Re: local vs global object IDs

[Marcus Brinkmann]

On Wed, Jun 04, 2003 at 08:54:03PM +0200, Espen Skoglund wrote:
> Sure.  In the system I am talking about there exists only a handful of
> system services that are transient.

What I am missing here is why the difference between persistent and
transient tasks matters.  How do persistent tasks communicate with each
other?  Only indirectly via the system services?

If you make a distinction between system services and other tasks, and don't
care about communication between other tasks, then of course I can follow
you.  However, in such a system IPC can never be as secure and robust
between those other tasks - they have to resort to other system services
like pipes.

The way I am thinking about "secure IPC" is that it is a system service
available to all tasks, not only to the system services.  So maybe that is
the fundamental difference.

> Not if the the thread number can not be used by any client (an
> arbitrary client is usually not allowed to stop and restart a system
> service).

Well, in a multi-server system like the Hurd you communicate with other
servers directly, and some servers might crash and be restarted
automatically.  So I guess I never made clear how broad my requirements for
this are :)

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/