On 10/09/2012 04:06 PM, Tom Lane wrote:
> Second, to my mind the point of a multi-table TRUNCATE is to ensure that
> all the referenced tables get reset to empty *together*. With something
> like this, you'd have no such guarantee. Consider a timeline like this:
>
> Session 1 Session 2
>
> TRUNCATE IF EXISTS a, b, c;
> ... finds c doesn't exist ...
> ... working on a and b ...
> CREATE TABLE c ( ... );
> INSERT INTO c ...;
> ... commits ...
>
> Now we have a, b, and c, but c isn't empty, violating the expectations
> of session 1. So even if there's a use-case for IF EXISTS on a single
> table, I think it's very very dubious to allow it in multi-table
> commands.
Hi,
I've to say that I don't understand your timeline :
- If c exist in Session 1, CREATE TABLE in Session 2 can't be done,
neither INSERT
- If c doesn't exists in Session 1, it will be ignored, then, Session 2
work fine.
In any case, TRUNCATE is sent before INSERT, but it can't lock an object
which still not exists.
I understand that people don't want TRUNCATE IF EXISTS, but, in my point
of view, even if TRUNCATE is not a DDL, it's the same use-case as DROP
TABLE IF EXISTS.
Regards,
--
Sébastien Lardière
PostgreSQL DBA Team Manager
Hi-Media
