Great! Would it be implemented in a next version? Seems it would be
very helpful, especially for people who commit database structure to
CVS/SVN once per minute to track changes history (or similar)...
On Sun, Sep 21, 2008 at 11:57 PM, Tom Lane <[EMAIL PROTECTED]> wrote:
> "Dmitry Koterov" <[EMAI
"Dmitry Koterov" <[EMAIL PROTECTED]> writes:
> CREATE TRIGGER t000_set_id
> -BEFORE INSERT OR DELETE OR UPDATE ON a
> +BEFORE INSERT OR DELETE OR UPDATE ON b
> FOR EACH ROW
> EXECUTE PROCEDURE i_trg();
> CREATE TRIGGER t000_set_id
> -BEFORE INSERT OR DELETE OR UPDATE ON b
>
Unfortunately, I cannot reproduce this with 100% effect.
But, time to time I execute diff utility for a database and notice
that two or more trigger or constraint definitions (or something else)
are permuted. Something like this:
+ALTER TABLE ONLY a
+ADD CONSTRAINT "fk_b_Id" FOREIGN KEY (b_i
"Dmitry Koterov" <[EMAIL PROTECTED]> writes:
> Utility pg_dump dumps the identical database schemas not always
> identically: sometimes it changes an order of SQL statements.
Please provide a concrete example. The dump order for modern servers
(ie, since 7.3) is by object type, and within a type
Hello.
Utility pg_dump dumps the identical database schemas not always
identically: sometimes it changes an order of SQL statements.
E.g.:
1. Dump of database A:
ALTER TABLE xxx ADD CONSTRAINT ...;
ALTER TABLE yyy ADD CONSTRAINT ...;
2. Dump of database B which has identical structure as A ("pg