C# 6 Preview: Parameterless struct Constructors

Update 20th October 2015: As Karl Fenech pointed out, this feature has been dropped in the final version of C# 6.0.

Up until C# 5, you couldn’t have a parameterless constructor in a struct. So if you try something like this:

    public struct Point
    {
        public int x;
        public int y;

        public Point()
        {
            this.x = 0;
            this.y = 0;
        }

        public Point(int x, int y)
        {
            this.x = x;
            this.y = y;
        }
    }

…then it’s not quite going to work:

struct-parameterless-constructors-vs2012

At the time of writing this article, using a parameterless constructor in a struct as above is now supported, but as an experimental feature. This means you need to edit your project’s .csproj file and add the Experimental language version as shown below:

  <PropertyGroup Condition=" '$(Configuration)|$(Platform)' == 'Debug|AnyCPU' ">
    <PlatformTarget>AnyCPU</PlatformTarget>
    <DebugSymbols>true</DebugSymbols>
    <DebugType>full</DebugType>
    <Optimize>false</Optimize>
    <OutputPath>bin\Debug\</OutputPath>
    <DefineConstants>DEBUG;TRACE</DefineConstants>
    <ErrorReport>prompt</ErrorReport>
    <WarningLevel>4</WarningLevel>
    <LangVersion>Experimental</LangVersion>
  </PropertyGroup>

Parameterless struct constructors still require you to initialise all members of the struct, just like any other struct constructor.

Additionally, parameterless struct constructors must be public:

struct-parameterless-constructors-vs14ctp4

The reason for this is explained in the C# Design Notes for Aug 27, 2014:

C#, VB and F# will all call an accessible parameterless constructor if they find one. If there is one, but it is not accessible, C# and VB will backfill default(T) instead. (F# will complain.)

It is problematic to have successful but different behavior of new S() depending on where you are in the code. To minimize this issue, we should make it so that explicit parameterless constructors have to be public. That way, if you want to replace the “default behavior” you do it everywhere.

C# 6 Preview: Exception filters

We already have the ability to branch our exception-handling code depending on the type of exception, by using multiple catch blocks:

            try
            {
                Console.WriteLine("Enter a number");
                string inputStr = Console.ReadLine();
                int input = Convert.ToInt32(inputStr);
                Console.WriteLine(5 / input);
            }
            catch(FormatException ex)
            {
                Console.WriteLine("Input must be a number");
            }
            catch(DivideByZeroException ex)
            {
                Console.WriteLine("Can't divide by zero");
            }
            catch(Exception ex)
            {
                Console.WriteLine("Doh");
            }

In C# 6, we will be able to filter exceptions by handling them only if a condition is true. For example, we may want special treatment for exceptions that have an InnerException:

        static void Main(string[] args)
        {
            try
            {
                throw new InvalidOperationException("Invalid operation",
                    new Exception("Justin Bieber is breathing"));
            }
            catch(Exception ex) if (ex.InnerException != null)
            {
                Console.WriteLine(ex.Message);
                Console.WriteLine("  -->{0}", ex.InnerException.Message);
            }
            catch(Exception ex)
            {
                Console.WriteLine(ex);
            }

            Console.ReadLine();
        }

An additional benefit of exception filters is described in the C# feature descriptions (PDF):

Exception filters are preferable to catching and rethrowing because they leave the stack unharmed. If the exception later causes the stack to be dumped, you can see where it originally came from, rather than just the last place it was rethrown.

The same PDF also describes a method to take advantage of the conditional expressions used in exception filters in order to perform some action that uses but does not actually handle the exception. The description and code below are taken from that PDF.

It is also a common and accepted form of “abuse” to use exception filters for side effects; e.g. logging. They can inspect an exception “flying by” without intercepting its course. In those cases, the filter will often be a call to a false-returning helper function which executes the side effects:

private static bool Log(Exception e) { /* log it */ ; return false; }
…
try { … } catch (Exception e) if (Log(e)) {}

C# 6 Preview: The Safe Navigation Operator (?.)

The safe navigation operator, first announced and described in this blog post, has been available since the first Visual Studio 14 CTP due to its popularity in Visual Studio User Voice..

The operator, written as ?. in code, has been referred to by many names: safe navigation, null-conditional, null propagation, and conditional access are the ones I’ve come across so far.

