Ondrej Hurt <[EMAIL PROTECTED]> writes: > I suppose that's the point of L4. Features needed by Hurd must be > implemented in the mk abstraction layer of Hurd. Do you think there > are some areas that are not easy to be implemented efficiently on L4 > ?
I don't think we want that sort of abstraction layer in general; we already have about the right sets of abstractions. The mailing list archives of the relevant lists have already gone into the difficulties that exist with L4; I don't want to repeat them again. > If your goal is layer very similar to some subset of the Mach API > (i.e. Mach emulation in fact) then it is only matter of writing some > typedefs, constants, wrapper routines/macros, putting them into a > few header files and replacing all Mach stuff with them. It's both more than that and less than that.
