Re: L4-hurd Digest, Vol 6, Issue 12

[Marcus Brinkmann]

On Fri, May 16, 2003 at 11:24:40AM -0300, Fortes Marcelo wrote:
> > > 4)About Unix compatibility how much it will be
> > > Unix-Like OS running "SCO Unix" binaries for
> > exemple?

[...]

>   So GNU is an operating system to replace Unix 
> The question is not about proprietary or non
> proprietary software (AKA: FreeSoftware). it is about
> technic compability with Unix (SystemV or BSD
> flavours)
> becouse hundred of thousands who uses programs to
> average Unix and may port it to GNU as a Unix Clone
> with full compatibility. I used "SCO Unix" only as an
> exemple.

No, you asked about binary compatibility, and that is a different issue than
portability.  The Hurd is (module bugs and missing features) POSIX
compatibility, and there was in its development a particular concern about
compatibility with BSD, and today it's about compatibility with BSD and
GNU/Linux (in particular on out-of-spec behaviour).

>  Ok Now i understand that the auth server does the
> work of security policy(Thank you Marcus). But the
> Superuser is the main user with special privilegies
> with auth server?

Not within the auth server.  Each component must decide itself if it gives
any user a special privilege.  The standard behaviour of the Hurd servers
is to give the user ID 0 special privileges, though.  (But for auth, 0 is
just a number).
 
> 1) The actual port to new variants of L4 wroted in C++
> gives to entire system the desired performance
> expected of an usable Unix-like system? i know about
> the super fast IPC, and minimum of system calls into
> L4 and non cacheable synchronous messages, and another
> different  things between Mach and L4 that does Mach
> so slow, running with multi-servers. but performance
> is yet a point to use to see as a reasonable thoughts.
>   So Hurd on top on new L4 implementations can have a
> performance comparated to GNU/Linux, SCO-Unix or
> another Unix monolithic kernels for exemple?

We hope so, and in fact, there is some result in L4 research that shows that
a multi server system can perform faster than a monolithical system under
certain conditions.

It's not only the IPC system, it is also caching and VM behaviour.  The port
to L4 will be written with some of the worst issues in mind, but probably
not all.  Beside performance, stability and robustness (ie safety against
DoS attacks) is another major concern.
 
> 2) what about people that works with Hurd on Mach ?
> why they dont join to L4 group ant throuw Mach away?
> becouse they knows Mach problems.

There is not one group and another.  We already are the same people.
As in any large project, not everybody is interested in the same aspects of
the system, but the core is certainly identical.

However, we can not throw Mach away before we have something to replace it,
right?  And there is some hope that the Hurd itself can be written to be
compiled on both Mach and L4 systems (with most code being identical for
both systems).

Thanks,
Marcus

-- 
`Rhubarb is no Egyptian god.' GNU      http://www.gnu.org    address@hidden
Marcus Brinkmann              The Hurd http://www.gnu.org/software/hurd/
address@hidden
http://www.marcus-brinkmann.de/