Just to demonstrate the concept in an extremely simple manner, consider the following code:

            List<string> list = null;
            int count = list.Count;

It’s pretty obvious that if you try to run the above code, you’re going to get a NullReferenceException, because you’re trying to access the Count property of an object that is null.

Now, the example above can seem dumb because it’s pretty obvious that the list is null, but sometimes objects can come from external sources or user input, and so you have to be careful how you handle them. In this case we can work defensively by adding a null check:

            List<string> list = null;
            if (list != null)
            {
                int count = list.Count;
            }

That’s okay. But it can get pretty messy when there is a chain of references involved. Consider a binary tree, for instance.

var something = binaryTree1.LeftChild.LeftChild.Name;

If we were to write defensive conditionals for this, you effectively need a null check for each access.

            List<string> list = null;
            int? count = list?.Count;

The use of the safe navigation operator involves putting a question mark before the dot operator, as shown above. If list happens to be null (which it is in this case), then the Count property is not accessed, and null is immediately returned. Since list?.Count may return either a number or null, the count variable must now be nullable.

The safe navigation operator pays off pretty nicely when you have more complex reference accesses:

var something = binaryTree1?.LeftChild?.LeftChild?.Name;

It also works pretty nicely with the null-coalescing operator.

            List<string> list = null;
            int count = list?.Count ?? 0;

We can even use it to guard access to array elements when the array is null:

            string[] names = null;
            int length = names?[1].Length ?? 0;

…or guard access to array elements which may themselves be null:

            string[] names = new string[] { "Steve", null, "John" };
            int length = names[1]?.Length ?? 0;

The safe navigation operator is quite convenient in properties which make use of object instances. Consider the following property:

        public string SessionId
        {
            get
            {
                if (this.Session != null)
                    return this.Session.Id;
                else
                    return null;
            }
        }

We can now simplify this to:

        public string SessionId
        {
            get
            {
                return this.Session?.Id;
            }
        }

It’s also convenient to use the safe navigation operator to raise an event only if some delegate is subscribed to it. Consider an INotifyPropertyChanged implementation:

        public event PropertyChangedEventHandler PropertyChanged;

        // Create the OnPropertyChanged method to raise the event 
        protected void OnPropertyChanged(string name)
        {
            PropertyChangedEventHandler handler = PropertyChanged;
            if (handler != null)
            {
                handler(this, new PropertyChangedEventArgs(name));
            }
        }

You can’t just stick in a ?. before the paranthesised argument list. So the recommended way to do this is by using the Invoke() method:

        public event PropertyChangedEventHandler PropertyChanged;

        // Create the OnPropertyChanged method to raise the event 
        protected void OnPropertyChanged(string name)
        {
            PropertyChangedEventHandler handler = PropertyChanged;

            handler?.Invoke(this, new PropertyChangedEventArgs(name));
        }

The safe navigation operator isn’t going to change our lives. But it’s one of those sweet syntactic sugar features that we’ll grow to love in time.

Coast of Bones

screen002

Coast of Bones is the 8th scenario of the human campaign in the Warcraft 2 expansion set, Beyond The Dark Portal. Your objectives are, as is often the case, to destroy all your enemies.

screen003

You start off with a small force on a barren piece of land, and you have a few ships that will help you move to another place where resources are hopefully present.

screen004

In the screenshot above, you can see the entire map, enabled thanks to the on screen cheat. Your enemies are the white orcs to the west, purple to the southwest, and yellow to the south. The white ones are not very well-defended, allowing you to take over their base, and there is also a spare gold mine on the same piece of land. The purple orcs are better defended, and the yellow orcs are tough and have a strong naval presence.

screen005

You have two dwarven demolition squads. You can put them on a transport and use them to take down a couple of cannon towers. The first place to land is shown above.

screen006

This will allow you to take down the yellow cannon tower on the island, which is in a strategic location and monitors sea traffic in the area.

screen007

Land the second demolition squad on the northwestern landmass.

screen008

Then, use it to destroy the white cannon tower that guards the entrance to the white base.

screen009

With the cannon tower out of the way, you can now ferry your troops to the white base, and engage the enemy.

screen010

After killing the defenders, take down any towers, and then finish off the white orcs by destroying their buildings.

screen011

Establish your own base, building a town hall, farms and a lumber mill.

screen012

Establish your economy by having peasants mine gold and harvest lumber, and keep building farms to sustain your base’s growth.

screen013

Invest in your defences, by manning each entry point to the base and setting up towers to assist them. Continue to grow by building a barracks and upgrading your town hall to a keep.

