Re: figure background color when printing

[Ben Abbott]

On Apr 24, 2010, at 4:39 PM, Ben Abbott wrote:

> On Apr 24, 2010, at 3:23 PM, David Bateman wrote:
> 
>> Ben Abbott wrote:
>>> When the figure background color was added for the gnuplot backend, David 
>>> checked to ensure the color property value was numeric before setting it.
>>> 
>>>     http://hg.savannah.gnu.org/hgweb/octave/rev/1e38d9ed1e28
>>> 
>>> This is important because Matlab sets the color to "none" when printing. 
>>> Unfortunately, Octave only supports numeric values or the color.
>>> 
>>> This change is beyond my understanding. Can someone take a look?
>>> 
>>> Ben
>>> 
>> Ben,
>> 
>> I no longer have access to matlab so I can't check want it does... Are you 
>> saying that matlab support "none" for the figure property and that when 
>> printing it is set to "none" automatically? If so the change is relatively 
>> simple and we have to make the figure color property support radio values 
>> like the axes color property does in graphics.cc and then in the print.m 
>> function in an unwind/protect block store the figure color, set it to none 
>> and reset it to the initial value in the protect block. Do you want me to do 
>> this? Is the axes color property also set to none when printing?
>> 
>> BTW, what is the default axes color in matlab? It would make a lot of sense 
>> to me if it was "none" and would allow the bug #29060 to be closed. If it 
>> isn't "none" is it set to "none" for imagess so that contours can overlay 
>> images.. Can you tell me what matlab gives for
>> 
>> close all
>> reset(0)
>> plot(1:10)
>> drawnow()
>> get(gca(), 'color')
>> close all
>> imagesc(hilb(4))
>> drawnow()
>> get(gca(),'color')
>> 
>> Cheers
>> David
> 
> David,
> 
> I get "none" for each. 
> 
> If I try ...
> 
> set (gcf, 'color', [0 0 1])
> set (gca, 'color', [0 0 1])
> plot (0:10)
> print -dpng test.png
> print -dpdf test.pdf
> 
> In both cases I get a white figure background with an axes with a white 
> background.
> 
> The fix for print.m is trivial.
> 
> --- /Users/bpabbott/Development/mercurial/octave/scripts/plot/print.m 
> 2010-03-27 19:36:43.000000000 -0400
> +++ print.m   2010-04-24 12:19:11.000000000 -0400
> @@ -586,6 +586,7 @@
>     p.paperposition = get (gcf, "paperposition");
>     p.paperpositionmode = get (gcf, "paperpositionmode");
>     p.paperorientation = get (gcf, "paperorientation");
> +    p.color = get (gcf, "color");
>     if (p.papersize(1) > p.papersize(2))
>       paperorientation = "landscape";
>     else
> @@ -637,6 +638,7 @@
>     unwind_protect
>       set (gcf, "paperunits", "inches");
>       set (gcf, "units", "pixels");
> +      set (gcf, "color", "none");
>       restore_properties = true;
>       if ((! output_for_printer || is_eps_file) && ! doprint)
>         ## If not PDF or PostScript, and the result is not being sent to a 
> printer,
> 
> SInce you're implying that I may be able to handle the change to the c++ 
> sources, it must be trival ;-)
> 
> Is this all that is needed?
> 
> -      color_property color , color_values (1, 1, 1)
> +      color_property color , color_property (color_values (1, 1, 1), 
> radio_values ("none"))
> 
> In any event, feel free to push a changeset.
> 
> Ben

I've pushed the change to graphics.h.in and print.m

        http://hg.savannah.gnu.org/hgweb/octave/rev/5475da58d39f

        http://hg.savannah.gnu.org/hgweb/octave/rev/79410269cd21

Ben