Troubleshooting
This page covers common issues and practical fixes.
GLE Executable Not Found
Symptoms:
Saving to
.pdf,.png, or.epsfailsErrors mention missing
glecommand
Fixes:
Install GLE from the official release page.
Ensure
gleis available on your PATH.Verify with:
gle -info
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:
Save as
.glefirst to validate script generation:
fig.savefig('plot.gle')
Compile manually for better diagnostics:
gle plot.gle -d PDF
If GLE reports
invalid font name, set the font to a GLE font name; see “Available Fonts” in Configuration.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:
Adjust layout using
fig.subplots_adjust(...).Move legend with
ax.legend(loc='...').Use
ax.text(..., bbox={...})for readability.
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 towhich='both'with a warning.Grid style is tick style (
xticks lstyle/lwidth/color), soax.grid(True, color='gray40')recolours that axis’ ticks as well.