tracker issue : CF-4199440

Problem Description: Elvis operator returns different results in CF 11 vs. CF 2016. The results under CF 11 make sense. Those under CF 2016 don't. We are having to back out our use of this operator.

Steps to Reproduce: Under CF 11 (Windows or Linux)  and CF 2016 run this:

 st = {a=1, b=0, c=false  };
    writedump( {
         a=st.a?:"null"
        ,b=st.b?:"null"
        ,c=st.c?:"null"
        ,d=st.d?:"null"
    });

CF11 Result: 

In the dump, CF 11 returns 1 for a, 0 for b, false for c, and null for d.

The results look good and make perfect sense. st.a is defined and equal to 1, st.b is defined and equal to 0, st.c is defined and equal to false, and st.d is undefined and, therefore, equal to null.

CF 2016 Result: 

In the dump, CF 2016 returns 1 for a (good!), null for b (what?!!!), null for b (again what!!!?), and null for d (I can't trust this!)

Any Workarounds:

Elvis has left the building for those of you born before 1970! I can no longer trust this operator.

Attachments:

Comments: