| From: | Bruce Momjian <bruce(at)momjian(dot)us> |
|---|---|
| To: | Michael Paquier <michael(at)paquier(dot)xyz> |
| Cc: | Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com>, PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: PG 14 release notes, first draft |
| Date: | 2021-06-21 03:50:21 |
| Message-ID: | 20210621035021.GA22121@momjian.us |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Tue, Jun 15, 2021 at 12:01:00PM +0900, Michael Paquier wrote:
> On Tue, Jun 15, 2021 at 11:49:21AM +0900, Masahiko Sawada wrote:
> > On Tue, Jun 15, 2021 at 10:36 AM Bruce Momjian <bruce(at)momjian(dot)us> wrote:
> >> OK, but I need more information on how users will see a difference based
> >> on this commit:
>
> +1. That would be good to have in the release notes.
>
> > I think that since with this commit the server on Windows can handle a
> > file over 4GB, COPY FROM loading data from an over 4GB file and
> > pg_dump dumping a large table work now.
>
> Segment files or WAL files larger than 4GB also gain from that.
> Anything for which we may finish to do a stat() on benefits from this
> change if running on Windows. For pg_dump, a workaround in PG <= 13
> was to use --no-sync as the stat() failure came from files with a size
> larger than 4GB. That's rather sad as that means sacrifying
> durability for more usability :(
OK, I went with this text and put it in the Source Code section since it
applies to several layers of Postgres.
--
Bruce Momjian <bruce(at)momjian(dot)us> https://momjian.us
EDB https://enterprisedb.com
If only the physical world exists, free will is an illusion.
| Attachment | Content-Type | Size |
|---|---|---|
| master.diff | text/x-diff | 1003 bytes |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Masahiko Sawada | 2021-06-21 03:50:31 | Re: Optionally automatically disable logical replication subscriptions on error |
| Previous Message | osumi.takamichi@fujitsu.com | 2021-06-21 03:18:10 | RE: locking [user] catalog tables vs 2pc vs logical rep |