> btoh{l,s}	-> bit endian to host
> ltoh{l,s}	-> little endian to host
> htob{l,s}	-> host to big endian
> htol{l,s}	-> host to little endian
>
> Maybe 'le' and 'be' instead of 'l' and 'b':
> betoh{l,s}, letoh{l,s}, htobe{l,s}, htole{l,s}. 

three things come to mind here.

1. this isn't, strictly speaking, a tech-kern thing.  these functions would
be useful in user mode also, and so the discussion should probably move
to the developers list.

2. l and s are misnomers and always were.  when we made the u_int32_t et al
types (dincha know those were bind throwoffs?) we just put the number of bits
into the names -- that's what we should do here.  "long" on an alpha is 64
bits.  l and s are vaxisms.

3. i like "l" and "b" better than "le" and "be" for what that's worth.