On Fri, Feb 21, 2003 at 03:39:55PM -0800, Daniel Rogers wrote: > Here is a copy op and tests for float and uint8 channel types as a > recursive unified context diff. Excellent. I will see about getting some of this into CVS. > I wrote support for uint16's but I didn't see any tests written for any > operations on uint16s. I don't have time write the test right now, so I > am posting what I have. In any case, the copying code is identical to > the other two cases, so uint16 copying _should_ work, since the tests > passed for float and uint8 types. > > There was one thing I was not sure of: > Can we assume that the destination will have the same number of color > channels as the source? Fade does, so my code does too, but it will > segfault if this is not true. Yes. I think that is safe for a copy op to assume. A color convert op couldnt assume that though. > Also, I didn't go over the testing code very well, except to make sure > that it compiled, didn't have properties associated with fade, and > passed the tests. I have also never written anything with GObject > before, so if someone could tell me whether I did it right, I would be > grateful. Ill take a look. Calvin
