Re: changes to graphics.cc and subplot.m

[Ben Abbott]

On Jul 19, 2013, at 02:28 PM, bpabbott <address@hidden> wrote:

I composed a simple test script to compare Matlab and Octave

clf ()
subplot (3,1,1)
plot (rand (3))
title ({'one', 'two', 'three'})
xlabel xlabel
subplot (3,1,2)
plot (rand (3))
xlabel xlabel
subplot (3,1,3)
xlabel xlabel
plot (rand (3))
xlabel xlabel

The results for Octave and Matlab are at the links below.

<http://octave.1599824.n4.nabble.com/file/n4655822/FLTK.png>

<http://octave.1599824.n4.nabble.com/file/n4655822/Matlab.png>

The results are consistent, but are also more esthetic than your example.
Can you provide a short script which I can run through Matlab to see how it
behaves?

In any event, I do think the margins between the subplots are too generous
and can be reduced.

 

For example, we could modify subplot.m subfunction -> subplot_position() and set margins.top = margins.bottom = margins.row.  And perhaps also margins.left = margins.right = margins.column.  Thus before calculating the subplot width and height we would ...

margins.top = min (margins.row, margins.top);

margins.bottom = min (margins.row, margins.bottom);

margins.right = min (margins.column, margins.right);

margins.left = min (margins.column, margins.left);

And then calculate the width and height.

width = 1 - margins.left - margins.right - (cols-1)*margins.column;

width = width / cols;

height = 1 - margins.top - margins.bottom - (rows-1)*margins.row;

height = height / rows;

To avoid problems with gnuplot, maybe this should only be done for non-gnuplot toolkits.

Ben