| From: | Yuefei Shi <shiyuefei1004(at)gmail(dot)com> |
|---|---|
| To: | Chao Li <li(dot)evan(dot)chao(at)gmail(dot)com> |
| Cc: | Сергей Шиндерук <s(dot)shinderuk(at)postgrespro(dot)ru>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
| Subject: | Re: Typos in smgr.c |
| Date: | 2026-07-29 08:05:39 |
| Message-ID: | CAD43U4X-ZskUJRHdoYrCsZ=Fd-MQs75CC-gxPi4cZQrd1yU8Bw@mail.gmail.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Hi,
I think this fix is reasonable. The current assignment:
sd->smgr.skip_fsync = skip_fsync && !SmgrIsTemp(smgr);
does not match the comment that temporary relations should never be
fsync'd, nor the existing smgr write path logic. The condition should be:
sd->smgr.skip_fsync = skip_fsync || SmgrIsTemp(smgr);
Although skip_fsync is not currently consumed, this fixes the semantics and
avoids carrying an incorrect condition into future AIO write support.
Thanks.
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Rafia Sabih | 2026-07-29 08:21:38 | Re: Add statistics refresh materialized view |
| Previous Message | Chao Li | 2026-07-29 07:37:08 | Re: Typos in smgr.c |