Complex insert clause for a btree cluster

Question

Hi,
I am reading Tom Kyte's book on Effective oracle by design -having difficulty in understanding of how to load data into a btree cluster. Tom says that to get full benifit of cluster, the data should be inserted in the tables (involved in the cluster) at about the same time. So he uses a rather complex insert statement which goes like:
insert
when (r=1) then
into table 1
values....
when (1=1) then
into table2
values
<select query data of which is to be inserted> - this query has:
row_number() over (partition by owner order by object_id) r
/I am not able to understand this syntax - what does when r=1 mean? does it mean only one row will go in first table? why in second table the condition should be when (1=1)? Taking a simple of example of EMP and dept table if I have to create a btree cluster on deptno, can someone show me the syntax and explain how it works?
Thanks,

Answer 1

orausern wrote:
Hi,
I am reading Tom Kyte's book on Effective oracle by design -having difficulty in understanding of how to load data into a btree cluster.I am sure Tom has provided all necessary scripts in order to
a) reproduce his test case and
b) verify the results
Did you run those scripts yourself?
As for understanding INSERT syntax, the documentation is more than sufficient, I guess.
Basically, the SELECT (in the INSERT) is joining header table (DEPT) to details table (EMP). This will generate resultset which contains data from both the tables such that header record (DEPT) is repeated for each detail record (EMP). The conditions in INSERT will be applied on the rows of this resultset. In order to insert only unique header records, the condition (r = 1) is used whereas in order to insert all detail records, the condition (1 = 1) will be used.