4 Oct
2017
4 Oct
'17
8:32 a.m.
On 10/04/2017 03:59 AM, Baa wrote:
.. 100% boolean coverage (0/0) ..
but 0/0 =/= 100% (an undefined limit ;) IMHO better is to be reported as 0. Is it a bug in HPC?
For all tests t, t passed -- vacuous truth to the rescue! Or, 100% of zero is zero. Don't think too hard about it =)
From a practical standpoint, you probably don't want your integration tests to reject a commit because you have 0% coverage on something that doesn't exist. For that reason, a "success" result is better when you have no coverage of no things.