Re: sql return only first matching record from join table

Not the most efficient approach, but it might help to get the creative juices flowing, so here goes with a correlated subquery:

select a.address, o1.occupant
from address_table a, occupant_table o1
where o.address_id = a.address_id
and o.rowid = (select min(rowid)

                        from occupant_table o2
                        where address_id = o1.address_id)

Should be OK performance if you're not dealing with a large number of rows. You could also use occupant_id instead of rowid as long as it's unique.

Martin

Barnoit wrote:

> Sorry, I guess I should have been more specific in my post. I realize
> that a relational table is unordered, I simply meant the first row
> returned, I'm not meaning the first record stored. But more
> importantly, when I simplified my example, I guess I failed to mention
> that this query can and most often will return more than one row, so
> using rownum = 1 will not work. It is only on that specific join
> column that I used in the example where I want to return only one
> record. I realize it might seem odd, but that is the nature of the
> datamodel I have been supplied (and cannot alter). So a query might
> return more than one row, but I only want it to give me one record
> when it does the join, even though more than one record may satisfy
> the where clause. Here is the real scenario, no simplification. I have
> a query that receives user input, which is a parcel (or parcel id -
> like a city block). On that parcel there may be 1 or more addresses.
> At each address there may be one or more occupants. I won't get into
> the details of the tables, but there are numerous joins. What I want
> then is when the user provides a parcel to query on, I want it to
> return me all the distinct addresses for that parcel with only one
> occupant (I don't care which one) per address. Sorry to be so long
> winded, but hopefully this should avoid confusion.
>
> Thanks in advance for any help.
>
> "Richard Foote" <richard.foote_at_bigpond.com> wrote in message news:<IBDU8.26883$Hj3.82905_at_newsfeeds.bigpond.com>...
> > Hi Barnoit,
> >
> > Simply add 'and rownum=1' to the where condition. Oracle will then only
> > return the first row that meets the other conditions.
> >
> > Good Luck
> >
> > Richard
> > "Barnoit" <jbarney_ca_at_yahoo.ca> wrote in message
> > news:cd2c60c8.0207030600.39d024b8_at_posting.google.com...
> > > Just to simplify my example, I have 2 tables that are joined on a key
> > > id (there are actually quite a few joins, but this is the one causing
> > > me trouble). The keyid on the first table is unique, but there is a
> > > one to many relationship in the second table (it is actually a street
> > > address table and an occupant table). What I need to be able to do is
> > > to a select using the keyid as the join criteria, but I only want to
> > > return the first matching record and disregard any other matches. So
> > > just to reconfirm, I want to do a select based on an address, join
> > > based on the keyid and return the name of the first person that it
> > > finds, disregarding any other occupants it finds. Any suggestions
> > > would be greatly appreciated.
> > > ex.
> > > select table1.address, table2.name
> > > from table1, table2
> > > where address = '123 Main St'
> > > and table1.keyid = table2.keyid
> > >
> > > What I am getting is
> > > 123 Main St, Joe Blow
> > > 123 Main St, John Q. Public ...
> > >
> > > but what I really want is whatever the first match is. I can't use
> > > pl/sql in this case, has to be sql only. I'm hoping I'm just
> > > overlooking something very simple, and BTW this is on oracle 8.0.5