Originally Posted by nohal

What is that NMEA.gpx file? NMEA feed converted to a track using GPSBabel? Could we see the original NMEA feed recording instead?

The NMEA file is here:
http://www.syeather.net/files/ocpn/marco-gulf-2.nmea

[nohal]

Just from very briefly looking at it, the reason is pretty obvious:

$ECRMC,153407,A,2557.846,N,08143.402,W,0.115,118.780,050514,,,*36
$GPRMC,153407.00,A,2557.84680,N,08143.40240,W,0.120,114.96,050514,,,D*7B
$ECRMC,153408,A,2557.846,N,08143.402,W,0.120,114.960,050514,,,*33

I would filter the EC sentences out in this configuration.

Pavel

[alsobrsp]

Originally Posted by nohal

I would filter the EC sentences out in this configuration.

Odd as I am filtering EC messages in Kplex before sending to OCPN.

Let me check some things.

Syd

[alsobrsp]

Originally Posted by alsobrsp

Odd as I am filtering EC messages in Kplex before sending to OCPN.

Okay, I have confirmed that OCPN is NOT receiving EC messages from the GPS server.

This screen shot shows the NMEA debug window on the right and a telnet session on the left. Both OCPN and telnet are connected to the same TCP port on the GPS server 172.29.35.4:10111.



The EC messages in the VDR file are being generated by OCPN.

Here is the VDR NMEA file for this test run.
http://www.syeather.net/files/ocpn/test.nmea

[nohal]

Syd...
I still bet that you were somehow sending the EC* sentences back to OpenCPN while recording that track.
I did try really hard to reproduce your problem, but it simply is impossible using just the GP* sentences you recorded as input.
When using the whole feed, including the ECRMC, I get exactly what you posted, but it's an issue with wrong configuration, not OpenCPN's fault.
If you see the same behaviour even in a situation you documented in your last post, which seems correct to me, please record just the input stream and post it together with your exact connections configuration (the DataConnections line from the configuration file will do)

Pavel

[muttnik]

If it's the EC sentences that are causing the problem and these are only being produced by OpenCPN then presumably either they're erroneously being used internally when generated or they're looping back somehow. If they're looping back either it's via the route that it's supposed to come through and the filtering in kplex doesn't work as intended or it's via another route which wasn't intended / considered.

The telnet test would seem to eliminate the broken filtering via that input route but why not also filter that input in opencpn just to be sure? If the problem then goes away it would point the finger at the filtering in kplex and we can focus on that. If it stays then either kplex filtering *and* OpenCPN filtering is broken, or more likely it's something else.

What does your opencpn data connections config look like?

[juehoe]

Hi there,
I'm new to the forum and no programmer, so please be patient...
I'm running 3.3.1731 on debian and persistently get seg faults (also on previous version).
I've loaded charts from OpenSeaMap, ONC-HamburgHafen_1.kap, ONC-HamburgHafen_2.kap, ONC-ElbeHamburgBrunsbuettel_1.kap,ONC-ElbeHamburgBrunsbuettel_2.kap, ONC-ElbeHamburgBrunsbuettel_3.kap and others.
The seg fault always happens, when quilting between the various small charts.
I also always get a seg fault, when I run opencpn -unit_test_1.
I'l try to attache a stack trace and a screen shot.
Please let me know what else to do.
Juergen

[nohal]

Juergen...
Are you sure the charts are not corrupted? Can it be isolated to any particular chart? I can't reproduce.

Thanks

Pavel

[CarCode]

Juergen,

OpenSeaMap sind keine Seekarten, sie heißen nur so.
Außerdem hat die OpenCPN Beta-Version 3.3.1731 noch extrem viele Fehler, verwende besser zu Anfang die Version 3.2.2.

Hilfe gibt es übrigens unter:
https://www.segeln-forum.de/board1-r...rd129-opencpn/

Das andere deutsche OpenCPN Forum ist so gut wie tot.

Gerhard

[juehoe]

Originally Posted by nohal

Juergen...
Are you sure the charts are not corrupted? Can it be isolated to any particular chart? I can't reproduce.

