I know I said it before, but I think I've really fixed the user/initdb problem now in Postgres. Part of recent problems was the change to flat packages changed script references. But also, I had separated the binary parts from the support resources to reduce the duplication of resources, then the installer runs the initdb script before the resources are installed and it fails.

Enough excuses, I hope this is it. The installer should also repair the broken user from the previous installer.