qualify rank() over (partition.....question - response (15) by ficklingIan

Hi
I have tried the above code but I dont seem to be getting the expected results. Example table
myTab(c1 varchar(10), c2 varchar(10), c3 varchar(10), sd date ed date);
example
data row1: 'A','B','C','2012-01-01','2012-01-02'
data row2: 'A','B','C','2012-01-03','2012-01-09'
data row 3 'A+','B','C','2012-01-10','2012-01-19'
data row 4 'A','B','C','2012-01-20','9999-12-31'
 
What I expected back was three rows in which data row 1 and data row2 are 'combined', however this is not the case.
The query I use is :-

Select c1
      ,c2
      ,c3
      ,sd
      ,coalesce(min(prev_end_time)
           over (partition by c1,c2,c3
                 order by sd
                 rows between 1 following and 1 following), max_end_time) as end_time
From 
     (
       Select c1
             ,c2
             ,c3
             ,sd
                   ,max(ed) 
                       over (partition by c1,c2,c3
                               order by sd
                               rows between unbounded  preceding and 1 preceding) as prev_end_time
                   ,max(ed) over (partition by c1,c2,c3) as max_end_time
        From insight.if_test1
        qualify sd > prev_end_time  or prev_end_time is null
       ) as dt

H E L P