why not fold the -unclobber patch into the standard db_ behavior and do without a new switch altogether? I can't think of anyone actually needing to clobber the variables in the calling code.
My 2 cents.
/Bart