Horton

Re: BufFileRead() error signalling

Robert Haas <robertmhaas@gmail.com>

View thread View on postgr.es
From: Robert Haas <robertmhaas@gmail.com>
To: Michael Paquier <michael@paquier.xyz>
Cc: Thomas Munro <thomas.munro@gmail.com>, Ibrar Ahmed <ibrar.ahmad@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2020-01-29T15:01:31Z
Lists: pgsql-hackers 
On Wed, Jan 29, 2020 at 1:26 AM Michael Paquier <michael@paquier.xyz> wrote:
> On Tue, Jan 28, 2020 at 03:51:54PM -0500, Robert Haas wrote:
> > I quickly reread that thread and I don't see that there's any firm
> > consensus there in favor of "read %d of %zu" over "read only %d of %zu
> > bytes". Now, if most people prefer the former, so be it, but I don't
> > think that's clear from that thread.
>
> The argument of consistency falls in favor of the former on HEAD:
> $ git grep "could not read" | grep "read %d of %zu" | wc -l
> 59
> $ git grep "could not read" | grep "read only %d of %zu" | wc -l
> 0

True. I didn't realize that 'read %d of %zu' was so widely used.

Your grep misses one instance of 'read only %d of %d bytes' because
you grepped for %zu specifically, but that doesn't really change the
overall picture.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

Commits

  1. Make BufFileWrite() void.

  2. Fix buffile.c error handling.

  3. Adjust WAL code so that checkpoints truncate the xlog at the previous