Some Thoughts on C# 6

I'm a Programming Languages freak. Though I've never been into compilers programming, I've ever had an enormous interest in their input, languages (and even their output, MSIL and so on). This explains why I follow with much attention the development of ES6, Java 8, anything related to Groovy... and I've been quite upset by the lack of news regarding C# 6.

Hopefully, this summer we got some information from the very Anders. There he mainly highlighted the importance of Roslyn, how the next C# will come with the new managed C# compiler (as a service), and how this will allow for further development of the language in the future. I quite like what he mentions about his work on TypeScript making him more appreciative of Dynamic features (I'd like to see the astonishing dynamic features added to C# in v.4 getting some more love in the near future)

Last week some more news about probable new features in C# 6 were released on the web. Well, quite a bunch of things, but in first instance I felt a bit disappointed.
The one I'm mostly interested in is that they call (oddly enough) "Monadic null checking" and that is just the equivalent to Groovy's Safe Navigation Operator. I've really missed this feature so many times that I even wrote a sort of workaround for it both for C# and JavaScript.
On second thought, I'll say that Read Only Autoproperties and Property Expressions seem also rather appealing to me, but anyhow, as some people mention in the comments, these are mainly cosmetic/wrist friendly additions, but nothing involving any real change. Some people mention that if you want a really modern language for the CLR, you should switch to F#. I don't like the idea, though multiparadigm, F# favors Functional Programming over OOP, so it involves a sort of paradigm shift that I haven't done yet (I'm used and rather keen of the soft functional features in JavaScript and C#, but I really need to devote some time to understand the advantages of a more purely functional approach).

All this has made me think about what features I would add to a modern language like C# (or Java 8), and well, indeed, there are not so many. C# is (for the most part) a mainly statically typed language, so some of the things I love in JavaScript (or Groovy) can't easily be applied/used (I'm referring to expanding classes or instances with new methods or altering the inheritance chain). Also, I'm not sure to what extent Python's metaclasses (little to do with Groovy's ones) could be used. Anyway, after some pondering, I've come up with this (wish) list:

• Some kind of Multiple Inheritance of Behaviour, like the one enabled by Java 8 with its Default Interface Implementation. Scala traits seem even more powerful, so they would be more than welcome (and would probably involve just some more compiler magic, but no IL changes).
• When writing this post, I realized that Iterators (Generators in normal Programming Languages parlor) could gain some more power (send, throw, close, delegated yield).
• Some more investment in the DLR and dynamic. Not sure if at this point (where Lightweight Code Generation has been superseded by Expression Trees in most cases) extending the access to the IL code could get us any benefits.
• Allow Attributes (Annotations in the Java world) to receive delegates (lambdas) as parameters. This is possible in Groovy land, but in C# all you're left with is an error prone hack.

In terms of the CLR, I think Microsoft have got it quite wrong with TypeScript. They've decided to follow that (to my mind utterly stupid) trend of avoiding JavaScript and writing something that compiles to it. Sure JavaScript is riddled with some quirks and gotchas, but it's one of the most powerful languages ever designed, and what we need is speeding up the development and adoption of ES6, as it fixes many of those problems. So, what I'd like to see is a modern JavaScript engine for the CLR (something like Oracle's Nashorn).