On Tue, 2009-08-04 at 23:00 +0100, Chris Young wrote: > Looking at the GTK source, it appears that the colour primitives > (which are ABGR on all platforms) have some bit-shifting "magic" in > the platform code to make them work on big- and little-endian systems. > > Conversely, the bitmaps - which are RGBA on BE and ABGR on LE systems > - are simply treated as RGB data by the platform code with no sign of > endianess complications. > > If the GTK platform code can handle bitmaps that have the colour > components "reversed" on BE/LE with no special handling, then I would > say this is the correct way of doing things and it is the colour > primitives that should be modified, not the bitmaps.
I've given up on this issue. I can see no sense whatsoever in making colour component ordering depend on endianness. J.
