Re: pg_column_toast_chunk_id: a function to get a chunk ID of a TOASTed value

On Tue, 2 Jan 2024 08:00:00 +0800
jian he <jian(dot)universality(at)gmail(dot)com> wrote:

> On Mon, Nov 6, 2023 at 8:00 AM jian he <jian(dot)universality(at)gmail(dot)com> wrote:
> >
> > minor doc issues.
> > Returns the chunk id of the TOASTed value, or NULL if the value is not TOASTed.
> > Should it be "chunk_id"?

Thank you for your suggestion. As you pointed out, it is called "chunk_id"
in the documentation, so I rewrote it and also added a link to the section
where the TOAST table structure is explained.

> > you may place it after pg_create_logical_replication_slot entry to
> > make it look like alphabetical order.

I've been thinking about where we should place the function in the doc,
and I decided place it in the table of Database Object Size Functions
because I think pg_column_toast_chunk_id also would assist understanding
the result of size functions as similar to pg_column_compression; that is,
those function can explain why a large value in size could be stored in
a column.

> > There is no test. maybe we can add following to src/test/regress/sql/misc.sql
> > create table val(t text);
> > INSERT into val(t) SELECT string_agg(
> > chr((ascii('B') + round(random() * 25)) :: integer),'')
> > FROM generate_series(1,2500);
> > select pg_column_toast_chunk_id(t) is not null from val;
> > drop table val;

Thank you for the test proposal. However, if we add a test, I want
to check that the chunk_id returned by the function exists in the
TOAST table, and that it returns NULL if the values is not TOASTed.
For the purpose, I wrote a test using a dynamic SQL since the table
name of the TOAST table have to be generated from the main table's OID.

> Hi
> the main C function (pg_column_toast_chunk_id) I didn't change.
> I added tests as mentioned above.
> tests put it on src/test/regress/sql/misc.sql, i hope that's fine.
> I placed pg_column_toast_chunk_id in "Table 9.99. Database Object
> Location Functions" (below Table 9.98. Database Object Size
> Functions).

I could not find any change in your patch from my previous patch.
Maybe, you attached wrong file. I attached a patch updated based
on your review, including the documentation fixes and a test.
What do you think about this it?

Regards,
Yugo Nagata

--
Yugo NAGATA <nagata(at)sraoss(dot)co(dot)jp>