Select statement problem. how to get data in one row

Question

SELECT
razm.datum_sklenitve,
razm.datum_prenehanja,
FROM
zket_zaposlenci zap,
zket_delovna_razmerja razm,
zket_delovne_razporeditve razp,
zket_sistemizacije_dm sistdm,
zket_organizacijske_enote oe,
zket_delovna_mesta delm
WHERE
zap.iid_zaposlenca = razm.iid_zaposlenca and
razm.iid_delovno_razmerje = razp.iid_delovno_razmerje and
sistdm.iid_sistemizacija_dm(+) = razp.iid_sistemizacija_dm and
nvl(sistdm.iid_organizacijske_enote, razp.iid_organizacijske_enote) = oe.iid_organizacijske_enote and
sistdm.iid_delovno_mesto = delm.iid_delovno_mesto and
razm.datum_sklenitve <= to_date('31.12.'||:v_datum, 'dd.mm.yyyy') and
nvl(razm.datum_prenehanja, to_date('31.12.'||:v_datum, 'dd.mm.yyyy')) >= to_date('01.01.'||:v_datum, 'dd.mm.yyyy') and
razp.datum_nastopa_dela <= to_date('31.12.'||:v_datum, 'dd.mm.yyyy') and
nvl(razp.datum_prenehanja_dela, to_date('31.12.'||:v_datum, 'dd.mm.yyyy')) >= to_date('01.01.'||:v_datum, 'dd.mm.yyyy') and
zkep_splosno.zkep_hierarh_oe_n(nvl(sistdm.iid_organizacijske_enote, razp.iid_organizacijske_enote), :v_oe, to_date('31.12.'||:v_datum, 'dd.mm.yyyy')) = 1 and
to_char(zkep_splosno.zkep_vrni_prvi_dr(razm.iid_delovno_razmerje, to_date('31.12.'||:v_datum, 'dd.mm.yyyy')), 'yyyy') = :v_datum
and zap.iid_zaposlenca = 1093507
order by razm.datum_sklenitve DESC
return this rows:
datum_sklenitve     datum_prenehanja
1: 01.10.2010 00:00:00 | 31.12.2010 00:00:00
2: 01.10.2010 00:00:00 | 31.12.2010 00:00:00
3: 01.04.2010 00:00:00     | 31.08.2010 00:00:00
4: 01.02.2010 00:00:00     | 31.02.2010 00:00:00
how can i get in one select just this result.
1: : 01.10.2010 00:00:00 | 31.08.2010 00:00:00 (first of A and then the next one of B which is not the same as top of B)
I try everything but i do not know how. If anyone know how to do this.
regards,
MB

Answer 1

Simma wrote:
But the above is not documented.Not only that, but it's string aggregation which isn't meeting the OP's requirements.
Though you could be easily forgiven as the OP hasn't clearly stated the logic behind the requirements.
It could be something like wrapping the original SQL with something like the following...
SQL> ed
Wrote file afiedt.buf
  1  with t as (select date '2010-10-01' as ds, date '2010-12-31' as dp from dual union all
  2             select date '2010-10-01', date '2010-12-31' from dual union all
  3             select date '2010-04-01', date '2010-08-31' from dual union all
  4             select date '2010-02-01', date '2010-02-28' from dual)
  5  --
  6  -- END OF TEST DATA
  7  --
  8  select max(case when rn = 1 then ds else null end) as ds
  9        ,max(case when rn = 2 then dp else null end) as dp
10  from (
11        select ds, dp, row_number() over (order by ds desc nulls last) rn
12        from (
13              select case when lag(ds) over (order by ds desc) = ds and lag(dp) over (order by ds desc) = dp then null else ds end as ds
14                    ,case when lag(ds) over (order by ds desc) = ds and lag(dp) over (order by ds desc) = dp then null else dp end as dp
15              from t
16              order by t.ds
17             )
18*      )
SQL> /
DS                  DP
01/10/2010 00:00:00 31/08/2010 00:00:00
SQL>but who knows?