They really just need to let us use C# and F# in the same sln/projects. Then structure can be in C#, nitty gritty implementations in F# and everyone is happy!
>C# seems to look more like Scala with every release
It does, but then that begs the question of "why use c# at all?". C# is still missing (along with F#) scala's most powerful features, and I prefer the JVM and openness of the ecosystem to .NET
The contact I'm on, we're having a lot of success porting away a large C# application to Scala.
You also get a step-wise path to rewrite the applications. Going to Scala would probably be a complete re-write which is risky and time consuming. I don't think Scala usage will grow much more now that Java 8 is released. The benefits of rewriting a Java application to Scala is probably not that great in comparison to the cost. And the C# to Scala re-write is probably even worse from a cost/risk perspective.
No, I'm pretty sure the reason they dropped CLR support was that there was no good way to get a decent .NET interop story with a language with a substantially different type system than .NET assumes (particularly, with higher-kinded types) given .NETs reified generics.
This is one respect in which .NET's superiority over the JVM as a platform for its core languages (C#/VB.NET on .NET vs. Java on the JVM) made it worse for flexibility. Reified generics are great for implementing languages with a C#-like-type-system, but they limit the power of the type systems you can have.
F# is a remarkably less powerful language than scala. Along with the cost of Visual Studio + windows licenses, the Scala/open source ecosystem is just a better choice.
EDIT: Interesting to see Neal Gafter is involved (See Roslyn mention here: http://www.gafter.com/~neal/), who is rather fond of Scala from what I've read: http://www.infoq.com/articles/neal-gafter-on-java