Re: Remove unused fields in ReorderBufferTupleBuf

Aleksander Alekseev <aleksander@timescale.com>

From: Aleksander Alekseev <aleksander@timescale.com>
To: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Cc: Amit Kapila <amit.kapila16@gmail.com>, Masahiko Sawada <sawada.mshk@gmail.com>
Date: 2024-01-22T14:02:03Z
Lists: pgsql-hackers

Attachments

Hi,

> > I played a bit more with the patch. There was an idea to make
> > ReorderBufferTupleBufData an opaque structure known only within
> > reorderbyffer.c but it turned out that replication/logical/decode.c
> > accesses it directly so I abandoned that idea for now.
> >
> > > Alternatively we could convert ReorderBufferTupleBufData macro to an
> > > inlined function. At least it will add some type safety.
> >
> > Here is v3 that implements it too as a separate patch.
> >
>
> But why didn't you pursue your idea of getting rid of the wrapper
> structure ReorderBufferTupleBuf which after this patch will have just
> one member? I think there could be hassles in backpatching bug-fixes
> in some cases but in the longer run it would make the code look clean.

Indeed. In fact turned out that I suggested the same above but
apparently forgot:

> On top of that IMO it doesn't make much sense to keep a one-field
> wrapper structure. Perhaps we should get rid of it entirely and just
> use HeapTupleData instead.

After actually trying the refactoring I agree that the code becomes
cleaner and it's going to be beneficial in the long run. Here is the
patch.

--
Best regards,
Aleksander Alekseev

Commits

  1. Remove ReorderBufferTupleBuf structure.

  2. Generational memory allocator