Thanks

Pavel

Since I also suspected that it has to do with the charts, I already downloaded the charts a second time, which made no difference. The crash does not occur the first time I move from one chart to another. But when I go up and down the river Elbe a few times, because I want to draw and manipulate a route for example, it crashes reliably. The scale/magnification also seems to make a difference and it does not seem to crash if there is a larger scale background chart.
The "display behaviour" of ElbeHamburgBrunsbuettel_3 is also a bit strange, because it is only displayed if it is the only visible chart (scale 1:21300 on my system). If I move around and any other chart appears in the bottom bar, it disappears at once and the background is displayed. I can than force it to be displayed, if I select it in the bottom bar, but the next I go back and forth it will be the same.
Even it the chart is corrupt, I would consider this a serious bug of opencpn though.
Jürgen

[nohal]

Juergen...
Still no luck making it crash and I did really go up and down the river more than a few times.
Could you post your ~/.opencpn/opencpn.conf so I can have exactly the same settings? Maybe even those charts, but as they don't always cause the crash, it's unlikely they are really corrupted and my today's download should be the same.
Just to make sure:

nohal@knut:~/MyCharts/ONC-Elbe$ md5sum *
530e52bb220f0c00697e1bd2b170efbc  ONC-ElbeHamburgBrunsbuettel_1.kap
0452e8ee38c02569438731325c18fb0c  ONC-ElbeHamburgBrunsbuettel_2.kap
a9a8de3f4add94f8691099e6ecbf3dd9  ONC-ElbeHamburgBrunsbuettel_3.kap
6ab53a4b0c9004b3d49af32ffcc5f657  ONC-HamburgHafen_1.kap
19f2b93dc3813926a023c40543b0da5c  ONC-HamburgHafen_2.kap

Are those 5 named the only charts available? How much memory do you have? Anything else to make it crash really reliably?

Pavel

[juehoe]

Originally Posted by nohal

Juergen...
Still no luck making it crash and I did really go up and down the river more than a few times.
Could you post your ~/.opencpn/opencpn.conf so I can have exactly the same settings? Maybe even those charts, but as they don't always cause the crash, it's unlikely they are really corrupted and my today's download should be the same.
Just to make sure:

nohal@knut:~/MyCharts/ONC-Elbe$ md5sum *
530e52bb220f0c00697e1bd2b170efbc  ONC-ElbeHamburgBrunsbuettel_1.kap
0452e8ee38c02569438731325c18fb0c  ONC-ElbeHamburgBrunsbuettel_2.kap
a9a8de3f4add94f8691099e6ecbf3dd9  ONC-ElbeHamburgBrunsbuettel_3.kap
6ab53a4b0c9004b3d49af32ffcc5f657  ONC-HamburgHafen_1.kap
19f2b93dc3813926a023c40543b0da5c  ONC-HamburgHafen_2.kap

Are those 5 named the only charts available? How much memory do you have? Anything else to make it crash really reliably?

Pavel

Hi Pavel!
I have plenty other charts installed, devided into groups Nordsee and Ostsee. This is because I am planning a round trip from Kieler Foerde, up the east coast of Sweden, through Goeta-Kanal, down the west coast of Sweden. I also used it to document some of my previous toerns in the Nordsee - and to play with it :-).
When I run opencpn -unit_test_1 it crashes all the time, 4 times in a row now.
I attached 2 sreenshots which show the memory usage during the last run.
Also a fresh log from the 4 crashes, the .conf file and a backtrace from an older run.
Thanks for your effort,
Jürgen

[juehoe]

Originally Posted by nohal

Juergen...
Still no luck making it crash and I did really go up and down the river more than a few times.
Could you post your ~/.opencpn/opencpn.conf so I can have exactly the same settings? Maybe even those charts, but as they don't always cause the crash, it's unlikely they are really corrupted and my today's download should be the same.
Just to make sure:

nohal@knut:~/MyCharts/ONC-Elbe$ md5sum *
530e52bb220f0c00697e1bd2b170efbc  ONC-ElbeHamburgBrunsbuettel_1.kap
0452e8ee38c02569438731325c18fb0c  ONC-ElbeHamburgBrunsbuettel_2.kap
a9a8de3f4add94f8691099e6ecbf3dd9  ONC-ElbeHamburgBrunsbuettel_3.kap
6ab53a4b0c9004b3d49af32ffcc5f657  ONC-HamburgHafen_1.kap
19f2b93dc3813926a023c40543b0da5c  ONC-HamburgHafen_2.kap

Are those 5 named the only charts available? How much memory do you have? Anything else to make it crash really reliably?

Pavel

Pavel...
forgot to confirm the md5sum, they are identical!
Since I noticed from my own screenshots, that memory consunmption was highest before the crash on the unit test, I experimented with this a bit further. I opened the gnome system monitor and tried to crash opencpn the usual way (causing charts to be displayed and disappear and display again...), I noticed that my memory is eaten up!!! (screenshot) The crash occurs, when about 60% are used up. The crash frees the memory immediately. I think, this indicates a problem with opencpn rather than the system (up to date debian 7.5, kernel 3.2.0-4-686-pae, gnome 3.4.2), doesn't it?
Btw: I had to convert the conf and log files to pdf, because the forum uploader is very restrictive with file extensions and allowed file size. Is there a better way to get around this?
Jürgen