I am amazed that noone has had a thought on the business problem here.
Frankly, I'm amazed that anyone expects exact results from a float-point value. By its very nature it is an approximation. The issue is not that the narrowing/widening behavior changed since it was ALWAYS allowed to, per specification. The issue is not that you can't get repeatable results since you ALWAYS could do so using explicit casts, and ONLY by doing so would you have _reliably_ gotten such results in the presence of mix-resolution values (Single v. Double). The issue is that you should NOT use floating-point values when you need precise decimal representation; if you need that, use Decimal. The correct reason for using floating-point values is when you need to deal with a large _RANGE_ of values in terms of absolute magnitude, not "just" to get decimal places. If we want to fix anything, let's not worry about enabling people to continue using the wrong datatype. This is just like when the designed the CLR to store character data as Unicode natively, and told people that wanted to play with the bytes directly to get them from the appropriate encoding. Without that design decision, we would be dealing with tons more buffer overflows, encoding breaks, etc. -- "One of the main causes of the fall of the Roman Empire was that, lacking zero, they had no way to indicate successful termination of their C programs." --Robert Firth Marc C. Brooks http://musingmarc.blogspot.com =================================== This list is hosted by DevelopMentor® http://www.develop.com View archives and manage your subscription(s) at http://discuss.develop.com
