Обсуждение: bug: repeated ALTER COLUMN SET DATA TYPE corrupt check constraint

Поиск
Список
Период
Сортировка

bug: repeated ALTER COLUMN SET DATA TYPE corrupt check constraint

От
jian he
Дата:
hi.

while working on let ALTER COLUMN SET DATA TYPE cope with trigger dependency.
I found this bug, it involves several ALTER COLUMN SET DATA TYPE calls.

DROP TABLE IF EXISTS main_table;
CREATE TABLE main_table (a int, b int, check(a = b));
ALTER TABLE main_table ALTER COLUMN a SET DATA TYPE numeric;
ALTER TABLE main_table ALTER COLUMN a SET DATA TYPE int;
ALTER TABLE main_table ALTER COLUMN a SET DATA TYPE text;
ALTER TABLE main_table ALTER COLUMN b SET DATA TYPE text;
INSERT INTO main_table VALUES ('a', 'a');

This also applies to statistics, indexes. for statistics: the error is only
observable after running ANALYZE manually.


--
jian
https://www.enterprisedb.com/



Re: bug: repeated ALTER COLUMN SET DATA TYPE corrupt check constraint

От
Srinath Reddy Sadipiralla
Дата:
Hi Jian,

On Sat, Dec 27, 2025 at 1:28 PM jian he <jian.universality@gmail.com> wrote:
hi.

while working on let ALTER COLUMN SET DATA TYPE cope with trigger dependency.
I found this bug, it involves several ALTER COLUMN SET DATA TYPE calls.

DROP TABLE IF EXISTS main_table;
CREATE TABLE main_table (a int, b int, check(a = b));
ALTER TABLE main_table ALTER COLUMN a SET DATA TYPE numeric;
ALTER TABLE main_table ALTER COLUMN a SET DATA TYPE int;
ALTER TABLE main_table ALTER COLUMN a SET DATA TYPE text;
ALTER TABLE main_table ALTER COLUMN b SET DATA TYPE text;
INSERT INTO main_table VALUES ('a', 'a');

This also applies to statistics, indexes. for statistics: the error is only
observable after running ANALYZE manually.

Yeah, I can easily reproduce this. 

--
Thanks,
Srinath Reddy Sadipiralla
EDB: https://www.enterprisedb.com/

Re: bug: repeated ALTER COLUMN SET DATA TYPE corrupt check constraint

От
jian he
Дата:
On Sat, Dec 27, 2025 at 4:32 PM Srinath Reddy Sadipiralla
<srinath2133@gmail.com> wrote:
>>
>
> Yeah, I can easily reproduce this.
>
hi.

thinking about it more...
I tend to think it's not a bug.
because SET DATA TYPE changes the CHECK constraint definition.

>> DROP TABLE IF EXISTS main_table;
>> CREATE TABLE main_table (a int, b int, check(a = b));
>> ALTER TABLE main_table ALTER COLUMN a SET DATA TYPE numeric;
>> ALTER TABLE main_table ALTER COLUMN a SET DATA TYPE int;

At this stage, the constraint definition is no longer "a = b",
it becomes "(a::numeric = b::numeric)", which is different from "(a = b)".

sorry for the noise.