[p2t2p]

I would say "it depends". I have seen when replacing bunch of joins with sub selects improved performance. I have seen otherwise as well. It all depends on the nature of your data, RDBMS you're using and statistics it has accumulated.

I haven't seen anybody hitting "in" clause limits in query like mine but we did hit such a limit once when we were passing a hardcoded list of IDS from client side. Something like "where blah in (id1, id2, id3.... idn)". I remember that one failing on MS SQL.