Thread: BUG #18124: PG16 release note document bug in "Add build option to allow testing of small WAL segment sizes"

The following bug has been logged on the website:

Bug reference:      18124
Logged by:          TAKATSUKA Haruka
Email address:      harukat@sraoss.co.jp
PostgreSQL version: 16.0
Operating system:   any
Description:

It seems that "WAL" is unnecessary in the following item.

doc/src/sgml/release-16.sgml:
|<!--
|Author: Andres Freund <andres@anarazel.de>
|2022-12-07 [d3b111e32] Add option to specify segment size in blocks
|-->
|
|     <listitem>
|      <para>
|       Add build option to allow testing of small <acronym>WAL</acronym>
|       segment sizes (Andres Freund)
|      </para>


On Thu, Sep 21, 2023 at 06:39:39AM +0000, PG Bug reporting form wrote:
> The following bug has been logged on the website:
> 
> Bug reference:      18124
> Logged by:          TAKATSUKA Haruka
> Email address:      harukat@sraoss.co.jp
> PostgreSQL version: 16.0
> Operating system:   any
> Description:        
> 
> It seems that "WAL" is unnecessary in the following item.
> 
> doc/src/sgml/release-16.sgml:
> |<!--
> |Author: Andres Freund <andres@anarazel.de>
> |2022-12-07 [d3b111e32] Add option to specify segment size in blocks
> |-->
> |
> |     <listitem>
> |      <para>
> |       Add build option to allow testing of small <acronym>WAL</acronym>
> |       segment sizes (Andres Freund)
> |      </para>

Uh, it is true that we don't have any segment sizes other than WAL, but
I am not sure people would easily know that, so I added WAL so people
knew.

-- 
  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
  EDB                                      https://enterprisedb.com

  Only you can decide what is important to you.



On Thu, Sep 21, 2023 at 6:47 PM Bruce Momjian <bruce@momjian.us> wrote:
> Uh, it is true that we don't have any segment sizes other than WAL, but
> I am not sure people would easily know that, so I added WAL so people
> knew.

But the commit in question added a new option that can be used to
control the relation segment size -- not the WAL segment size.
Obviously, that's what TAKATSUKA-san meant.

--
Peter Geoghegan



On Thu, Sep 21, 2023 at 06:55:09PM -0700, Peter Geoghegan wrote:
> On Thu, Sep 21, 2023 at 6:47 PM Bruce Momjian <bruce@momjian.us> wrote:
> > Uh, it is true that we don't have any segment sizes other than WAL, but
> > I am not sure people would easily know that, so I added WAL so people
> > knew.
> 
> But the commit in question added a new option that can be used to
> control the relation segment size -- not the WAL segment size.
> Obviously, that's what TAKATSUKA-san meant.

Oh, I see now, my mistake.  How is the attached patch?  Thanks.

-- 
  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
  EDB                                      https://enterprisedb.com

  Only you can decide what is important to you.

Attachment
On Thu, Sep 21, 2023 at 7:06 PM Bruce Momjian <bruce@momjian.us> wrote:
> Oh, I see now, my mistake.  How is the attached patch?  Thanks.

LGTM.

--
Peter Geoghegan



Peter Geoghegan <pg@bowt.ie> writes:
> On Thu, Sep 21, 2023 at 6:47 PM Bruce Momjian <bruce@momjian.us> wrote:
>> Uh, it is true that we don't have any segment sizes other than WAL, but
>> I am not sure people would easily know that, so I added WAL so people
>> knew.

> But the commit in question added a new option that can be used to
> control the relation segment size -- not the WAL segment size.
> Obviously, that's what TAKATSUKA-san meant.

Yeah.  The release note entry is simply wrong to say it's WAL segment
size.  I would also argue that d3b111e32's installation.sgml changes
were poorly worded, because they only say "segment size" which can
easily be misunderstood, just as happened here.  Better would be
"relation segment size" or "table segment size".

            regards, tom lane



Hi,

On 2023-09-21 22:08:18 -0400, Tom Lane wrote:
> Peter Geoghegan <pg@bowt.ie> writes:
> > On Thu, Sep 21, 2023 at 6:47 PM Bruce Momjian <bruce@momjian.us> wrote:
> >> Uh, it is true that we don't have any segment sizes other than WAL, but
> >> I am not sure people would easily know that, so I added WAL so people
> >> knew.
> 
> > But the commit in question added a new option that can be used to
> > control the relation segment size -- not the WAL segment size.
> > Obviously, that's what TAKATSUKA-san meant.
> 
> Yeah. The release note entry is simply wrong to say it's WAL segment size.

Agreed. Bruce are you committing that bit?


> I would also argue that d3b111e32's installation.sgml changes
> were poorly worded, because they only say "segment size" which can
> easily be misunderstood, just as happened here.  Better would be
> "relation segment size" or "table segment size".

Hm. Yea. I copied the language from --with-segsize, but there there's
subsequent sentences that do clarify what the option relates to. I prefer
"relation" over "table" as it affects indexes as well.

Pushed that adjustment.

Greetings,

Andres Freund



On Mon, Sep 25, 2023 at 10:45:04AM -0700, Andres Freund wrote:
> Hi,
> 
> On 2023-09-21 22:08:18 -0400, Tom Lane wrote:
> > Peter Geoghegan <pg@bowt.ie> writes:
> > > On Thu, Sep 21, 2023 at 6:47 PM Bruce Momjian <bruce@momjian.us> wrote:
> > >> Uh, it is true that we don't have any segment sizes other than WAL, but
> > >> I am not sure people would easily know that, so I added WAL so people
> > >> knew.
> > 
> > > But the commit in question added a new option that can be used to
> > > control the relation segment size -- not the WAL segment size.
> > > Obviously, that's what TAKATSUKA-san meant.
> > 
> > Yeah. The release note entry is simply wrong to say it's WAL segment size.
> 
> Agreed. Bruce are you committing that bit?

Yes, I will try to do it tomorrow.

> > I would also argue that d3b111e32's installation.sgml changes
> > were poorly worded, because they only say "segment size" which can
> > easily be misunderstood, just as happened here.  Better would be
> > "relation segment size" or "table segment size".
> 
> Hm. Yea. I copied the language from --with-segsize, but there there's
> subsequent sentences that do clarify what the option relates to. I prefer
> "relation" over "table" as it affects indexes as well.
> 
> Pushed that adjustment.

Thanks.

-- 
  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
  EDB                                      https://enterprisedb.com

  Only you can decide what is important to you.



On Mon, Sep 25, 2023 at 06:32:48PM -0400, Bruce Momjian wrote:
> On Mon, Sep 25, 2023 at 10:45:04AM -0700, Andres Freund wrote:
> > Hi,
> > 
> > On 2023-09-21 22:08:18 -0400, Tom Lane wrote:
> > > Peter Geoghegan <pg@bowt.ie> writes:
> > > > On Thu, Sep 21, 2023 at 6:47 PM Bruce Momjian <bruce@momjian.us> wrote:
> > > >> Uh, it is true that we don't have any segment sizes other than WAL, but
> > > >> I am not sure people would easily know that, so I added WAL so people
> > > >> knew.
> > > 
> > > > But the commit in question added a new option that can be used to
> > > > control the relation segment size -- not the WAL segment size.
> > > > Obviously, that's what TAKATSUKA-san meant.
> > > 
> > > Yeah. The release note entry is simply wrong to say it's WAL segment size.
> > 
> > Agreed. Bruce are you committing that bit?
> 
> Yes, I will try to do it tomorrow.

Patch applied.

-- 
  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
  EDB                                      https://enterprisedb.com

  Only you can decide what is important to you.