Quoth Liane Praza on Wed, Jan 09, 2008 at 03:02:26PM -0800: > >If we suspect this will be obsoleted, shouldn't that result in a lower > >stability level? Otherwise, aren't we saddling ourselves with > >supporting this for two releases? > > The functionality will need to still exist, right? I guess I'm just > missing the big support burden.
The part of refresh which notifies the restarter that configuration has changed will still be necessary, but the part that commits changes won't. Those semantics don't require that there be a way to _produce_ uncommitted changes, though, so when the current modification interfaces are replaced by transactional interfaces, there will be no way to produce uncommitted changes, and those semantics of refresh will be moot. So you're right. I did notice, however, that svcadm refresh and smf_refresh_instance() are Evolving = Uncommitted. So it seems a little strange for svccfg refresh to have a higher stability level, but not really a problem that I can see. David
