math.isFinite missing

[juancarlospaco]

math is missing the maximum representable finite float that is not +Inf (but math has the minimum).
It is not the same as +MinFloatNormal.

• https://github.com/nim-lang/Nim/blob/devel/lib/pure/math.nim#L124

I dont know if this is correct:

const MaxFloatNormal* = 1.7976931348623157e+308   ## Maximum representable finite `float`.

People find this useful, to check that your float did not gone Inf.

Other langs

• https://docs.python.org/3/library/sys.html#sys.float_info
• https://www.cplusplus.com/reference/cfloat

[timotheecour]

nim --eval:'import fenv; echo float.maximumPositiveValue'
1.797693134862316e+308

but maybe docs can improve cross-referencing

also this doc comment seems not the best description:

fenv module for handling of floating-point rounding and exceptions (overflow, zero-divide, etc.)

[juancarlospaco]

OK, maybe I am wrong, but then why MinFloatNormal is on math but not the Maximum?. 🤔

[timotheecour]

actually they're different:

nim --eval:'import fenv,math; echo float.minimumPositiveValue - MinFloatNormal'
4.940656458412465e-324

I'd rather not introduce an alias, but I'm all in favor for cross-referencing, and explaining in doc comment the difference.

[juancarlospaco]

I seen is important for end users to check if their float gone -Inf or +Inf, maybe 1 tiny proc to check can be added, isInf() ?.

[timotheecour]

let's add math.isFinite instead (logical to add after isNaN)

found in lots of langs:
https://dlang.org/library/std/math/is_finite.html
https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Number/isFinite
http://www.cplusplus.com/reference/cmath/isfinite/
https://numpy.org/doc/stable/reference/generated/numpy.isfinite.html
https://kotlinlang.org/api/latest/jvm/stdlib/kotlin/is-finite.html

isInfinite is less common IIUC (note that it's not opposite of isFinite because of NaN), because it's probably not more efficient than a == Inf or a == -Inf