Hi, I've brought up the idea of having a more powerful replacement for std::numeric_limits a while ago. I've now gotten around to drafting up a paper. Let me know what you think of the direction.

In a nutshell, there is a lot of information that std::numeric_limits doesn't provide, and we cannot fix that because it would break users' specializations. Having more floating-point information would enable users to implement a lot more things portably using the information they get from the standard library.