Re: ORDER BY pushdowns seem broken in postgres_fdw

On Mon, Sep 6, 2021 at 1:17 AM Ronan Dunklau <ronan(at)dunklau(dot)fr> wrote:

> Le vendredi 3 septembre 2021, 22:54:25 CEST David Zhang a écrit :
> > The following review has been posted through the commitfest application:
> > make installcheck-world: tested, failed
> > Implements feature: tested, passed
> > Spec compliant: not tested
> > Documentation: not tested
> >
> > Applied the v6 patch to master branch and ran regression test for
> contrib,
> > the result was "All tests successful."
>
> What kind of error did you get running make installcheck-world ? If it
> passed
> the make check for contrib, I can't see why it would fail running make
> installcheck-world.
>
> In any case, I just checked and running make installcheck-world doesn't
> produce any error.
>
> Since HEAD had moved a bit since the last version, I rebased the patch,
> resulting in the attached v7.
>
> Best regards,
>
> --
> Ronan Dunklau
>
Hi,
bq. a pushed-down order by could return wrong results.

Can you briefly summarize the approach for fixing the bug in the
description ?

+ * Returns true if it's safe to push down a sort as described by 'pathkey'
to
+ * the foreign server
+ */
+bool
+is_foreign_pathkey(PlannerInfo *root,

It would be better to name the method which reflects whether pushdown is
safe. e.g. is_pathkey_safe_for_pushdown.

Cheers