> -----Original Message-----
> From: Tom Lane
>
> Tatsuo Ishii <t-ishii@sra.co.jp> writes:
> >> Ooops. I guess libpq needs to supply a copy of this function?
>
> > Simply copying the function won't work since the way to know what
> > encoding is used for this session is different between backend and
> > frontend.
>
> Good point --- in fact, the encoding itself might be different between
> the backend and frontend. That seems to imply that "truncate to
> NAMEDATALEN bytes" could yield different results in the frontend than
> what the backend would get.
>
> > Even better idea would be creating a new function that returns the
> > actual rule name (after being shorten) from given view name. I don't
> > think it's a good idea to have codes to get an actual rule name in two
> > separate places.
>
> Given the above point about encoding differences, I think we *must*
> do the truncation in the backend ...
>
I agree with Tatsuo.
However we already have relkind for views.
Why must we rely on rulename to implement isViewRule()
in the first place ?
Regards.
Hiroshi Inoue