> Tom>  This seems strange.  When can the result here be NULL?  And what does
> Tom>  that mean?  It seems like this either ought to be impossible, or it
> Tom>  should be an error of some kind.
>
> Sami>  Yes this is incorrect. value_x_unop never returns null. If the
> Sami>  operator is not found it throws an error. I added a try catch
> Sami>  statement and tests for the failing case.
>
> Why a try/catch?  I would think that letting any error propagate would
> be more appropriate.
>

At first I wanted to avoid the error message changing. I expected an 
error message like "You cant use -> on a non-pointer" but it turns out 
you can :). So the catch is to allow that extension to continue working. 
Although personally I don't think these extensions should be allowed.

Test case added.