C# is the past, the present and the future
blog.filipekberg.se
blog.filipekberg.se
"I spend most of my days working with the latest technologies from Microsoft."
"In the darkest alley-ways we still have the people not wanting to touch C# with a pitch-fork because it’s behind Microsoft"
In my opinion he lacks enough neutrality to make that statements(how he know MS tech is so great if he is totally ignorant of other options?).
Instead of telling people to learn only Microsoft tech he could take his own advise and learn non Microsoft tech too. The world is much bigger than a single company.
In my work as a software engineer I had to use what was good for my company, not what was good for Microsoft, Apple, or any religion(GNU or whatever).
I had to use Java, C#, objective C, python, (and lisp and c and c++). We will never be in the hands of a single company ever.
Finally, neither C# 3.0, 4.0 or 5.0 have been submitted to Ecma. So we can hardly refer to C# as an Ecma language for crying out loud! :)
Features of C# 3.0 and 4.0 have too been included in the ECMA standard. So yes, we can definitely refer to C# as an ECMA language...for crying out loud. The standard is not updated as often as Microsoft or Mono release a new version of their runtime (ohnoes!), just like web standards when a browser Mfr. comes out with a new feature.
In my view, the future of C# is hanging in the balance. Microsoft has lost significant mind share, and I am not sure if Windows is a good platform anymore to write new apps.
And yeah this isn't at all specific to C#. Anything not natively supported suffers - just ask Flash who in a handful of years have gone from a defacto standard to a dinosaur.
So let's give it a go. An opinionated past, present, and future for C#:
The past:
.NET emerged seemingly in 2002/2003 with a feature set similar to Java. This was pre-generics, and all code written in C# in this era is alien and involves a bit of casting.
Microsoft released Visual Studio 2002, containing support for (at least) Visual C#, VB.net, Visual C++ .NET (of course. let's blend two completely different object systems. my actual favourite: Objective-C++ .NET), real C++, and Visual J#, the successor to Visual J++ (which Sun had sued the crap out of them for). Of course you're not jealous of the JVM, Microdollar! We believe you.
C# did have structs at this point, distinct from classes, which are a pretty cool feature. They're more lightweight than classes, and I believe are passed by value rather than by reference like with classes.
Also, unsigned integer types. As a man who had to make a small machine 'emulator' in Java as part of a not incredibly well thought-out practical, I appreciate not having to bend my mind around poking the sign bit to try and get the real number I'm actually after.
And automatic unboxing and boxing of types was cool too. Int basically the same as int. So they did have a few original thoughts.
Event handling also deserves a mention, so there we go.
I personally was coding in VB6 at this time, and viewed VB.net with a mixture of bewilderment and contempt. So many keywords! Although the My.* namespace looked really cool. Actually being able to find out and change things about the current system, without being e.g. only allowed to change registry settings in a VisualBasic area of the registry.
That was about the time that I was messing around with threads thanks to the AddressOf operator, so really, a move was probably on the cards. (You had to bind the Windows threading API yourself, and when you tried to pause in the IDE it would only pause one thread, get confused, and crash.)
I eventually moved to VB.net (in the 2.0 era) and turned to treating C# with a mixture of bewilderment and disgust. (Clearly, I'm a quick learner.)
Java and .NET got generics around about the same time, a few years later. Java (sanely, to be honest) implemented generics via type erasure - so the compiler does the type checking, and then emits code that has no type information for generic containers - whereas the .NET team elected to break backwards compatibility* and make the runtime understand generics.
I think Mono appeared around this time.
This is the time when .NET started to get going, to be honest. Never settle for writing < 2.0 targetted code! Not that you can, unless you're running Visual Studio pre-2005.
XNA ran 2.0, although Microsoft's bored of trying to get people who aren't amateurs to use that now. Running my own code on the Xbox 360 was pretty cool though. But they seriously needed to put a better garbage collector on that thing. It was, er, rubbish.
