Horton

Re: ExecTypeSetColNames is fundamentally broken

Aleksander Alekseev <aleksander@timescale.com>

View thread View on postgr.es
From: Aleksander Alekseev <aleksander@timescale.com>
To: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, Robert Haas <robertmhaas@gmail.com>
Date: 2022-03-16T12:47:15Z
Lists: pgsql-hackers 
Hi hackers,

> Anyone else have thoughts?

I came across this thread while looking for the patches that need review.

My understanding of the code is limited, but I can say that I don't
see anything particularly wrong with it. I can also confirm that it
fixes the problem reported by the user while passing the rest of the
tests.

I understand the concern expressed by Robert in respect of backward
compatibility. From the user's perspective, personally I would prefer
a fixed bug over backward compatibility. Especially if we consider the
fact that the current behaviour of cases like `select row_to_json(i)
from int8_tbl i(x,y)` is not necessarily the correct one, at least it
doesn't look right to me.

So unless anyone has strong objections against the proposed fix or can
propose a better patch, I would suggest merging it.

-- 
Best regards,
Aleksander Alekseev

Commits

  1. Don't bother to attach column name lists to RowExprs of named types.

  2. Revert applying column aliases to the output of whole-row Vars.