Re: Separate HEAP WAL replay logic into its own file

Li, Yong <yoli@ebay.com>

From: "Li, Yong" <yoli@ebay.com>
To: Melanie Plageman <melanieplageman@gmail.com>
Cc: "Debnath, Shawn" <sdn@ebay.com>, "Shyrabokau, Anton" <antons@ebay.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2024-06-19T06:44:58Z
Lists: pgsql-hackers

> On Jun 18, 2024, at 20:42, Melanie Plageman <melanieplageman@gmail.com> wrote:
> 
> External Email
> 
> On Mon, Jun 17, 2024 at 9:12 PM Li, Yong <yoli@ebay.com> wrote:
>> 
>> As a newcomer, when I was walking through the code looking for WAL replay related code, it was relatively easy for me to find them for the B-Tree access method because of the “xlog” hint in the file names. It took me a while to find the same for the heap access method. When I finally found them (via text search), it was a small surprise. Having different file organizations for different access methods gives me this urge to make everything consistent. I think it will make it easier for newcomers, and it will reduce the mental load for everyone to remember that heap replay is inside the heapam.c not some “???xlog.c”.
> 
> That makes sense. The branch for PG18 has not been cut yet, so I
> recommend registering this patch for the July commitfest [1] so it
> doesn't get lost.
> 
> - Melanie
> 

Thanks for the positive feedback. I’ve added the patch to the July CF.

Yong

Commits

  1. Move logic related to WAL replay of Heap/Heap2 into its own file