| From: | Arseniy Mukhin <arseniy(dot)mukhin(dot)dev(at)gmail(dot)com> |
|---|---|
| To: | Andres Freund <andres(at)anarazel(dot)de> |
| Cc: | Melanie Plageman <melanieplageman(at)gmail(dot)com>, Pg Hackers <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: Finalizing read stream users' flag choices |
| Date: | 2025-10-12 14:55:39 |
| Message-ID: | CAE7r3MJ5NHb1BMo1oCNWmfrG=Ytx-GKX44YNdA21FuQQQeq_Qg@mail.gmail.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Hi,
Could you please help to clarify the READ_STREAM_SEQUENTIAL usage
policy. There are several places that use the callback
'block_range_read_stream_cb'. So it means all of them have the same
sequential access pattern. But only one of them (amcheck
verify_heapam) uses READ_STREAM_SEQUENTIAL. Why is it? Is it correct
to say that 'block_range_read_stream_cb' produces a simple enough
pattern so we don't need READ_STREAM_SEQUENTIAL with it? OTOH we know
in advance that we want to read data strictly sequentially, so why not
let OS detect the pattern, if it does a better job as the comment
around READ_STREAM_SEQUENTIAL says. So amcheck usage looks valid too..
Best regards,
Arseniy Mukhin
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Sergey Fukanchik | 2025-10-12 16:04:40 | Is there public API to fetch errcode? |
| Previous Message | Srinath Reddy Sadipiralla | 2025-10-12 14:12:15 | Re: foreign key on virtual generated column |