Postgresql update each row with a random value

I think that Postgres might optimize the subquery. Doesn’t this work?

update foo
   set bar = floor(random() * 9 + 1);  -- from 1 to 10, inclusive