Re: Shared m68k packages (Was: CVS commit: src/sys/arch/m68k/m68k)

[email protected] (John Klos) writes:

>Or perhaps even better, binaries which have both soft float and floating 
>point instructions which are chosen at run time would be good so we could 
>have one set of binaries for everyone.

You may be able to select some more complex math functions (like sqrt/sin)
at run-time, but for plain arithmetic this usually nullifies the "hard float"
advantage, see AmigaOS mathieeedoubbas.library. Most often people will
just use hard-float with a trap handler that emulates the instructions
if a FPU isn't available. Yes, that's slower than soft-float, but you
rarely use a FPU-less CPU for math anyway.