Nick

@pcjmuenster : Open a PR and we can see what you're doing.

@jtlap : Do you mind posting the diff from your patch or making a PR? Thanks!

@jtlap : You have to fork the repo, then the PR must be "compared across forks". Your diff is sufficient for me to get it working, but of course I'd...

I don't understand this. There is no obligation to use precompiled headers from our end.

Actually all `int` comparisons are a cause of pain, since ideally we'd like the library to work with floats, multiprecision, autodiff variables, interval types, and now unit types.

Can you compare multiprecisions against any int quickly? Or just zero? I note that I have really great antipathy towards units libraries, because we've had the Buckingham Pi theorem for...

Yes, I think that's fine. Could you give me a main I could try to get working? BTW, I recommend pchip or makima interpolation over barycentric rational for almost all...

> What does everyone else feel about this as a breaking change? I think it should be done, and there should be a slow but steady commitment to doing the...

@jzmaddock : Wowsers the linear regression test looks bad. The Whittaker-Shannon also looks like it's beyond reasonable "fudge the tolerance bounds." Will try to look into this soon.

@duhadler : Thanks! I was just thinking last night that I needed to add some MPFR quadrature tests; looks like this will be a good opportunity to get that done....