14 Apr
2010
14 Apr
'10
5:58 p.m.
On 2010-04-14 13:59, roconnor@theorem.ca wrote:
There is some notion of value, let's call it proper value, such that bottom is not one.
In other words bottom is not a proper value.
Define a proper value to be a value x such that x == x.
So neither undefined nor (0.0/0.0) are proper values
In fact proper values are not just subsets of values but are also quotients.
thus (-0.0) and 0.0 denote the same proper value even though they are represented by different Haskell values.
The trouble is, there are functions that can distinguish -0.0 and 0.0. Do we call them bad functions, or are the Eq instances for Float and Double broken? -- Ashley Yakeley