Paul Fultz II

Actually, another option could be for Fit to use Boost.Config when available and then fallback on another mechanism when its not. This would allow for Fit to be installed modularly...

Ok, I am using the sd6 macros and falling back on C++ version to detect C++ features. I think this is more robust than Boost.Config. Of course, there is no...

> which only checks whether _MSC_VER is defined, but not its value. What value should I check for? > In addition, checking _MSC_VER before __clang__ is wrong because clang-cl defines...

> which one did you test with when you added the workaround? I originally tested this with 14 and 15. It looks like the workaround stops working in 19.20 in...

> The "normal" (not Clang) implementation works under VS2017 Yea they updated VS2017 in appveyor and there have been test failures. I should probably drop this version. > It however...

Technically, they should go to `develop`, but right now i haven't been using the develop branch since there is not too much activity to warrant a separate branch. However, I...

I think this could be a good idea. A couple of things. It needs to keep the constraints of the static function, so using `FIT_RETURNS` needs to be used instead...

Thanks, I am going to be making a few API breaking changes in the next month or so due to feedback from the review, and then I'll let you know...

The nice thing about inheritance is easy access to the base class. I think it would be better to keep inheritance and the user can use `final` or `fit::protect` to...

For example, the `capture` decorator can't be implemented using `fit::decorate`. However, adding the second parameter to `decorate`, `capture` could be implements like this: ``` cpp auto capture = decorate([](auto&& p,...