Re: local vs global object IDs

On Tue, Jun 03, 2003 at 03:53:11PM +0200, Ludovic Courtès wrote:
> Note that global IDs may cause troubles if they are used within a
> persistent task as we discussed it some time ago [1]

Other requirements now seem to force us to use another level of indirection
anyway, so it might be that after all resistance you will get your local IDs
anyway.  The reason is that if a server dies, we want to release the
reference to its task ID quickly.  This would not be possible if the server
thread ID is stored and referenced directly by a thread, but only if there
is an intermediate capability name that points to the server and object id.

In fact, such "dead objects" are exactly what would then need to be detected
and caught transparently in the persistent case.  However, they also need to
be detected (with an error message) in the non-persistent case.

Thanks,
Marcus

--

-- 
`Rhubarb is no Egyptian god.' GNU      http://www.gnu.org    marcus <at> gnu.org
Marcus Brinkmann              The Hurd http://www.gnu.org/software/hurd/
Marcus.Brinkmann <at> ruhr-uni-bochum.de
http://www.marcus-brinkmann.de/