Sorry about the delay... On Mon, 2008-04-14 at 00:07 -0700, Danek Duvall wrote: > > > This is doable for these examples, I guess. It is the expectation > > that each postinstall script is either eliminated by changing the > > code such that no post-install processing is needed or becomes an > > SMF service + an action that pings the service? > > Correct. If you find a case where that won't work, let us know, but we > think this model should cover anything that's not required either for > booting or really early in the boot sequence.
I agree, it should work, since it basically pushes the responsibility of postinstall tasks from the packaging system onto smf. It makes sense, since it's a well defined and controlled environment -- logs, timeouts, dependencies. Much better than what we have now. I'm not sure how preremove scripts can be substituted, but I can live without them for now. So I started thinking about how to implement an smf service that imports the gconf schemas. What I'm trying to achieve is that the service only does anything at system boot, if there are new schemas installed or updated since the service last imported them into gconf. Will the IPS action that pings my service set a property or something to let the service know that it has work to do? Thanks, Laca _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
