Bug report for Octave 2.1.73 in conjunction with gnuplot 4.2

[Ben Abbott]

On Wednesday, September 03, 2008, at 07:42AM, "Zidek, Herbert" <Herbert.Zidek at epcos.com> wrote:
>Description:
>-----------
>
>The following script fails with `fails_in_gnuplot42==true.'
>
>  The error is "line 0: invalid command."
>
>  It is working properly (but with an additional, undesired output,
>in case
>
>   fails_in_gnuplot42==false,
>   if an explicit legend is supplied, e.g., plot( x,y, 'r-; ;' ), or
>   if the number of curves is being reduced, e.g., 64 instead of 128.
>
>Repeat-By:
>---------
>
>Run the script with fails_in_gnuplot42==true.
>
>Script:
>------
>
>fails_in_gnuplot42 = true;
>
>x = 1:500;
>y = [];
>
>for i = 1:128
>  y = [ y; sin(x/(50+i)) ];
>end
>
>if ( fails_in_gnuplot42 )
>
>  __gnuplot_set__ terminal postscript;
>  __gnuplot_set__ output 'xx.ps';
>  plot( x,y, 'r-' );
>
>else
>
>  plot( x,y, 'r-' );
>  __gnuplot_set__ terminal postscript;
>  __gnuplot_set__ output 'xx.ps';
>  replot
>
>end
>--
>
>  Best regards,
>  Herbert Zidek

Hi Herbert,

The developers of have made significant changes to Octave in the last year. The functionality you are using has been obsoleted in favor of compatibility with Matlab.

The most recent release of Octave is 3.0.2.

Am I correct in inferring that your example works with an earlier version of gnuplot? I believe Octave 2.1.73 should work with gnuplot 4.0. Perhaps the  problem is a change in gnuplot (anyone?).

Ben