The previous fix introduced a not-yet-understood regression in compiling
CodePeer. For now, we attempt a quick workaround for the problem.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* sem_util.adb (Build_Discriminant_Reference): In the unexpected
case where we pre
Given a discriminated type T1 with discriminant D1 having a component C1
of another discriminated type T2 with discriminant D2 and a propagated
discriminant constraint (that is, "C1 : T2 (D2 => D1);" and, for
example, a parameter of type T1, the compiler will sometimes build an
anonymous subtype to