screen015

Build stables to be able to train knights, and build a gnomish inventory to prepare for a naval presence (you will need gnomish flying machines to sniff out enemy giant turtles).

screen016

There’s an oil patch to the southwest. It’s pretty close to purple’s base, but is also quite convenient. It’s a good place to start setting up oil operations.

screen017

Build a shipyard there, but be prepared to face some heavy resistance from purple and yellow orc ships. One or two well-placed ballistas should keep them at bay (pun not intended).

screen018

At this point you will also see more activity from the yellow orcs, who will frequently make landings at the east side of the base. Make sure you are always well-defended.

screen019

My first shipyard couldn’t withstand the strength of the enemy naval forces, so I had to build another. I also reinforced my defences with a second ballista, seen on the right in the above screenshot. The ballista is out of range of enemy ships, and they conveniently get stuck trying to attack it.

screen020

When the main gold mine runs out of gold, you can build a town hall and start using the second one.

screen021

Continue investing in your naval operations by building a second shipyard, a refinery, and a foundry. Make sure you have a steady supply of oil.

screen022

Once you have a few battleships, you can begin to attack coastal buildings. Weakening purple’s naval presence is a good start.

screen023

When the oil platform runs out of oil, there’s another oil patch to the southeast you can use.

screen024

Be careful though – it’s an easy target for purple and yellow orc ships, and also for yellow orc troops.

screen025

Continue attacking purple’s coastal area…

screen026

…but keep your own operations safe. The central oil platform is an easy target for ships, land troops, and death knights’ whirlwinds.

screen027

As the tide of the battle permits, destroy the last remnants of purple’s naval presence.

screen028

Continue the attack towards the yellow orcs’ island coast, but be careful – it’s easy for your ships to get trapped in the narrow canals, in which they are easy prey for catapults and whirlwinds.

screen029

Make a landing into purple orc territory.

screen030

They may be grounded, but they can still put up a fight.

screen031

Bring in additional troops if you need.

screen032

Put an end to any death knights, because they weaken your forces pretty quickly.

screen033

Then work on paralysing their production by destroying strategic buildings, such as the fortress.

screen034

In this scenario, orcs usually don’t use air units; but they may do that when they get desperate. Having a couple of archers handy is a good idea.

screen035

With most of the purple forces neutralised, send fresh ships into the heart of the yellow orcs’ gulf.

screen036

Destroy the naval defences there.

screen037

Once the ships are destroyed, focus on destroying coastal buildings to completely annihilate yellow’s naval presence.

screen038

You will also find that many other buildings and forces are within reach of your battleships from the gulf, so you can easily bring the yellow orcs to their knees.

screen039

Complete the last preparations for an invasion by destroying orc units and buildings along the rest of the coast.

screen040

With that done, you can now make a landing on the yellow orcs’ island.

screen041

Bring in additional forces as needed.

screen042

Destroy strategic buildings first…

screen043

…and then finish off the remaining farms.

screen044

Enjoy your victory.

screen045

C# 6 Preview: Autoproperty enhancements

C# 6 extends the power of autoproperties.

Up to now, if you wanted to have a read-only property, then you had two options. You could either give up on autoproperties altogether, and use a readonly backing field exposed by a property:

        private readonly string name = "Snoop";

        public string Name
        {
            get
            {
                return name;
            }
        }

Or you could use an autoproperty with a private setter, which isn’t really read-only (it could be modified somewhere in the same class), and also means you’ll have to set the value in a constructor:

        public string Name { get; private set; }

        public Dogg()
        {
            this.Name = "Snoop";
        }

Fear not! C# 6 supports autoproperty initializers…

        public string Location { get; set; } = "NYC";

…which, quite conveniently, also work for getter-only properties:

        public string Name { get; } = "Snoop";

So it is now possible to declare an autoproperty without a setter, and its backing field will be implicitly declared as readonly.

These autoproperty initializers would have worked pretty nicely with a feature called primary constructors. But since this feature has been dropped, the usefulness of getter-only autoproperty initializers is restricted to readonly properties.

It is also worth mentioning this note about mutability from the C# feature descriptions [PDF]:

“This is about expressing types more concisely, but note that it also removes an important difference in the language between mutable and immutable types: auto-properties were a shorthand available only if you were willing to make your class mutable, and so the temptation to default to that was great. Now, with getter-only auto-properties, the playing field has been leveled between mutable and immutable.”

