[OPENJPA-2735] IN with sub-query doesn't work with compound IDs

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 2.4.2
• Component/s: jdbc
• Assignee: Mark Struberg
• Reporter: Pawel Veselov
• Created: 25/Apr/18 13:04
• Updated: 14/Feb/19 14:20

Description

Ran into this problem now.
With two entities, A and B, where A has multi-field ID (say id1 and id2), and B has a field that references A. Running this query:
select b from B b where b.ref in ( select a from A ); causes problems.
I don't know how to create a test case for it, because this outright doesn't work with Derby - derby doesn't allow for sub-queries that return more than one field.
With postgres, this fails, because the query translates into:

select * from b where b.ref_id1 in ( select a.id1, a.id2 from a );

The problem is in InSubQExpression.appendTo() method:

    public void appendTo(Select sel, ExpContext ctx, ExpState state, 
        SQLBuffer buf) {
        InSubQExpState istate = (InSubQExpState) state;
        _sub.calculateValue(sel, ctx, istate.subqState, null, null);
        _val.calculateValue(sel, ctx, istate.valueState, null, null);
        _val.appendTo(sel, ctx, istate.valueState, buf, 0);
        buf.append(" IN ");
        _sub.appendTo(sel, ctx, istate.valueState, buf, 0);
    }

The _val.appendTo() is called with index of 0, but that can be a compound object (which it is in this case). I'm gonna try replacing the appendTo() call with appendType(). I don't understand enough of this to know what I'm doing though.