Actually, now that I think about it, is your system little-endian?
I can never tell what the case would be with Macs.
Since RGBA is just defining what each byte is in memory, then using a literal might be an issue since all hex literals are written as big-endian, so you would need to keep in mind that the that the bytes at the end of the literal would end up first in memory if your system is little-endian.
The more I think about it, the more confident I am that this is just a mistake in the understanding of how the literals work.
If you do something like u32be(0x0000ffff), I think that should enforce big-endian enocding. But you need to make sure you transmute() it and not cast() it whenever you need it as u32.
Otherwise you’re just going to have to think about your hex literals in reverse.