Redshift 中的 update + case + join 语句

Question

I have this code from SQL Server that I have to implement in Redshift.

update t1 set some_flag = (case when b.some_value is null then false else true end)
from t1 a
left join t2 b on a.code = b.code;

I get an error in Redshift that says:

 Target table must be part of an equijoin predicate

How can I amend the above to make it work in Redshift? Thanks.

Answer 1

Try removing the reference in the from clause:

update t1
    set some_flag = (case when b.some_value is null then false else true end)
from t2 b 
where t1.code = b.code;

Postgres (and Postgres-derived databases) allow a FROM clause for UPDATE . However, the table references are in addition to the referenced table in the UPDATE . This differs from SQL Server, where the table reference in the UPDATE is assumed to come from the FROM clause if possible -- even ignoring the alias if necessary.

You should be able to simplify this to:

update t1
    set some_flag = (b.some_value is not null);
from t2 b 
where t1.code = b.code;

Answer 2

You must fully qualify the table name with column instead of alias. Redshift does not support table alias in update command i guess.

update merge_demo1
set lastname = (case when merge_demo1.lastname = merge_demo1.lastname then merge_demo1.lastname else merge_demo2.lastname end)
from merge_demo2
where merge_demo1.id = merge_demo2.id;