Basically, if a test combined both NA and NaN in the same sum/mean/prod calculation then the result is essentially up in the air as to whether it will be another NA or NaN.
While I was actually trying to test this difference before by using expect_identical(), it isn't really worth the effort since the base equivalents don't make any promises about it either.
The easiest fix is to just switch to expect_equal() where NA_real_ and NaN are considered equivalent.
Closes #198
Basically, if a test combined both
NA
andNaN
in the samesum/mean/prod
calculation then the result is essentially up in the air as to whether it will be anotherNA
orNaN
.While I was actually trying to test this difference before by using
expect_identical()
, it isn't really worth the effort since the base equivalents don't make any promises about it either.The easiest fix is to just switch to
expect_equal()
whereNA_real_
andNaN
are considered equivalent.