Question on correlated update

Question

Greeting gurus. I expect the following demo will get id populated as 1.2.3.....100 but its all 1. why or why not working ?
create table ynot ( id number, sd date)
insert into ynot(sd)
select sysdate + rownum from dual connect by level < 101
select * from ynot
update ynot xx set xx.id = ( select row_number() over (partition by null order by sd ) from ynot yy where xx.sd = yy.sd )
select * from ynot

Answer 1

Table ynot has 100 different values for sd (today +1 to today +100)
The subquery asks for rows
from ynot yy where xx.sd = yy.sd
There can only be one row from ynot where sd matches the value in the current xx.sd, so the subquery only finds one row.
E.g. when updating the row where sd = sysdate + 50 the correlation value xx.sd is sysdate + 50 and there is only one row with that value, so the
row_number() over (partition by null order by sd )
is counting rows from a set of 1 rows, which is why it always gives 1 as the result.
You seem to be expecting the analytic clause to look outside the subquery, but it can't do that.
One way to do what I think you want to do is via a  merge:
merge into YNOT dst
using (select rowid rid, row_number() over (partition by null order by sd ) rnum  from ynot) src
on ( dst.rowid = src.rid )
when matched then update set id = src.rnum