the design of CLP

Today I took a quick look at WSLT (from BEA) as a scripting solution (written in Jython) for WebLogic admin tasks.

For the last couple of days, I implemented a mini CLP for some WAS DynaCache cache. It is simple and quite extensible. I followed the DB2 CLP design.

This brings me an interesting thought on this issue.
Which way is better?

#1: design CLP commands only in a list of string seperated by white space characters and let
platform shell to take care of flow control, variable,name space if any.

#2: a full scripting language such as WebSphere 's wsadmin and Weblogic's WSLT. It provides flow control, variable, etc. That's why Jacl or Jython has been directly exposed to the end user (i.e. administrator).

Personally, i think #1 is better in that after a while, I don't need to look at the command reference any more. If I want to control it, always go back to my favorite shell. Secondly, for anyone who don't know Tcl or Jython (as most admin from Perl or Shell) syntax well, to look up the reference is pain in the butt.

DB2 CLP design gets this right: to keep simple things simple.

Victor