Script to reproduce the problem. It drops the tables at the end of the
script.
Stephan Szabo wrote:
>
> There was a bug (which should be fixed for 7.1) that got the
> arguments wrong for the alter time check of the existing data.
> I think I should be able to get a patch together to fix it once
> I get a copy of the 7.0.3 source.
>
> Can you send the table schema as well so I can test it out?
>
Script to reproduce the problem. It drops the tables at the end of the
script. If data is not inserted into mtable there is no problem.
========================================================================================
create table utable (
uk int not null,
pk int not null,
thetext text,
primary key (uk, pk)
);
create table mtable(
id serial not null primary key,
mtext text,
o int, /* corresponds to uk */
p int /* corresponds to pk */
);
insert into utable (uk,pk,thetext) values (2,4,'blah 2 4');
insert into utable (uk,pk,thetext) values (2,5,'blah 2 5');
insert into mtable (mtext,o,p) values ('m 2 4',2,4);
alter table mtable add FOREIGN KEY (p,o) REFERENCES utable (pk,uk);
drop sequence mtable_id_seq;
drop table mtable;
drop table utable;
--
Joseph Shraibman
[EMAIL PROTECTED]
Increase signal to noise ratio. http://www.targabot.com
