Re: Attribute of type record has wrong type error with MERGE ... WHEN NOT MATCHED BY SOURCE THEN DELETE

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Tender Wang <tndrwang(at)gmail(dot)com>
Cc: Dean Rasheed <dean(dot)a(dot)rasheed(at)gmail(dot)com>, Duncan Sands <duncan(dot)sands(at)deepbluecap(dot)com>, pgsql-bugs(at)lists(dot)postgresql(dot)org
Subject: Re: Attribute of type record has wrong type error with MERGE ... WHEN NOT MATCHED BY SOURCE THEN DELETE
Date: 2025-03-10 15:00:46
Message-ID: 765994.1741618846@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

Tender Wang <tndrwang(at)gmail(dot)com> writes:
> When the query has NOT MATCHED BY SOURCE, commit d7d297f84 add "src IS NOT
> NULL" join condition.
> In this case, the src is view(e.g. subquery), so in makeWholeRowVar(), it
> will call below code:
> result = makeVar(varno,
> InvalidAttrNumber,
> RECORDOID,
> -1,
> InvalidOid,
> varlevelsup);

> the vartype is RECORDOID, but te reltype of src is not RECORDOID, so
> $SUBJECT error reports.

Hmm. I tried adjusting the example to make _country_or_region
be a materialized view or plain table instead of a view, and
those cases did not fail. I wonder why the difference...

> I add the below codes to makeWholeRowVar() default branch:
> if (rte->relkind == RELKIND_VIEW)
> toid = get_rel_type_id(rte->relid);
> else
> toid = RECORDOID;
> It can work.

I'm quite uncomfortable with the idea of changing makeWholeRowVar()
itself in this way --- the potential blast radius from that seems
rather large. A localized fix in transform_MERGE_to_join() might
be a wiser answer.

regards, tom lane

In response to

Browse pgsql-bugs by date

  From Date Subject
Next Message Dean Rasheed 2025-03-10 15:11:16 Re: Attribute of type record has wrong type error with MERGE ... WHEN NOT MATCHED BY SOURCE THEN DELETE
Previous Message Tom Lane 2025-03-10 14:23:47 Re: Window Functions with identical PARTITION BY and ORDER BY clauses evaluated separately