Heh. The debate was even broader. There were arguments that success meant 1. puppet itself ran and did not crash 2. Puppet ran and applied a complete catalog successfully 3. Puppet ran, retrieved a catalog and applied it successfully (no cache) 4. Puppet ran, applied a catalog successfully, and all resources with providers were successful and even more weird edge cases like "no resources resolved != 1 providers" and such. I was and still am in camp 2.