Re: Do we want a hashset type?
| От | Joel Jacobson |
|---|---|
| Тема | Re: Do we want a hashset type? |
| Дата | |
| Msg-id | 0f6d4856-95b8-415c-b820-b319da61ce1c@app.fastmail.com обсуждение исходный текст |
| Ответ на | Re: Do we want a hashset type? ("Joel Jacobson" <joel@compiler.org>) |
| Ответы |
Re: Do we want a hashset type?
|
| Список | pgsql-hackers |
On Sun, Jun 25, 2023, at 11:42, Joel Jacobson wrote:
> SELECT hashset_contains('{}'::int4hashset, NULL::int);
>
> would be False, according to the General Rules.
>
...
> Applying the same rules, we'd have to return Unknown (which we represent as
> null) for:
>
> SELECT hashset_contains('{null}'::int4hashset, NULL::int);
>
Aha! I just discovered to my surprise that the corresponding array
queries gives the same result:
SELECT NULL = ANY(ARRAY[]::int[]);
?column?
----------
f
(1 row)
SELECT NULL = ANY(ARRAY[NULL]::int[]);
?column?
----------
(1 row)
I have no more objections; let's stick to the same null semantics as arrays and multisets.
/Joel
В списке pgsql-hackers по дате отправления: