From: Tom Lane <tgl@sss.pgh.pa.us>
To: r.zharkov@postgrespro.ru
cc: pgsql-bugs@lists.postgresql.org
Subject: Re: BUG #17465: Wrong error message while trying to change system
 column type.
In-reply-to: <17465-f2a554a6cb5740d3@postgresql.org>
References: <17465-f2a554a6cb5740d3@postgresql.org>
Comments: In-reply-to PG Bug reporting form <noreply@postgresql.org>
	message dated "Mon, 18 Apr 2022 11:13:28 -0000"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <2283490.1650296211.1@sss.pgh.pa.us>
Date: Mon, 18 Apr 2022 11:36:51 -0400
Message-ID: <2283491.1650296211@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/2283491.1650296211%40sss.pgh.pa.us>
Precedence: bulk

PG Bug reporting form <noreply@postgresql.org> writes:
> accidentally I tried to change a system column data type on Windows and got
> a strange error message "no owned sequence found":

> postgres=# alter table t1 alter column cmin type text;
> ERROR:  no owned sequence found
> postgres=# alter table t1 alter column cmax type text;
> ERROR:  cannot alter system column "cmax"

Yeah, same here.

> I think the ATParseTransformCmd() function needs the same check as the check
> in the ATPrepAlterColumnType() function.

That seems pretty duplicative.  I think it's sufficient to do something
like

-                    if (TupleDescAttr(tupdesc, attnum - 1)->attidentity)
+                    if (attnum > 0 &&
+                        TupleDescAttr(tupdesc, attnum - 1)->attidentity)

in the faulty code, and let the actual error message come out where
it does now.

Will fix, thanks for the report!

			regards, tom lane