NACK. I think that perhaps the decision we made about this has been misinterpreted. The problem with the code in respawn() was that it was untested and essentially duplicated the code in spawn(). But we still have the idea that we want to use respawn() for updates. As such, we still want to keep the respawn() infrastructure - and fill it in as needed with update code. We may even decide to call this from a spec file on update eventually. This is similar to the upgrade scripts run on IPA upgrades. When I originally did the cleanup, I left the code that started logging for the respawn operation and the infrastructure in general. I think those things still need to be there. Ade On Wed, 2012-11-14 at 10:18 -0600, Endi Sukma Dewata wrote: