Re: a proper whiteout function

[Carl Sorensen]

On 4/28/15 8:57 AM, "Kieren MacMillan" <address@hidden>
wrote:

>Hi all,
>
>What would be involved in making a real stencil whiteout function which
>could be applicable to all grobs?
>
>For it to Do The Right Thing, I imagine it should:
>1. follow exactly the grob/glyph outline (i.e., not just a rectangle/box,
>as currently implemented);
>2. include a parameter to set the thickness of the outline; and
>3. include a parameter to determine whether the whiteout was filled
>throughout, or allowed ³holes² inside (as per the grob/glyph outline).

I think that doing all 3 things is virtually impossible, given my
understanding of the current state of stencil handling.  There are
multiple ways of creating stencils, and at this point there is no check
for how the stencil is created.  A stencil is a procedure for creating
some graphical output, along with the bounding box of the output.

I think that a stencil whiteout for simple stencils could be created by
combining two stencils:

1) the original stencil
2) a scaled version of the original stencil, colored white, and placed on
a lower layer

The amount of scaling would determine the thickness of the outline.

There would be no way of dealing properly with holes.

Composite stencils (e.g. lines of text/words) would not work with this
method.

To really do it correctly would require a redefinition of every stencil
procedure to include whiteout properties.  And I'm not even sure how some
of the stencils would have whiteouts defined (e.g. font glyphs).

Thanks,

Carl