Andrei Alexandrescu wrote: > Rainer Deyke wrote: >> Jason House wrote: >>> Do you have any common cases where a range would generate nulls as >>> part of its normal output? I think using the natural null feature of >>> reference types is a clean and reasonable solution. >> >> Iterating over an array that contains nulls? >> >> Nullable!(T) must be able to hold all values of T, plus null. By >> inference, Nullable!(Nullable!(T)) must be able to hold all values of T, >> plus two different null values. 'Maybe' in Haskell can do this. (The >> two null values are called 'Nothing' and 'Just Nothing'.) >> 'boost::optional' in C++ can do this. > > They can't store a nullable ref int. They'd have to store a nullable > pointer to an int. Then wait, there exists a null pointer to an int. > > Andrei
The way C# solves this is to have IsSet and Get properties or something.
