Re: Use XLOG_CONTROL_FILE macro everywhere?

Fujii Masao <masao.fujii@oss.nttdata.com>

From: Fujii Masao <masao.fujii@oss.nttdata.com>
To: "Anton A. Melnikov" <a.melnikov@postgrespro.ru>, Kyotaro Horiguchi <horikyota.ntt@gmail.com>
Cc: daniel@yesql.se, peter@eisentraut.org, tgl@sss.pgh.pa.us, robertmhaas@gmail.com, michael@paquier.xyz, pgsql-hackers@postgresql.org
Date: 2025-04-05T03:11:48Z
Lists: pgsql-hackers

On 2025/04/05 8:27, Anton A. Melnikov wrote:
> Hi!
> 
> On 04.04.2025 19:32, Fujii Masao wrote:
>> I'm not sure there's clear consensus yet on the changes in the 0001 and
>> 0002 patches, and it might not be worth rushing them in right before
>> the feature freeze. So for now, I reviewed and updated only the 0003 patch,
>> since there seems to be agreement on that one.
>>
>> I've attached the updated version. It fixes a typo and replaces
>> the remaining hardcoded control file name with the XLOG_CONTROL_FILE macro.
> 
> Thanks! Looks good for me.

Thanks for checking! Barring any objections, I'll go ahead and commit the patch.


> Maybe extend this to perl tests since there are several hardcoded names too?
> The patch attached tries to do this.

I had the same thought, i.e., we could use scan_server_header() to pull
XLOG_CONTROL_FILE from xlog_internal.h and replace the hardcoded "global/pg_control"
in the TAP tests. But for now, that feels like overkill, since the TAP tests already
have other hardcoded values like the WAL directory (pg_wal), and we haven't used
macros like XLOGDIR there either. Replacing just "global/pg_control" might feel
inconsistent, and handling the other cases properly would take more time beyond
feature freeze. So I'm thinking it's better to use XLOG_CONTROL_FILE only in
C code for now.

Regards,

-- 
Fujii Masao
Advanced Computing Technology Center
Research and Development Headquarters
NTT DATA CORPORATION




Commits

  1. Use XLOG_CONTROL_FILE macro consistently for control file name.

  2. Doc: Update link to the mentioned subsection