* An annoying but possibly necessary decision. To this day, .NET 1.0/1.1 and 2.0+ apps require two separate runtimes to be installed, on Windows. Although Mono just deals with it, I think. Still, coming across the odd remaining 1.1 program can be a nasty surprise.
Man, why would anyone want any of this LINQ or closure stuff? I don't get it. We already have event handlers, and for blocks can do anything LINQ can do (that I'd use it for)!
It was about this time that I got into Haskell, somewhat out of necessity as that's what our uni course starts teaching. I've only grown more fond of Haskell since those days, mostly after learning a little OCaml/F# and seeing how comparatively nasty ML syntax is. Yeuch.
VB.net kinda lagged implementation-wise for these features vs. C# - I guess I wasn't the only person who moved from the former to the latter - but in C# they were fairly glorious.
No more would we be teased about the relative verbosity of our language! Fine, we still might go a bit long horizontally, but now we too can fit an amount of functionality on one line sufficient to guarantee the confusion of the reader and thus job security. Not that I have a job. Does job security keep you out of a job if you don't have one? Hm.
LINQ uses method extensions (a hilarious but occasionally useful bit of the language) to add new functions to all the collections types whenever LINQ is imported. Mapping and 'aggregating' (folding) and what have you arrived in C# 3.0, with surprisingly nippy implementations, and a whole lot more came with 3.5/4.0 (I forget).
Cool stuff. Cooler with closures. With events we didn't have to worry about writing inner Listener classes willy-nilly (insert comedy projectile vomiting here), but you still had to define the type of the function you were writing explicitly. Not so with closures. They're cool. I sometimes use them like private inner functions, but I'm pretty sure that's my fault.
Something else most definitely worth mentioning that came around here that is undoubtedly a good and cool thing was type inference, or the super magic 'var' keyword. (Like the C++ auto keyword, THAT CAME AFTERWARDS (possibly). We're inventing! Yay!)
Out with the List<string> stringList = new List<string>()! In with the var sl = new List<string>()! Glorious.
While I'm going long here, I need to also mention automatic property implementations.
So we don't have getWhatever and setWhatever functions in C# - (though we do agree with our Java brethren that generally exposed fields Are A Bad Thing) - we have: public whatever { get { return _whatever; } private set { _whatever = value; } }.
But in 3.0, you can do e.g. public whatever { get; private set; } and the boilerplate is written for you. If you ever want to swap them out with some code that actually does something at a point after that, you can then go ahead and do that. Awesome stuff. Java coders love this when they learn about it.
I think Mono also came of age here and stopped sucking. A new generational garbage collector (sgen), far greater API support, and the demands of game developers embedding the runtime left, right and center undoubtedly helped. (Mono runs on Windows and consoles, too, you know!)
Their System.Drawing implementation is still a bit crappy, though. I've had to do some weird things to alpha channels and what have you.
.NET 4.0 gave unto us the second revision of the F# language, and the Async and Await keywords. These are extremely cool, and I have had a great deal of fun writing seemingly-synchronous code (e.g. a character generation module for a MUD) that turns out to be asynchronous with the right words in the right places. Marvelous, great job, etc.
F# didn't begin existing at this point, but it basically did for me. F# is open source as heck, so it has a bit of the cool factor, and it really is just another take on OCaml. Unfortunately F# 'objects' (things, I mean) aren't real .NET objects, but you can explicitly write real .NET objects within the language with only slightly more verbose syntax and going the other way is really easy.
F# is a lot of fun and a real functional language. If you haven't tried it, you really should. Async and await are amazing here too.
Also gave us the .NET Framework 4 vs. .NET Framework 4 Client Profile as distinct framework targets. Clearly the latter is a cut-down version of the former, except that there's almost no difference in the available functionality, and the downloads for both runtimes are almost the same. Hm. Mono ignores the distinction completely, and I think we should too.
Whatever. Clearly C# is a labor of love, and we are going to get all the features we can ever pretend to understand over the coming years. I look forward to the eventual relaxing of having to put startup code in void main instead of just loose in the file (maybe), the IDE having more understanding of code (even if .NET intellisense in VS blows the C++ version out of the water and, indeed, at least three levels of atmosphere), and (optimistically) a more coherent F#.
