Re: Persistance and boot scripts (was: Re: L4-HURD , POSIX, UNIX)

[Jonathan S. Shapiro]

On Mon, 2005-11-07 at 15:02 +0100, Marcus Brinkmann wrote:
> I definitely agree that there is a range of options between
> persistence and non-persistence.  Well, in fact there is exactly one
> third option, and that is orthogonal persistence, but of course you
> can vary the extent on which the persistence attribute applies.

Just to make sure that we are having the same conversation, I am aware
of only three foundational persistence mechanisms:

  1. Non-persistence
  2. Per-process persistence, which can be aggregated into sessions by
     persisting multiple processes as a group.
  3. Machine-wide persistence

Is there a fourth mechanism that I am missing?

The performance of per-process persistence is bad, and the complexity is
very high.

> > session management ... can be still considerably less invasive than
> > transparent persistence

Since transparent persistence is completely invisible to the
application, it is hard to imagine how this can be true.

For example: when the local system fails and restarts, a client with a
network connection sees that the connection has timed out. Either the
client already knows how to deal with this (because it is intrinsic to
*any* network connection) or the client is broken. In either case, the
presence of transparent persistence did not change the situation.

shap