Re: Add pg_walinspect function with block info columns

At Tue, 7 Mar 2023 16:18:21 +0900, Michael Paquier <michael(at)paquier(dot)xyz> wrote in
> On Tue, Mar 07, 2023 at 03:49:02PM +0900, Kyotaro Horiguchi wrote:
> > Ah. Yes, that expansion sounds sensible.
>
> Okay, so, based on this idea, I have hacked on this stuff and finish
> with the attached that shows block data if it exists, as well as FPI
> stuff if any. bimg_info is showed as a text[] for its flags.

# The naming convetion looks inconsistent between
# pg_get_wal_records_info and pg_get_wal_block_info but it's not an
# issue of this patch..

> I guess that I'd better add a test that shows correctly a record with
> some block data attached to it, on top of the existing one for FPIs..
> Any suggestions? Perhaps just a heap/heap2 record?
>
> Thoughts?

I thought that we needed a test for block data when I saw the patch.
I don't have great idea but a single insert should work.

regards.

--
Kyotaro Horiguchi
NTT Open Source Software Center