www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.bugs - [Issue 813] New: regression: optimizer & scope

reply d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=813

           Summary: regression: optimizer & scope
           Product: D
           Version: 1.00
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P3
         Component: DMD
        AssignedTo: bugzilla digitalmars.com
        ReportedBy: thomas-dloop kuehne.cn


http://dstress.kuehne.cn/run/s/scope_16_A.d











fails only if compiled with "-O"


-- 
Jan 07 2007
parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=813


bugzilla digitalmars.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID





The problem here is that 2.2 is not representable exactly. Constant folding
happens in different ways depending on if the optimizer is run or not,
resulting in slightly different answers due to roundoff error. Generally
speaking, an algorithm that depends on this is always going to have trouble.

To make the example work, use a constant that is exactly representable, like
1.25 and 2.25.

Not a compiler bug.


-- 
Feb 02 2007