[GENERAL] sequence behavior - is this correct

[Marek Lewczuk]

Hello,
I would like to ask if my problem with sequence is a proper behavior or 
this is a bug (probably not)...

I have a table:
CREATE TABLE "testtable" (
  "serialfield" SERIAL,
  "someotherfield" TEXT,
  PRIMARY KEY("serialfield")
) WITH OIDS;
After creation of this table, sequence "testtable_serialfield_seq" is 
automatically created, with start value set to "1".

CREATE SEQUENCE "public"."testtable_serialfield_seq"
    INCREMENT 1  MINVALUE 1
    MAXVALUE 9223372036854775807  START 1
    CACHE 1;
And now I have to insert into this table some records from other rdbms 
(like MySQL) using sql inserts generated by e.g. mysqldump - it is a 
simple import of data, where "serialfield" has got some value...

INSERT INTO testtable VALUES (1, 'sdsdsd');
INSERT INTO testtable VALUES (5, 'sdsdsd');
INSERT INTO testtable VALUES (10, 'sdsdsd');
And my question is: how sequence should react for this ? Right now start 
value of the sequence is still "1", so if I will make a query like 
"INSERT INTO testtable (someotherfield) VALUES ('sdsdsd')" then new 
record's serialfield will have value set to "1" - of course this record 
will not be created, because of primary key - but next insert will 
create a new record with serialfield set to "2".

I didn't notice this problem before, because I have never made any 
imports of data in this way..

Window/Cygwin
PostgreSQL 8.0.0beta3
Regards,
ML

---------------------------(end of broadcast)---------------------------
TIP 6: Have you searched our list archives?
              http://archives.postgresql.org