Steve Simms wrote:

> Hmm, I think my PL/pgSQL is good enough to suggest a fix for that.  
> The attached patch does a check before looking for a primary key to 
> make sure that the table exists.
>
> I've given it some brief testing, and it seems to work.  Please 
> double-check before committing it.  I also made an attempt to follow 
> the coding style -- please explain where I went wrong so that I know 
> what to do with future patches.  :-)
>
That seems a good change; committed to HEAD...

(Jan, I think we were planning to add another committer to CVS???)