Re: NSColor patterns and NSImage draw...

[Fred Kiefer]

Uli Kusterer wrote:
>  that's what I get for not testing my beautiful progress indicator code 
> under GNUstep after writing it on OS X...
>
>  Are there any docs or other infos (subjects of interesting threads on a 
> mailing list?) on how the graphics stuff in GNUstep works? Because 
> without patterns and the NSImage drawXXX operations, there's not much 
> fun in progress indicators. And the sources say "// Need another PS 
> function for this".
>
>  So, what would be involved in writing one?
>
>  And while I'm here: Does anyone know the difference between the 
> drawAtPoint: and compositeToPoint: methods of NSImage? Couldn't one just 
> use composite to implement draw?

You should not confuse these two. The operations compositeXXX and 
disolveXXX are very similar and all of them should be handle by just one 
operation implementing [NSImage 
compositeToPoint:fromRect:operation:fraction:] (this is the other point 
where I put in the request for a new PS function). But as this requires 
a change in all backends, it always seemed to complicated to achive up 
to now. The code for this is already there in xlib, cairo and winlib. 
Changing ART as well should be easy, but it will be an incompatible 
change, you will need to have matching GUI and backend components. I am 
waiting for Alexander to declare that it is ok, to make such changes. 
Before the last GUI release I hoped that incompatible changes would be 
allowed after that release, but somehow this did not happen.

The drawAtPoint: and drawInRect: methods are different from this set in 
that they apply the current transformation to the image. That is, the 
image will be scaled and rotated, the later would not happen with the 
other operations. I would expect that only the art backend is prepared 
for such an operation. Only if we had full image transformation 
implemented in GUI could the compositeXXX methods be used to give us the 
rest of the operation.


As for pattern colours, Alexander and I discussed this some time ago. My 
idea was a cheap implementation on the GSC level (common layer of the 
backends), where we would add the image as an ivar on the GState and 
when performing a fill operation use this value, if available, instead 
of the colour values (Setting a colour would unset the image of course). 
For the image than the normal draw operations, plus the clippath would 
be used. The idea comes of course from the cameleon engine. Perhaps slow 
and not very elegant, but usable at once for all different backend. 
Alexander, if I remember correctly, suggested some interesting pattern 
composition only available for the art backend. In the end nothing came 
out of this discussion. Again any solution here would require a parallel 
change to GUI and all backends.

Cheers
Fred