Indeed, I learned early to use a revision control system. The code itself hasn't changed, but there was something odd with this one set of data.

As it turns out, it was a faulty assumption. I assumed that the resident state data would be present. When it wasn't, $stash->set( 'resident', undef ) occurred inside the [% PERL %] block, giving me my strange error. I'm a little disappointed that TT didn't give me a better error message as to where this was happening.