On 2024-Jan-29, Dean Rasheed wrote:
> Yeah, for all practical purposes, that check in CheckValidResultRel()
> has been dead code since d751ba5235, but I think it's still worth
> doing, and if we're going to do it, we should do it properly. I don't
> like using elog() in some cases and ereport() in others, but I also
> don't like having that much duplicated code between this and the
> rewriter (and this patch doubles the size of that block).
>
> A neater solution is to export the rewriter functions and use them in
> CheckValidResultRel(). All these checks can then be reduced to
>
> if (!view_has_instead_trigger(...))
> error_view_not_updatable(...)
>
> which eliminates a lot of duplicated code and means that we now have
> just one place that throws these errors.
This looks quite nice, thanks. LGTM.
--
Álvaro Herrera 48°01'N 7°57'E — https://www.EnterpriseDB.com/
"No me acuerdo, pero no es cierto. No es cierto, y si fuera cierto,
no me acuerdo." (Augusto Pinochet a una corte de justicia)