Fuzzy filling

[Boudewijn Rempt]

On Tue, 24 May 2005, Bart Coppens wrote:

> Now I noticed this a while ago, but forgot about it until today. The problem
> is that I believe the behaviour of the bltSelection functions of KisPainter
> is wrong. If you use a very fuzzy brush, and paint a selection, I would
> expect when I fill it with a color or a pattern, that the fuzziness of the
> selection would be seen in the filling. This would be consistent with the
> apidocs specification of these functions:
>   * A version of bitBlt that applies the destination selection mask
>   * first to the source device and
>   * only then blits. This means that the source device is permanently altered.
> But in the source, there is absolutely no alteration of the source device, or
> any applying of the selection whatsoever.
>
> Would it be wise to fix this now, or should I wait with this until after the
> freeze?

If you have a fix, please apply it. It's one of those places where I cut a corner
with the idea to make it work correctly later -- because the selection attribute
of the iterator was a bool at the time, iirc. Anyway, you're right, and a fix would
improve Krita a lot.

Boudewijn