Re: should we include mount? (was Re: umount missing ?)
On Tue, Oct 05, 1999 at 07:23:29PM +0200, Norbert Nemec wrote:
> > > i have installed with the big tarball and cannot found the umount command
> > This would appear to be a good reason *not* to include a "mount"
> > command---it encourages people not to learn about translators (and if they
> > don't know about translators, they might as well be using a monolithic
> > kernel).
> Why would you want to force people into their luck? IMO, it would be most
> important to make it as easy as possible to switch to hurd. Once you have
> the people using it, those that are interested will find out about all the
> goodies.
Originally I was going to answer this with "SO?? You want us to have DIR
aliased to ls -l by default?", but I realised that a) it already is, and b)
I wasn't going to win any favours with that argument. :-)
The main reason I'm not entirely happy about having "mount" is that the
effects of doing a settrans under Hurd are very different from the effects
of doing a mount under a Unix clone; in a way, a settrans is like loading a
kernel driver, doing a mount and making a symlink all in one.
The point is not that we should not have a "mount", but that the
documentation should talk about "settrans" and clearly explain that "mount"
is just a wrapper around settrans.
> Perhaps there could be a separate package that offers a whole bunch
> of scripts for that purpose. I guess mount/umount is not the only point
> where the fundamental differences can be hidden by a simple script.
Such as other things done with settrans, like network configuration (IIRC; I
don't have Hurd installed right now).
On a completely unrelated matter, would it be possible to produce a
"usermode GNU Mach" like there's a usermode Linux kernel?
--
Adam Sampson
azz@gnu.org
Reply to: