Re: [HACKERS] Enhancing PGSQL to be compatible with Informix SQL

On Thu, 6 Jan 2000, Bruce Momjian wrote:

> > > Thomas has tried to explain the ANSI syntax for outer joins, and I must
> > > say I am quite confused by it. A simple OUTER added before the column
> > > name would be a quick and simple way to do outers, perhap get them into
> > > 7.0, and allow new users to do outers without having to learn the quite
> > > complex ANSI syntax.
> > >
> > > At least that was my idea.
> >
> > First, I'm for getting OUTER JOINs in ASAP...but, I'm a little concerned
> > with thought of throwing in what *sounds* like a 'stop gap' measure...
> >
> > Just to clarify..."A simple OUTER added before the column" would be a
> > PostgreSQL-ism? Sort of like Oracle and all the rest have their own
> > special traits? Eventually, the plan is to implement OJs as "SQL92 spec",
> > and leave our -ism in for backwards compatibility?
>
> Yes, OUTER is an Informix-ism. Oracle uses *=. I think the first is
> easier to add and makes more sense for us. *= could be defined by
> someone as an operator, and overloading our already complex operator
> code to do *= for OUTER may be too complex for people to understand.
>
> It would be:
>
> SELECT *
> FROM tab1, OUTER tab2
> WHERE tab1.col1 = tab2.col2

What about >2 table joins? Wish I had my book here, but I though tyou
could do multiple OUTER joins, no?

Marc G. Fournier ICQ#7615664 IRC Nick: Scrappy
Systems Administrator @ hub.org
primary: scrappy(at)hub(dot)org secondary: scrappy(at){freebsd|postgresql}.org