Troubleshooting

This page covers common issues and practical fixes.

GLE Executable Not Found

Symptoms:

  • Saving to .pdf, .png, or .eps fails

  • Errors mention missing gle command

Fixes:

  1. Install GLE from the official release page.

  2. Ensure gle is available on your PATH.

  3. Verify with:

gle -info
  1. If needed, provide an explicit path via GLECompiler(gle_path='...').

Save to Vector/Raster Format Fails

Symptoms:

  • fig.savefig('plot.pdf') raises a compile error

Fixes:

  1. Save as .gle first to validate script generation:

fig.savefig('plot.gle')
  1. Compile manually for better diagnostics:

gle plot.gle -d PDF
  1. If GLE reports invalid font name, set the font to a GLE font name; see “Available Fonts” in Configuration.

  2. If GLE reports include file not found, install GLE’s include library; see GLE’s include library.

Unexpected Data File Names

Symptoms:

  • Sidecar files use global names like data_12.dat

Fixes:

Name them with data_prefix on the figure or data_name on a series; see Configuration.

Text or Legend Overlaps

Symptoms:

  • Labels overlap data points or panel boundaries

Fixes:

  1. Adjust layout using fig.subplots_adjust(...).

  2. Move legend with ax.legend(loc='...').

  3. Use ax.text(..., bbox={...}) for readability.

Shared Axes Not Displaying All Labels

This is expected: shared axes label only the outer panels, as described in Figures and Subplots. Disable sharing if each panel must show all tick labels.

Lines Look More Angular Than They Used To

Symptoms:

  • A figure regenerated with gleplot 1.9.0 or later has visibly kinked lines where the old one had flowing curves, and peaks no longer overshoot

This is the fix, not a regression. Before 1.9.0, GLEGraphConfig.smooth_curves defaulted to True, so every line series was emitted with GLE’s smooth qualifier and rendered as a spline fitted near the data instead of a polyline through it. The new default draws the data. Any jaggedness you now see is in your points and was previously being smoothed away.

If the smoothed look is what you want (a guide to the eye, a densely sampled model curve), ask for it explicitly:

fig = glp.figure(graph=glp.GLEGraphConfig(smooth_curves=True))
# or, once per script, for every figure it creates:
glp.GlobalConfig.graph.smooth_curves = True

See the Curve Smoothing section of CONFIGURATION.md.

Grid Call Has No Visual Effect

ax.grid(...) draws a real grid (GLE xaxis grid, which stretches that axis’ ticks across the graph). Two things to know if it does not look the way matplotlib would:

  • A grid always covers the main ticks; which='minor' alone is not expressible in GLE and is normalized to which='both' with a warning.

  • Grid style is tick style (xticks lstyle/lwidth/color), so ax.grid(True, color='gray40') recolours that axis’ ticks as well.