Tue 13 Aug 2013 05:38:46 PM UTC, comment #1:
The problem is only with gnuplot. The issue is that the "ytickmode" property is set to "auto". In this case, Octave calculates ytick values, but so does gnuplot. Ideally, only Octave should calculate the ytick values and pass them down to gnuplot so that the display remains in sync with what Octave believes is being displayed.
This one-liner is a workaround you can use until this gets fixed. It grabs the current ytick values that Octave knows about, and then sets them on the current axis. This forces the "ytickmode" to "manual" and installs the ytick values that Octave knows about.
|
Mon 22 Jul 2013 02:17:10 PM UTC, original submission:
I can't really tell in which cases this happens, but I have an example that does return wrong tick values reliably for me:
There are ticks at 0.5, 1.5 etc. in the plot, so the expected and correct results would be:
Plotting 0:3 works fine and so does 0:5.
1:5 and 0:2:8 both have the same issue.
I'm running Octave 3.6.2 with Gnuplot 4.6 patchlevel 0 on Debian Wheezy. More specifically, I have these octave packages installed:
liboctave1 (3.6.2-5+deb7u1, amd64)
octave (3.6.2-5+deb7u1, amd64)
octave-common (3.6.2-5+deb7u1, all)
octave-htmldoc (3.6.2-5+deb7u1, all)
octave-info (3.6.2-5+deb7u1, all)
octave-io (1.0.19-1, amd64)
octave-java (1.2.8-6, amd64)
octave-miscellaneous (1.1.0-1, amd64)
octave-nan (2.5.5-2, amd64)
octave-octcdf (1.1.4-2, amd64)
octave-plot (1.1.0-2, all)
octave-statistics (1.1.3-1, all)
octave-strings (1.1.0-1, amd64)
|