Re: oversight in EphemeralNamedRelation support

On Tue, Oct 10, 2017 at 2:35 AM, Julien Rouhaud <rjuju123(at)gmail(dot)com> wrote:
> Hugo Mercier (in Cc) reported me an error in a query, failing since pg10.
>
> Simple test case to reproduce:
>
> CREATE TABLE public.test (id integer);
> WITH test AS (select 42) INSERT INTO public.test SELECT * FROM test;
>
> which will fail with "relation "test" cannot be the target of a
> modifying statement".
>
> IIUC, that's an oversight in 18ce3a4ab22, where setTargetTable()
> doesn't exclude qualified relation when searching for special
> relation.

I agree.

> PFA a simple patch to fix this issue, with updated regression test.

Thanks!

I suppose we could consider moving the schemaname check into
getRTEForSpecialRelationType(), since otherwise both callers need to
do that (and as you discovered, one forgot).

--
Thomas Munro
http://www.enterprisedb.com