(I forgot to cc -bugs...)
Mike Landis wrote:
>
> Two things strike me as odd about that...
> 1) What's the logic behind the owner of a table not automatically
> getting a readonly privilege like SELECT?
Owner always has select on a table they have created.
> 2) I think it would be more logical to refuse to return a NULL result
> from PQexec() than to return a result with the proper number of tuples
> and columns, but inaccurate contents.
COUNT always returns a number (never null). Recall you are select from
the information schema table.
>
> It's also strange that Linux and Vista would produce different answers.
I suspect they do not. Its all in the permissions.
Cheers
Mark