std::numeric_limits<fp>::min() vs. lowest()

[Klummel69]

According to https://en.cppreference.com/w/cpp/types/numeric_limits
min() and lowest() return different values for floating point types:

min():
Returns the minimum finite value representable by the numeric type T.
For floating-point types with denormalization, min() returns the minimum positive normalized value.
Note that this behavior may be unexpected, especially when compared to the behavior of min() for integral types._

fpm min() and lowest() return the same value (smallest negative number).
In my opinion, fpm min() should behave more like floating point min() here.

Is there a reason why min() is implemented this way?

[Klummel69]

Who can read...
Sorry, now I get it.

[MikeLankamp]

Indeed. To be clear (and for future readers), fpm's fixed-point type behaves like a floating-point type without denormalization so according to the letter of the spec, the floating-point rule for min() doesn't apply.

You could argue about the spirit of the spec and that fpm is supposed to pretend to be a float, but it seems that the reason the rule exists is to let users know when to expect a loss of precision when moving from normal into subnormal floating-point numbers. With fpm, you don't have this problem and you always have to be wary of precision when compared to floats.