[changeset] clarification to pcolor doc-string

[Ben Abbott]

On Tuesday, September 23, 2008, at 10:24AM, "John W. Eaton" <jwe at bevo.che.wisc.edu> wrote:
>On 23-Sep-2008, Francesco Potorti` wrote:
>
>| >>>+## @code{shading} modifies an attribute determining the manner by which the
>| >>>+## face color of each cell is interpolated from the values of @var{c},
>| >>>+## and the visibility of the cells' edges. By default the attribute is
>| >>>+## "flat", which renders a single color for each cell's face with the edge
>| >>>+## visible.
>| >>
>| >>The text is correct, apart from the fact that this is not the "flat"
>| >>attribute, but "interp" or "faceted" (I can not tell which).
>| >
>| >The code for pcolor.m appears to be setting each patch to "flat"
>| >     set (tmp, "facecolor", "flat");
>| >Why is it you think that is not the case?
>| 
>| This is what I observe with Octave 3.0.1 and gnuplot 4.2.3 on Debian:
>| 
>| ## produce a plot with 16 (4x4) squares
>| octave> pcolor(magic(5))	# edges are visible (black lines)
>| octave> shading("flat")		# edges become invisible
>| octave> shading('interp')	# no change
>| octave> shading('faceted')	# edges become visible again
>| 
>| So apparently the default is "faceted".  I cannot find this as a
>| property of the plot though, so I can only deduce that by looking at the
>| plot.
>
>In the current development sources, I see:
>
>  octave:9> h = pcolor (magic (5))
>  h = -2.8337
>  octave:10> get (h, 'facecolor')
>  ans = flat
>  octave:11> get (h, 'edgecolor')
>  ans =
>
>     0   0   0
>
>  octave:12> shading ('flat')
>  octave:13> get (h, 'facecolor')
>  ans = flat
>  octave:14> get (h, 'edgecolor')
>  ans = none
>
>This appears to be compatible behavior.
>
>Does gnuplot directly support an interpolated image viewing mode?  If
>not, then I suppose someone would need to write the code in
>__go_draw_axes__.m to support this feature by taking the given data,
>doing the interpolation, and passing something else along to gnuplot.
>But don't wait for me to do it, I'm not that someone.  But I would
>consider patches.
>
>Thanks,
>
>jwe
>

For a compatible behavior whom ever might try to do this should examine how changing the number of rows in the colormap modifies the result.

>From what I infer, Matlab's surface object produces discrete patches to represent the interpolation mode. The discrete patchs are indicative of the discrete rows of the colormap.

So as not to confuse, learning c++ remains a future goal for me, so "I'm not that someone" either :-(

Ben