> All I know about it is that it doesn't even have an operating system
> identification in executables (needed for proper emulation modes or
> at least recognition for which OS a binary is) so that various OSes
> have kludged around this in incompatible ways.

Actually, the binutils crowd has a solution to this problem that
allows executables to be appropriately tagged after linking, and
they have a standard way of doing the tag that should be in binutils
by now, so it's really a non-problem, IMHO.

> Or would ELF (as in a fairy tale ;-) magically make shared libs much
> more efficient even on NetBSD platforms where they already exist?

No, it would magically mean we'd be using a standard object file
format that is widely supported and therefore portable.   It would
mean, e.g., that we could link against third-party Linux application
libraries and have some hope of winnage.   We're not talking about
doing this because ELF is "better" than a.out - we're talking about
doing it because of the strategic win.   It's also a superior object
module format on many architectures, but since we use it on those
architectures already, that's a non-issue.

			       _MelloN_