On Fri, May 3, 2013 at 10:28 AM, Bruce Momjian <bruce@momjian.us> wrote:
I have removed both pg_ctl idempotent-commit items from the TODO list:
<listitem> <para> Allow pg_ctl --idempotent to a 'success' return code if the requested start/stop action fails, but the cluster is already in the requested state (Peter Eisentraut) </para> </listitem>
<listitem> <para> Change pg_ctl to return an error code if start fails because the server is already running (Peter Eisentraut) </para> </listitem>
My question is do we really want the second item reverted?
Are those parts easily severable? They don't appear to be to me.
I think that if we had time to sort out the issue before beta, we could just as easily sort out both issues and so not revert either one of them.
Also, changing the long-standing behavior and not providing an option to neutralize that change for people who depended on it doesn't seem like a good idea to me.