Re: BUG #18158: Assert in pgstat_report_stat() fails when a backend shutting down with stats pending - Mailing list pgsql-bugs

On Mon, Jun 8, 2026 at 1:49 PM Michael Paquier <michael@paquier.xyz> wrote:
>
> On Sat, Jun 06, 2026 at 08:45:26PM +0800, Xuneng Zhou wrote:
> > Sorry for being late. One nit for the patch -- would it be better to
> > use AmWalSenderProcess() for identification to save the include?
> > +#include "replication/walsender.h"
>
> In the startup sequence, MyBackendType is updated to B_WAL_SENDER a
> couple of lines after am_walsender.  One or the other don't really
> matter; they're both right.

Yeah, both of them work well.  To use AmWalSenderProcess(), we would
need to include src/include/miscadmin.h, whereas using am_walsender
requires including “replication/walsender.h”. I’m just wondering
whether using AmWalSenderProcess() would result in less coupling
between pgstat and replication.


--
Regards,
Xuneng Zhou
HighGo Software Co., Ltd.