<thorpej@nas.nasa.gov> (Jason Thorpe) writes:

> Now, the SYSV way, while better in many respects, isn't perfect either.
> However, it *IS* a fairly standardized way of doing things (SVR4 uses it,
> as does Linux), and I think adopting a standardized way is going to be
> more beneficial to us in the long run than designing a Perfect scheme.

The semantics and implementation of runlevels are sufficently
different in the systems that use them (say Solaris, Irix, Tru64, and
various Linuxes) that claiming that there is a "standardized" thing we
can adhere to is not sufficentrly true to be useful.

        - Nathan