Kathy Kam

As a C# developer, I don't really mind Nullable<T>, since I tend to use T? anyway.  It's a shame VB developers have to put up with Nullable(Of T) for now.  Then again, they put up with the verbose casting syntax.

Collection<T> is pretty good, too.  I wouldn't like CollectionBase<T> as much because it's not a clean name when used directly (without inheriting from it).

The nice thing about Collection<T> is you can have an internal class FooCollection that inherits from Collection<Foo>, but expose your collection property as Collection<Foo>.  If you were certain you were never going to add any public methods, this makes for a very clean API (no cluttering the namespace with collection classes that nobody ever codes against directly).

Still, I tend to use FooCollection anyway since it versions better.  I'm not sure, but changing a non-virtual read-only property to a derived type is still a breaking change.  Maybe co/contra-variance needs to be added to runtime versioning support.

Once again, I am very excited that the new CTP is available. :) My second feature for Orcas is available

Collection<T>, although a useful class, is horribly named. "Collection" is a very general term for a structure which holds things (see ICollection and ICollection<T>). Collection<T> is essentially a List<T>, with some overrides included for customizing the behavior if desired. That means it is basically a slightly MORE specific version of List<T>. But it has a LESS specific name. Am I the only one who thinks this makes no sense whatsoever? What was the justification for calling it Collection<T> in the first place?