Re: Separate HEAP WAL replay logic into its own file - Mailing list pgsql-hackers

From Melanie Plageman
Subject Re: Separate HEAP WAL replay logic into its own file
Date
Msg-id CAAKRu_Zn5WQm1UA-ickT9h7rZQ_44K3cZdxgu354UarDFjJnig@mail.gmail.com
Whole thread Raw
In response to Separate HEAP WAL replay logic into its own file  ("Li, Yong" <yoli@ebay.com>)
Responses Re: Separate HEAP WAL replay logic into its own file
List pgsql-hackers
On Mon, Jun 17, 2024 at 2:20 AM Li, Yong <yoli@ebay.com> wrote:
>
> Hi PostgreSQL hackers,
>
> For most access methods in PostgreSQL, the implementation of the access method itself and the implementation of its
WALreplay logic are organized in separate source files.  However, the HEAP access method is an exception.  Both the
accessmethod and the WAL replay logic are collocated in the same heapam.c.  To follow the pattern established by other
accessmethods and to improve maintainability, I made the enclosed patch to separate HEAP’s replay logic into its own
file. The changes are straightforward.  Move the replay related functions into the new heapam_xlog.c file, push the
commonheap_execute_freeze_tuple() helper function into the heapam.h header, and adjust the build files. 

I'm not against this change, but I am curious at what inspired this.
Were you looking at Postgres code and simply noticed that there isn't
a heapam_xlog.c (like there is a nbtxlog.c etc) and thought that you
wanted to change that? Or is there some specific reason this would
help you as a Postgres developer, user, or ecosystem member?

- Melanie



pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: DROP OWNED BY fails to clean out pg_init_privs grants
Next
From: Nathan Bossart
Date:
Subject: Re: Introduce XID age and inactive timeout based replication slot invalidation