eokumus79@gmail.com wrote:

> A B
> 1 10000
> 1 10001
> 1 10002
> 1 10006
> 1 10007
> 2 10008
> 2 10009
> 2 10010
> 2 10011
> 2 10012
> 2 10015
> 3 10021
> 3 10022
> 3 10023
> 3 10031
> 3 10032
> 3 10033
> 3 10034
> 3 10035
> 4 10041
> 4 10042
> 4 10044
>
> i want to get following output
>
> A B C
> 1 10000 10002
> 1 10006 10007
> 2 10008 10015
> 3 10021 10023
> 3 10031 10035
> 4 10041 10042
> 4 10044 10044

select
a,
min(b),
max(b)
from
(
select
a - rank() over (partition by A order by b asc) as grp,
a
from mytab
) dt
group by a, grp
order by 1,2;

If a/b is not unique, replace rank with dense_rank...

Dieter