fxruby users - May 2005 - Extending fxruby with c++ operating on an FXImage?

Hello,

I''m currently doing a little app that works on images and corrects lens
distortion. It is fully written in c++ using fox (1.4), but I''m 
interested in converting it to ruby+fxruby. However, the c++ code that 
does the actual work on the images will need to stay c++, for 
performance reasons. It currently isn''t too speedy, so converting that 
part to ruby too will make it to slow to use.

The image processing actually works on an input FXImage object and 
outputs a new processed one. I''m wondering how easy it would be to 
create a ruby extension that interfaces with the processing code while 
still allowing the newly created FXImage''s to be used by fxruby (as I 
want to display the results obviously). Will a simple swig interface 
definition for a c++ function like

FXImage* processImage(const FXImage* img, <processing parameters>);

be enough or does fxruby do more tricks underwater that I would need to 
duplicate?

Thanks in advance,
Paul

On May 3, 2005, at 7:58 AM, Paul Melis wrote:
> Hello,
>
> I''m currently doing a little app that works on images and corrects
> lens distortion. It is fully written in c++ using fox (1.4), but
I''m
> interested in converting it to ruby+fxruby. However, the c++ code that 
> does the actual work on the images will need to stay c++, for 
> performance reasons. It currently isn''t too speedy, so converting
that
> part to ruby too will make it to slow to use.
>
> The image processing actually works on an input FXImage object and 
> outputs a new processed one. I''m wondering how easy it would be to
> create a ruby extension that interfaces with the processing code while 
> still allowing the newly created FXImage''s to be used by fxruby
(as I
> want to display the results obviously). Will a simple swig interface 
> definition for a c++ function like
>
> FXImage* processImage(const FXImage* img, <processing parameters>);
>
> be enough or does fxruby do more tricks underwater that I would need 
> to duplicate?
Unfortunately, there''s a lot more to it than that and it''s not
something that can be easily simplified at this point. What I am 
wondering, however, is you could compromise by still doing all of the 
compute-intensive image processing in your C++ extension code and then 
construct the new FXImage object back in the Ruby layer. I haven''t put 
as much thought into this as you have, obviously, but I''m thinking of 
something long these lines:

	memBuffer = inputImage.data
	colorArray = memBuffer.data # an array of FXColor values
	newColorArray = processImageData(colorArray, ... other parameters ...)
	newImage = FXImage.new(app, newColorArray, ...)

In in approach like this, your extension code (that implements the 
processImageData method) doesn''t know anything about FOX or FXRuby per 
se, it''s just manipulating a Ruby Array of color values (32-bit 
unsigned longs).

Hope this helps,

Lyle