From: Tom Lane <tgl@sss.pgh.pa.us>
To: Peter Geoghegan <pg@bowt.ie>
cc: pgsql-hackers@lists.postgresql.org
Subject: New compiler warning from btree dedup code
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <3171400.1649275153.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Wed, 06 Apr 2022 15:59:13 -0400
Message-ID: <3171401.1649275153@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/3171401.1649275153%40sss.pgh.pa.us>
Precedence: bulk

I just noticed that if I build without asserts on my Mac laptop
(using Apple's latest clang, 13.1.6) I get

nbtdedup.c:68:8: warning: variable 'pagesaving' set but not used [-Wunused=
-but-set-variable]
        Size            pagesaving =3D 0;
                        ^
1 warning generated.

Apparently, late-model clang can figure out that the variable
is incremented but not otherwise used.  This is enough to
shut it up, but I wonder if you have another preference:

-   Size        pagesaving =3D 0;
+   Size        pagesaving PG_USED_FOR_ASSERTS_ONLY =3D 0;

			regards, tom lane