How can I find out if a Float value is a negative zero (and not a positive one)?
Unfortunately:
-0.0 == 0.0 # => true
-0.0 === 0.0 # => true
My initial solution works but is ugly:
x.to_s == '-0.0'
From this question, I found
x == 0 and 1 / x < 0
Is there a better, more Ruby-like way?
Ruby's BigDecimal class has a sign method that produces the correct result for negative zero. You can convert a Float to a BigDecimal with the to_d method if you require 'bigdecimal/util'.
require 'bigdecimal'
require 'bigdecimal/util'
0.0.to_d.sign
#=> 1
-0.0.to_d.sign
#=> -1
Combine this with zero? and you're good to go:
def negative_zero?(x)
x.zero? && x.to_d.sign == -1
end
negative_zero?(0.0)
#=> false
negative_zero?(-0.0)
#=> true
The angle method (and it's aliases arg and phase) returns zero for positive floats and Pi for negatives.
p 0.0.angle #=> 0
p -0.0.angle #=> 3.141592653589793
In Ruby the Float equality operator for -0.0 and 0.0 returns true, as per ordinary arithmetic.
However if you convert the two floats to bytes using little-endian or big-endian byte order, you'll see they do not in fact match.
[-0.0].pack('E')
#=> "\x00\x00\x00\x00\x00\x00\x00\x80"
[0.0].pack('E')
#=> "\x00\x00\x00\x00\x00\x00\x00\x00"
[-0.0].pack('E') == [0.0].pack('E')
#=> false
If your purpose is to prevent "negative zero", then this is how rails does it:
number = number.abs if number.zero?
Cause ruby determines them as the same object the only way to detect it by "-" sign after string conversion, as you described: -0.0.to_s.start_with?('-').
