Re: ORDER BY pushdowns seem broken in postgres_fdw - Mailing list pgsql-hackers

From Ronan Dunklau
Subject Re: ORDER BY pushdowns seem broken in postgres_fdw
Date
Msg-id 3165278.bG4bnikssK@aivenronan
Whole thread Raw
In response to Re: ORDER BY pushdowns seem broken in postgres_fdw  (Ranier Vilela <ranier.vf@gmail.com>)
Responses Re: ORDER BY pushdowns seem broken in postgres_fdw  (David Rowley <dgrowleyml@gmail.com>)
Re: ORDER BY pushdowns seem broken in postgres_fdw  (Ranier Vilela <ranier.vf@gmail.com>)
List pgsql-hackers
Le jeudi 22 juillet 2021, 02:16:52 CEST Ranier Vilela a écrit :
> Unfortunately your patch does not apply clear into the head.
> So I have a few suggestions on v2, attached with the .txt extension to
> avoid cf bot.
> Please, if ok, make the v3.

Hum weird, it applied cleanly for me, and was formatted using git show which I
admit is not ideal. Please find it reattached.


>
> 2. appendOrderbyUsingClause function
> Put the buffer actions together?
>
Not sure what you mean here ?

> 3. Apply style Postgres?
> + if (!HeapTupleIsValid(tuple))
> + {
> + elog(ERROR, "cache lookup failed for operator family %u",
> pathkey->pk_opfamily);
> + }
>

Good catch !


> 4. Assertion not ok here?
> + em = find_em_for_rel(pathkey->pk_eclass, baserel);
> + em_expr = em->em_expr;
>   Assert(em_expr != NULL);
>

If we are here there should never be a case where the em can't be found. I
moved the assertion where it makes sense though.



Regards,


--
Ronan Dunklau
Attachment

pgsql-hackers by date:

Previous
From: Pavel Stehule
Date:
Subject: Re: psql - add SHOW_ALL_RESULTS option
Next
From: Julien Rouhaud
Date:
Subject: Re: Hook for extensible parsing.