Okay, once again:
1) You can already start the updating process from a network share.
2) If your network share is unmapped, UpdateInstaller.exe will automatically map it for you.
3) If you prefer a direct call of Update.cmd or DoUpdate.cmd, you'll have to map your network share by "net use ..." before. This is reasonable in my opinion.
4) To avoid this technical limitation of cmd.exe, one could use your TCC shell. Much quicker and easier though, one could hack the registry as described in the Microsoft article mentioned above (
http://support.microsoft.com/kb/156276). Anyway, what would you earn? You'ld just save one "net use" call, but
5)
This would not bring back the missing autorecall feature! The automatic recall feature would still be unavailable, because the new, temporary user which is automatically logged on won't have access to the share, mapped or unmapped, persistent or not. That's the main restriction when you use a network share as source, that's the feature people cry for, it's a logical problem and therefore I suppose we won't get it solved in a satisfying way.
I hope this clarifies.
Regards
Torsten Wittrock