On Thu, Aug 02, 2018 at 06:40:56PM +0200, Andrea Bolognani wrote:
If all we achieve is reducing the depth by one for a single
test case, the additional complexity (not to mention breaking
the principle of least surprise) is not worth it: let's use
simpler, more predictable code instead.
This basically reverts fec6e4c48c9c (with a few adjustments).
Yeah, it doesn't make sense. I recall there was yet another reason for that and
it should've been used later on in a test that was never written (and me being
the only one interested in making that test didn't help). So:
Reviewed-by: Martin Kletzander <mkletzan(a)redhat.com>