copy_copyout/copyin_copyout depend on order of clauses

[alisterj1138]

The tests copy_copyout and copyin_copyout (C and C++) depend on the order of the data clauses, which the spec prohibits.

From v2.7 of the spec:

680 • A program must not depend on the order of evaluation of the clauses, or on any side effects
681 of the evaluations.

From v3.2 of the spec:

1056 • A program must not depend on the order of evaluation of the clauses or on any side effects of
1057 the evaluations.

    for(int x = 0; x < n; ++x){
        test[x] = 1.0;
    }
   #pragma acc parallel loop copy(test[0:n]) copyout(test[0:n])
   for(int x = 0; x < n; ++x){
        test[x] += 1.0;
   }

   for(int x = 0; x < n; ++x){
        if(fabs(test[x] - 2.0) > PRECISION){
            err++;
        }
    }

It's unclear what the best path forward is for these tests.

[ajarmusch]

Hi @alisterj1138,

In these tests, we were looking at what happens when users combine commonly used clauses. You're totally right that programs shouldn't depend on the clause order.

So in cases like copyin(a) copyout(a) or copyout(a) copyin(a), shouldn't those behave the same way? From our perspective, the clauses should be allowed in any order and it's up to the compiler to handle them correctly. That said, the spec doesn't full define what should happen in this case.

See the version 3.3 of the spec:

889 • The order in which clauses appear is not significant unless otherwise specified.
890 • A program must not depend on the order of evaluation of expressions in clause arguments...

Until the spec is clearer, we should add a description in the test. Thank you for taking a look.