(j3.2006) (SC22WG5.6009) IEEE_INT

[John Reid]

WG5,

Do we have a problem with IEEE_INT (see 17.11.11 in N2137)? It has the form

     IEEE_INT (A, ROUND [, KIND])

where ROUND is of type IEEE_ROUND_TYPE, so has one of the values

IEEE_NEAREST, IEEE_TO_ZERO, IEEE_UP, IEEE_DOWN, IEEE_AWAY and IEEE_OTHER

The value of A is converted to an integer according to the rounding mode 
specified by ROUND using either convertToInteger{round} or the 
convertToIntegerExact{round} operation.

My version of the IEEE standard is the FDIS and provides the functions
ConvertToIntegerTiesToAway,
ConvertToIntegerTiesToEven,
ConvertToIntegerTiesTowardNegative,
ConvertToIntegerTiesTowardPositive,
ConvertToIntegerTiesTowardZero,
and the corresponding Exact versions.

It is clear what happens with ROUND of value IEEE_TO_ZERO and IEEE_AWAY.

I suppose IEEE_UP and IEEE_DOWN give ConvertToIntegerTiesTowardPositive 
and ConvertToIntegerTiesTowardNegative, that is, towards positive and 
negative infinity, respectively. I think this needs to be said.

Does IEEE_OTHER give ConvertToIntegerTiesToEven?

What about IEEE_NEAREST?

Cheers,

John.