Group by preserving the order

Question

Say for eg: my data is like this
--order by col3
col1 col2  col3
1       5     1
1      10    2
2      2      3
2      5      4
1     15     5
I hav to get o/p as
col1  sum_col2
1       15
2       7
1      15
I have to group by preserving the order
Pls give me suggestions..
jeneesh                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   

Answer 1

I have reserched solution which this problem can solve.
Therefore, I created this solution.
Below solution consults "Tabibitozan".
"Tabibitozan" is Japanese language.
"Tabibitozan" is one of math problem. (http://www.manabinoba.com/index.cfm/4,757,73,html?year=2002)
I do not know what "Tabibitozan" is called in English.
with ValTable as(
select 1 ID,5 Val,1 SortOrder from dual
union all select 1,10,2 from dual
union all select 2,2,3 from dual
union all select 2,5,4 from dual
union all select 1,15,5 from dual
union all select 3,25,6 from dual
union all select 3,10,7 from dual
union all select 3,5,8 from dual
union all select 3,15,9 from dual
union all select 4,5,10 from dual)
select ID,sum(Val) as Val
from (select ID,Val,SortOrder,
       Row_Number() over(order by SortOrder)
      -Row_Number() over(partition by ID order by SortOrder) as makeGroup
      from ValTable)
group by ID,makeGroup
order by min(SortOrder);
ID  VAL
1   15
2    7
1   15
3   55
4    5In above solution,
I imagined that there are 5 people called X,A,B,C, and D.
X always walks 1m. (Row_Number() over(order by SortOrder))
When ID = 1, A walks 1m. (Row_Number() over(partition by ID order by SortOrder))
When ID = 2, B walks 1m. (Row_Number() over(partition by ID order by SortOrder))
When ID = 3, C walks 1m. (Row_Number() over(partition by ID order by SortOrder))
When ID = 4, D walks 1m. (Row_Number() over(partition by ID order by SortOrder))
Then,
I derived distance between X and A.
I derived distance between X and B.
I derived distance between X and C.
I derived distance between X and D.
Then, I made group using these distance. (group by ID,makeGroup)
similar threads
SQL - Hierarchical View
Version Key creation
Group by preserving the order
Is it possible to reset agg function row_number()