> > I agree with you. What about having an "execd" (exec daemon)? The kernel
> > directly understands elf and/or a.out, and #!-scripts. If something else is
> > executed, it sends a message to execd. If it knows how, execd could either
> > execute the command, or instruct the kernel to do it (i.e. return the new
> > command line).
> >
> > (What do you think of this?
>
> I think all "non-native machine executables" (scripts, interpreted
> code, etc.) should have the full name of their interpreter in them.
> If they also had a unique (i.e. common) pattern for identification,
> it would be rather nice. Personally I don't like the presence of too
> many daemons, but if you can keep the kernel simple, why not?
I understand the desire to keep the kernel simple, but there is such a thing
as adding excessive user-space cruft in an attempt to avoid fairly simple
kernel features....
OTOH, if we can figure a way to have this daemon handle ELF, a.out and
script executables as well as other formats, I'd favor it. ;)
:) lilo