Commit 86dc90056 - Rework planning and execution of UPDATE and DELETE - Mailing list pgsql-hackers

With the commit mentioned in the $subject, I am seeing the

change in behaviour with the varlena header size.  Please

consider the below test:

postgres@83795=#CREATE TABLE test_storage_char(d char(20));
CREATE TABLE
postgres@83795=#INSERT INTO test_storage_char SELECT REPEAT('e', 20);
INSERT 0 1
postgres@83795=#SELECT d, pg_column_size(d) FROM test_storage_char;
          d           | pg_column_size
----------------------+----------------
 eeeeeeeeeeeeeeeeeeee |             21
(1 row)

postgres@83795=#ALTER TABLE test_storage_char ALTER COLUMN  d SET STORAGE PLAIN;
ALTER TABLE
postgres@83795=#SELECT d, pg_column_size(d) FROM test_storage_char;
          d           | pg_column_size
----------------------+----------------
 eeeeeeeeeeeeeeeeeeee |             21
(1 row)

postgres@83795=#UPDATE test_storage_char SET d='ab' WHERE d LIKE '%e%';
UPDATE 1
postgres@83795=#SELECT d, pg_column_size(d) FROM test_storage_char;
          d           | pg_column_size
----------------------+----------------
 ab                   |             24
(1 row)

After changing the STORAGE for the column and UPDATE, pg_column_size 

now returns the size as 24. 

postgres@129158=#SELECT d, pg_column_size(d) FROM test_storage_char;
          d           | pg_column_size
----------------------+----------------
 ab                   |             21
(1 row)

I am not sure whether this change is expected? Or missing something

in the toasting the attribute?

Thanks,

Rushabh Lathia