[MLton-commit] SF.net SVN: mlton:[7571] mlton/trunk/mlton/atoms/real-x.fun

[<fl...@us...>]

Revision: 7571
          http://mlton.svn.sourceforge.net/mlton/?rev=7571&view=rev
Author:   fluet
Date:     2012-06-11 20:57:21 +0000 (Mon, 11 Jun 2012)
Log Message:
-----------
Revert RealX.equals to return false for two nan values.

nan values represented by distinct bit-patterns are not equal.  There
are a number of functions (MLton.Real{32,64}.castToWord, Real.signBit,
Pack{Real,Real32,Real64,LargeReal}{Big,Little,Host}.*) that reveal a
floating-point value's bit pattern.  In particular, the following
should result in two distinct nan values with different sign bits:

val nan = posInf + negInf
val pnan = copySign (nan, 1.0)
val psb = signBit pnan
val nnan = copySign (nan, ~1.0)
val nsb = signBit nnan

However, if these are constant folded and common-subexpression
eliminated (with RealX.equals returning true for two nan values), then
psb and nsb will (incorrectly) be equal.

Modified Paths:
--------------
    mlton/trunk/mlton/atoms/real-x.fun

This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.


------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
MLton-commit mailing list
MLt...@li...; mlt...@ml...
https://lists.sourceforge.net/lists/listinfo/mlton-commit