Regression list time

[Adrian Page]

Boudewijn Rempt wrote:
> On Thursday 09 March 2006 21:42, Boudewijn Rempt wrote:
>> On Thursday 09 March 2006 21:13, Boudewijn Rempt wrote:
>>> * gradient on a selected area on cmyk8 doesn't get reduced to the
>>> selection (or crashes)
>> Update: not just gradient -- for cmyk8, and only cmyk8, the selection is
>> the bounding box of the selection.
> 
> Ah, Casper has arrived and has found the problem: we simply haven't 
> implemented the masking in the bitBlt for cmyk8 over composite ops -- and in 
> almost none of the other composite ops in the other colorspaces...

The masking is done in the prolog macros so all the colourspaces and ops 
that use them have masking, which is most of them...unless I'm 
misunderstanding. CMYK8 was neglected though, it's true, and maybe a few 
others.

Adrian