I believe that I sent a patch on Sep 17 for this to -patches although
I don't know if anyone saw it (it's in the archives, so I know it
went through).
Stephan Szabo
sszabo@bigpanda.com
On Mon, 16 Oct 2000, Bruce Momjian wrote:
> Has this been resolved?
>
> > On Fri, 15 Sep 2000, Tatsuo Ishii wrote:
> >
> > > It seems that foreign key does not work in current, if specified with
> > > primary key definition. Take a look at following example(works in
> > > 7.0.2.):
> > >
> > > test=# CREATE TABLE PKTABLE ( ptest1 int, ptest2 int, ptest3 int, ptest4 text, PRIMARY KEY(ptest1, ptest2,
ptest3));
> > > NOTICE: CREATE TABLE/PRIMARY KEY will create implicit index 'pktable_pkey' for table 'pktable'
> > > CREATE
> > > test=# CREATE TABLE FKTABLE ( ftest1 int, ftest2 int, ftest3 int, ftest4 int, primary key
(ftest1,ftest2,ftest3,ftest4), CONSTRAINT constrname3 FOREIGN KEY(ftest1, ftest2, ftest3) REFERENCES PKTABLE);
> > > NOTICE: CREATE TABLE/PRIMARY KEY will create implicit index 'fktable_pkey' for table 'fktable'
> > > NOTICE: CREATE TABLE will create implicit trigger(s) for FOREIGN KEY check(s)
> > > ERROR: columns referenced in foreign key constraint not found.
> >
> > Hmm, that's very strange. I wonder which columns it think didn't exist.
> > It shouldn't be checking the pktable in that case, which would imply
> > it doesn't believe the existance of ftest1,ftest2,ftest3. Probably
> > a stupid mistake on my part. As soon as I clear off space to compile
> > current, I'll look.