Use isfinite and isnan provided by <cmath>

[a-andre]

The included provides std::isfinite and std::isnan. There is no need to fallback to C functions.

[svigerske]

The included provides std::isfinite and std::isnan.

With C++11 it became part of the standard, but before that?

There is no need to fallback to C functions.

But is there a harm in doing so?

[a-andre]

It seems I misinterpreted the configure file and assumed C++11 or newer was already enforced.

Anyway, C++11 seems well supported nowadays. So why not required it for the current development version of CoinUtils and simplify code where possible?

[svigerske]

I recall some discussions where a C++11 dependency was introduced the first time, and the decision back then was to not make it a requirement. In my recollection, the argument was that it would only be worth to go for C++11 if the whole code base would be cleaned up to use C++11 features. Increasing the C++ requirement just to save a few lines in CoinFinite.cpp wouldn't seem worth it.

[a-andre]

Based on 1a83f9b I assume the discussion happened 5 years ago. Things might have changed since then.

If there is a probability of being accepted, I can looking into making more use of C++11 (or newer?) features.

[svigerske]

Things might have changed since then.

Not in this corner of COIN-OR.

If there is a probability of being accepted, I can looking into making more use of C++11 (or newer?) features.

I don't think there is anything good at requiring higher C++ versions just because one could. If it were to provide some new interesting functionality or a (measurable) gain in performance, then that would be an argument. Also, if you wanted to take over the development of this project, then I would understand if you want a cleaner code base to work with.