Comparing float values about aunit HOT 4 CLOSED

Hi,

Thanks for using AUnit and posting this question. I should have added some notes about this in the README.md.

You are correct that AUnit does not provide the equivalent of ASSERT_FLOAT_EQ and ASSERT_DOUBLE_EQ from Google Test. The reason is that floating point comparison is subtle and complex. The Google Test ASSERT_FLOAT_EQ() interprets equality to mean 4 ULPs (Units in the Last Place) from each other. The definition of Units in the Last Place is incredibly complicated.

For embedded environments like Arduino, floating points are not used often so it didn't seem worth spending a lot of time on this.

The ASSERT_NEAR() macro from Google Test seems much easier to implement and is the same as your workaround. If this functionality is good enough for you, I think I could get that incorporated without too much effort. I would probably name it assertNear(val1, val2, error) in AUnit for consistency. I would implement it for floating types, and extend it to integer types if it's straightforward.

from aunit.

Yes I think that adding assertNear would be good enough.
I'm now wondering if that could be useful also for integers comparison

from aunit.

Please take a look at assertNear() and assertNotNear() in v1.1 that I just released. They work for integral types, but I suspect that they will be mostly used for floating point types.

from aunit.

Thanks

from aunit.