Re: Support for 8-byte TOAST values, round two

On Thu, Sep 10, 2026 at 08:18:57AM +0900, Michael Paquier wrote:
> Hmm. Okay. I'll think about something here.

Attached is a rebased v16 following cb298616463d. The patch set has
been slightly re-organized, with the pieces that seem worth
introducing first on top of the rest:
- 0001 for binary upgrades.
- 0002 for the reloption toast_value_type
- 0003 for adding more Oid8 in the TOAST APIs compared to Oid.
- 0004 for pg_column_toast_chunk_id() switched to Oid8
- 0005 for adding toast_value_type=oid8, with all the TOAST changes
not related to the new vartag.
- 0006 is the battery of tests for all the cases found during
development, grouped together: core things, amcheck, test_decoding..
- 0007 adds the new vartag_external.
- 0008 is the tweak for toast_tuple_find_biggest_attribute(), which
should perhaps be merged with 0007.

Most of the previous comments were related 0002 and the new reloption
(double whitespace, doc improvement regarding no effects after
initial creation of the TOAST table, parens in the defaultarg of
rel.h. regression tests). I've addressed all of them, as so as whole
0002 feels kind of OK as-is.
--
Michael