After working out a stupid bug that was holding me back (I needed to
set the alpha value to see any output!), I'm debugging my loadimage()
implementation. I have some of the bit-shifts wrong. As image(6)
notes, the middle-endianness is confusing!
I'm mostly implementing it from the man pages rather than the
libmemdraw code, which is probably making it harder than it should be.
CMAP8 loading should be easy to implement in terms of cmap2rgb().
I haven't touched compressed images yet; /sys/src/libmemdraw/cload.c
was way too simple, so I feel like I'm missing something.