.NET 4.0/4.5 has a bunch of features I can't even remember right now, and I look forward to learning more about the language, even if Littlecurrency politics seem to currently neglect .NET for C++ and Javascript in Metro apps a bit. (WinRT is a really cool framework though.) C'mon, only C++ apps allowed to use Direct3D? What year is this?
I'm still going to keep using C#, and I'm not the only one. (I use other languages too. Doesn't everyone? Go rules, Ruby has some very exciting ideas, and I wrote a node.js app the other day that to my surprise wasn't actually a piece of shit. So that's nice.)
I think that 'a slightly better Java than Java' is an idea that will prove promising for years to come.
A little about Mono:
Mono is a great platform, for what it's worth. With their new garbage collector (that can be tuned!), almost complete* support for the latest .NET 4.5 feature set, and their own additional features (tasklets and SIMD instructions, a REPL, and Compiler as a Service, anyone?) on top, I think it's pretty cool, and you might, too.
At the risk of preemptively lowering the tone - the current discussion here isn't that bad - the current best defense of Mono is an old Jo Shields rant at http://apebox.org/wordpress/rants/124/, though its tone is a little inflammatory at the start.
To summarize: Mono is independent from Microsoft, covered by the Open Invention Network re. patents, is a free implementation of an international standard by a convicted monopolist (a la the entire GNU system for UNIX) and is actually quite awesome in its own right. It's not just a compatibility framework. It isn't 100% compatible with the .NET Framework, and it doesn't need to be.
* Mono doesn't support WPF.
Speaking of spelling mistakes...
Also, most of tooling seems to be proprietary, Windows-only too.
So I'd stick to JVM.
I don't want to be a second class citizen ever. I don't want to be a second class citizen. I don't want to be a second class citizen ever.
"Really good" is so much different from "The Most Awesome Edition" you get with languages like Java or Ruby.
I can tinker with a "Really good" tech on home projects all right (except I prefer Clojure), but not in the workplace.
There is no alternative spelling. Sorry, you just used it so much!
When adopting Java, you can trust it will just work, but with Mono you have to bet it's going to work, you're not going to run into compatibility problems and tooling will be adequate.
Also every game made in Unity is using Mono, which is _a lot_ of games. The Playstation Vita and Wii U uses Mono as well for third party development.
C# is a great language and has evolved beautifully, but many people will still continue to avoid it because it's owned by MS and stuck on Windows (without the help of Mono). The next Google, Facebook or Twitter won't be written with C#.
Does Sun/Oracle sponsor all the different implementations of Java? Does the Python Foundation sponsor the various implementations of Python?
I'm guessing the work Microsoft are currently doing with Roslyn (compiler as a service) will benefit the various implementations also.
The C# spec is open and standardised, and if a particular group wants to pick it up and implement it, I find it a very weak argument to suggest that any company should throw money behind it.
Most of the additions have been to the CLR, where Microsoft has earned a reputation for changing behavior in the 'green' bits (meaning: supposedly unmodified) when updates are installed. Also, Windows 8 brings the new WinRT; it's impossible to know how long either runtime will be 'the one true way' (see Silverlight).
It's hard to get behind a software development tech created by a company that makes its money selling new versions of operating systems and development tools (the cool stuff always requiring the latest versions, of course!). Microsoft is always moving the cheese.
EDIT : couldn't find 2012 but heres 2011 http://punchcut.com/sites/punchcut.com/files/win8_1-8_eb-msf... business + server looks around 60% which I guess is where my figure comes from
Desmond: So should we be looking to move away from managed platforms and languages?
Petzold: No, no, no, not at all.
This article would have received a bit more respect if it had been a bit more subtle about the C# cheerleading.
RMS had been trying to spread that particular flavor of FUD for over a decade - albeit C#'s popularity has extended beyond Linux into mobile space.
I am no friend of Microsoft, and not a huge fan of Mono - but I think together they provide a viable alternative to the cross-platform JVM.