#1158 Grid shows in wrong color with cairolatex

[Michael Bölling]

With gnuplot 4.7 on windows (http://www.gnuplot.info/development/binaries/gp470-20120707-win32-mingw.exe) the grid in an heatmap shows up in red even when it was set to black.
When I use the attached black_grid_problem.plt file with gnuplot 4.6 I get a black grid (see black_grid_problem46.pdf) whereas in gnuplot 4.7 I get a red grid (see black_grid_problem47.pdf)
The grid is set explicite to black by
set grid front noxtics noytics x2tics y2tics lt 1 lc rgb "black"
Is the something I do wrong?

[Michael Bölling]

zip file containing plt file that generates a heatmap using epslatex. In gnuplot 4.6 the pdf has black grid whereas in gnuplot 4.7 deving tex and pdf file

[Michael Bölling]

And by the way, if I use epslatex instead of cairolatex the white aereas (0-values) end up as yellow ones. That is why I use cairolatex. And I have to use 4.7 as there was a bug in 4.6 that don't show the color palette.
I am using windows 64 bit.

[Ethan Merritt]

I have not found the source of this bug yet, but if you need an immediate work-around, use

set grid front noxtics noytics x2tics y2tics lt -1

rather than

set grid front noxtics noytics x2tics y2tics lt 1 lc rgb "black"

[Michael Bölling]

Thanks you very much for your quick fix, that really helps me.

[Ethan Merritt]

The grid color bug is now fixed in CVS for 4.6 and 4.7

The failure of the postscript terminal to map a palette color that is zero-width on the cb scale is a separate bug. Most other terminals handle it correctly.

[Ethan Merritt]

Update/work-around:

I think the PostScript oddity is due to the colorbar axis precision being treated as 1 part in 1024. Your test script works if the range [-1.e-9 : 1.e-9] is replaced by [-1.e-3:1.e-3] to stay within this limited precision. I don't see a simple was to relax this restriction.