Pipeline runtime error in PlotIndvNetMet function

[gourie]

Hi,

Thanks for the great package!

I keep running into the following pipeline runtime error with a few consecutive runs:

Can you help me find out what causes this error and how to workaround/solve it?

Thanks
Joeri

[Timothysit]

Hi! Thanks for using the package.

I haven't seen this before, my current guess is that the particular figure is too big / somehow maybe the figure name is too long. I will recommend entering debug mode upon error (eg. using dbstop if error) and maybe trying to replace print with exportgraphics(fig, fname) / maybe send me a file with the variables during that step so I can try to troubleshoot it.

Other unlikely causes that I can think of:

• check the figure handle exists (ie. if for some reason the figure window was closed), you can check that by running isgraphics(figHandle, 'figure') just before pipelineSaveFig is called. But this is unlikely because it seems earlier figures were saved
• check the figure path exists (ie. check figPath input to pipelineSaveFig)

I am also happy to troubleshoot it if you are willing to send me an example file that creates this error; maybe you know which specific file is causing this error and if so I can try to just start running from step 4A.

[gourie]

Hi Timothy

Thanks for your prompt reply.

I have been reproducing the error consistently a number of times using the debug mode.
I can confirm that the figure handle is there and the path exists, but exportgraphics() does not work either.

I used the following code snippet to generate a debug file. Can you share your email address to send this over?
If this doesn't work, I would be happy to do a quick, joint remote debugging session if that works for you.

Best
Joeri

[Timothysit]

Yup, my email is tim.sit.18@ucl.ac.uk

[WR1171]

I think I got this exact error running MEA-NAP on macOS Apple Silicon (MATLAB ARM64 native build). Happened in pipelineSaveFig.m at line 76 where print(figHandle, '-dpng', '-r300', figFileName) fails on complex network plots.

MATLAB's print function with the -dpng renderer can choke on figures with many graphic elements.

If your issue is the same this is what I did to get it to work:

Wrap the print call in a try-catch with a saveas fallback in Functions/pipelineSaveFig.m (lines 72-77). Replace:

if strcmp(renderer, 'opengl')
fprintf('Using opengl renderer \n')
print(figHandle, '-dpng', '-r300', '-opengl', figFileName);
else
print(figHandle, '-dpng', '-r300', figFileName);
end

With:

try
if strcmp(renderer, 'opengl')
fprintf('Using opengl renderer \n')
print(figHandle, '-dpng', '-r300', '-opengl', figFileName);
else
print(figHandle, '-dpng', '-r300', figFileName);
end
catch ME
warning('PNG export failed for %s: %s. Trying saveas fallback...', figFileName, ME.message);
try
saveas(figHandle, figFileName);
catch ME2
warning('Fallback also failed for %s: %s. Skipping figure save.', figFileName, ME2.message);
end
end

This way the pipeline continues running even if a particular figure can't be exported. The saveas fallback often succeeds where print fails, and if both fail, it skips the figure instead of crashing the entire analysis.

Alternative workaround: Run this before starting the pipeline to force a different renderer:

set(0, 'DefaultFigureRenderer', 'painters');

[gourie]

Yup, my email is tim.sit.18@ucl.ac.uk

Dear Timothy
I have sent the data package via email
Thanks for your support!

[gourie]

I think I got this exact error running MEA-NAP on macOS Apple Silicon (MATLAB ARM64 native build). Happened in pipelineSaveFig.m at line 76 where print(figHandle, '-dpng', '-r300', figFileName) fails on complex network plots.

MATLAB's print function with the -dpng renderer can choke on figures with many graphic elements.

If your issue is the same this is what I did to get it to work:

Wrap the print call in a try-catch with a saveas fallback in Functions/pipelineSaveFig.m (lines 72-77). Replace:

if strcmp(renderer, 'opengl')
fprintf('Using opengl renderer \n')
print(figHandle, '-dpng', '-r300', '-opengl', figFileName);
else
print(figHandle, '-dpng', '-r300', figFileName);
end

With:

try
if strcmp(renderer, 'opengl')
fprintf('Using opengl renderer \n')
print(figHandle, '-dpng', '-r300', '-opengl', figFileName);
else
print(figHandle, '-dpng', '-r300', figFileName);
end
catch ME
warning('PNG export failed for %s: %s. Trying saveas fallback...', figFileName, ME.message);
try
saveas(figHandle, figFileName);
catch ME2
warning('Fallback also failed for %s: %s. Skipping figure save.', figFileName, ME2.message);
end
end

This way the pipeline continues running even if a particular figure can't be exported. The saveas fallback often succeeds where print fails, and if both fail, it skips the figure instead of crashing the entire analysis.

Alternative workaround: Run this before starting the pipeline to force a different renderer:

set(0, 'DefaultFigureRenderer', 'painters');

Thanks for the suggestion!
I agree that your proposal is safer in safeguarding the run process, but I would love to get all figures exported and not loose data.
From your experiments, does save as work for all figures that you catch the exception for? Alternatively, does changing the renderer realize exporting all figures without errors?

[Timothysit]

Hello, sorry for the late response! And thanks @WR1171 for offering a solution. I managed to save the plot from the debug file you provided. I will email you to follow up on the details of this particular error and then report back the general solution here.

In our experience in the SAND group, we haven't encountered this saving error.