Guest section DW wrote:
> What do you mean by `traditional' here?
> Do you mean last year's Linux? Or are you thinking of other systems?
Last year's Linux :)
> [That is, if we compare with traditional Unix, we are fine today.
> But if we also compare with POSIX, there is a difference.
Yep, that's what I mean: we had it right with respect to POSIX,
so what is the advantage we gain by doing what breaks it, and
why would it be unacceptable to fix it again ?
> Alexander says: forget about that, POSIX doesnt know about mount.
> That is a point of view.]
Okay ... my interpretation: POSIX is a convention for programmers
who wish to write portable programs. On a POSIX-compliant system,
programs written for the interfaces described in POSIX can be
expected to behave as described there.
If the claim is that any system using mount becomes non-POSIX,
that seems consistent to me. Is this the statement we wish to
make ?
If not, it seems to me that the system should either (a) continue
to provide an environment in which programs written for POSIX
continue to behave as specified by POSIX, or (b) to avoid that
programs written for POSIX can enter an area of the system in
which POSIX semantics may not be available.
You can implement (b) in many ways, e.g. (1) by disallowing
mounting over busy subtrees, (2) by killing every process in such
a subtree at mount time, (3) by requiring the system
administrator who wishes to have a POSIX-compliant system to
mount only on top of non-busy directory trees, etc.
All this looks quite wasteful, considering that there appears to
be a trivial change that yields (a).
So, again: what is the problem with solution (a) ?
- Werner
-- _________________________________________________________________________ / Werner Almesberger, ICA, EPFL, CH werner.almesberger@ica.epfl.ch / /_IN_N_032__Tel_+41_21_693_6621__Fax_+41_21_693_6610_____________________/- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Mar 07 2000 - 21:00:22 EST