Question

I have the following example code:

create table #tempmembers (
    memberid int
)

update Members set 
Member_EMail = NULL
where Member_ID in (select member_id from #tempmembers)

The subselect contains an error, since #tempmembes does not contain a column named member_id, but the sql statements run WITHOUT any errors and update no rows.

If I then add just ONE row to #tempmembers:

create table #tempmembers (
    memberid int
)
insert into #tempmembers select 1

update Members set 
Member_EMail = NULL
where Member_ID in (select member_id from #tempmembers)

it still runs without any errors – but this time ALL records in Members will be affected.

Why does the SQL statement not fail completely? And if the failing subselect is evaluated to NULL – should updating all rows in Members not only occur if it had been:

update Members set 
Member_EMail = NULL
where Member_ID not in (select member_id from #tempmembers)

Answer 1

It’s inheriting member_id from the outer query so is equivalent to:

   ...
   (select Members.member_id from #tempmembers)

This will fail as expected:

   ...
   (select #tempmembers.member_id from #tempmembers)