gnuplot v5 crashes when plotting waterfall.
Modify gp file or make gnuplot command customisable?
Hi @drid – can you post more details? It would really help track down what’s going on if you could post the error message you see. Does this happen with a particular observation? Are you able to share the data that caused the crash?
There’s an earlier thread about Gnuplot using too much memory on a Raspberry Pi; the error message shown is “out of memory for expanding curve points”. Does that match what you see, or is it something different?
Log record is “Segmentation fault”
When v5 is default:
gnuplot -e "inputfile='receiving_waterfall_7560_2017-10-12T17-51-57.dat'" -e "outfile='waterfall_7560_2017-10-12T17-51-57.png'" -e "height=1600" /usr/share/satnogs/scripts/satnogs_waterfall.gp
Segmentation fault
gnuplot4 -e "inputfile='receiving_waterfall_7560_2017-10-12T17-51-57.dat'" -e "outfile='waterfall_7560_2017-10-12T17-51-57.png'" -e "height=1600" /usr/share/satnogs/scripts/satnogs_waterfall.gp
Works
Problem is caused by final plot (last line)
plot inputfile every 2 using ($1*1e-3):2:3 binary matrix with image
It’s not happy with every 2
I used update-alternatives so gnuplot points to gnuplot4-nox and waterfall is generated but someone might want gnuplot5 as default gnuplot.
Will do more testing and post results
1 Like
So i plotted with and without every 2 with v4 and v5.
v4 with every 2 | without every 2 | v5 without every 2
i tested on v5 patchlevel 6 and it works with the every 2 on my system. If I recall well, segfault may occur if the binary file has not properly closed.
Hopefully, in a couple of days we will get rid of gnuplot 
https://gitlab.com/surligas/gr-satnogs/commits/waterfall