Eventually you’ll also be able to initialize getter-only autoproperties from within constructors, but at the time of writing this article, it is not yet supported in the latest CTP (Visual Studio 14 CTP4).

C# 6 Preview: await in catch and finally blocks

Up until now, you couldn’t put an await in a catch or finally block. That changes in C# 6.

I’ve got this Logger class, whose implementation is pretty trivial. This is how I’m using it in a simple WPF application’s button event handler:

        private async void Button_Click(object sender, RoutedEventArgs e)
        {
            using (var logger = new Logger())
            {
                try
                {
                    await logger.LogAsync("Executing operation...");
                }
                catch (Exception ex)
                {
                    await logger.LogAsync(ex.ToString());
                }
                finally
                {
                    await logger.FlushAsync();
                }
            }
        }

As you can see, in C# 6 you can now await asynchronous operations from within your catch and finally blocks, which is very useful in cases such as the above.

C# 6 Preview: Declaration Expressions

Note: although this feature appears to be implemented as an experimental language feature in Visual Studio 14 CTP4, the Roslyn language features page seems to indicate that it won’t be supported. In fact this post seems to confirm that declaration expressions won’t make it into the next version of .NET. Keep this in mind as you continue reading.

Update 14th April 2017: Although this feature did not make it into C# 6, it is now available as from C# 7 (released in 2017).

One of the new features in C# 6 is that we’ll be allowed to declare a variable from within an expression. The most common example of this is in the out parameter of a TryParse() call. Consider this code:

            Console.WriteLine("How old are you?");
            string input = Console.ReadLine();

            int x = 0;
            if (int.TryParse(input, out x))
                Console.WriteLine("You don't look {0}!", x);
            else
                Console.WriteLine("That doesn't sound quite right.");

            Console.ReadLine();

That x declaration is a little annoying – it wastes a line just to declare a variable so that we can get an out parameter into it. In C# 6, we can finally get rid of it:

            Console.WriteLine("How old are you?");
            string input = Console.ReadLine();

            if (int.TryParse(input, out int x))
                Console.WriteLine("You don't look {0}!", x);
            else
                Console.WriteLine("That doesn't sound quite right.");

            Console.ReadLine();

To actually get this working in Visual Studio 14 CTP4 (which is the latest at the time of writing this article), you’ll need to set the LangVersion to Experimental by editing the .csproj file in a text editor and adding the line highlighted below:

  <PropertyGroup Condition=" '$(Configuration)|$(Platform)' == 'Debug|AnyCPU' ">
    <PlatformTarget>AnyCPU</PlatformTarget>
    <DebugSymbols>true</DebugSymbols>
    <DebugType>full</DebugType>
    <Optimize>false</Optimize>
    <OutputPath>bin\Debug\</OutputPath>
    <DefineConstants>DEBUG;TRACE</DefineConstants>
    <ErrorReport>prompt</ErrorReport>
    <WarningLevel>4</WarningLevel>
    <LangVersion>Experimental</LangVersion>
  </PropertyGroup>

Now, I particularly love this feature because I use concurrent collections a lot. And those are full of out parameters. Here’s an example using a ConcurrentDictionary:

            var dict = new ConcurrentDictionary<string, Guid>();
            bool added = dict.TryAdd("user1", Guid.NewGuid());

            bool retrieved = dict.TryGetValue("user1", out Guid user1Guid);
            Console.WriteLine(user1Guid);

            Console.ReadLine();

Now, declaration expressions are useful for more than just out parameters. In fact, if you read Scott Allen’s article on the subject, you’ll see a couple of examples of declarations within conditional and loop expressions. Below is an example of how declaration expressions may be used in a loop:

            foreach (var num in var numbers = Enumerable.Range(1, 10))
                Console.WriteLine("Processing {0} of {1}...", num, numbers.Count());

…and here’s the output of that, just so you get the idea:

Processing 1 of 10...
Processing 2 of 10...
Processing 3 of 10...
Processing 4 of 10...
Processing 5 of 10...
Processing 6 of 10...
Processing 7 of 10...
Processing 8 of 10...
Processing 9 of 10...
Processing 10 of 10...

Finally, here is an example from the Roslyn C# feature descriptions (CTP3) [PDF] which shows how declaration expressions can facilitate the use of out parameters in queries, where declarations were previously not possible:

from s in strings
select int.TryParse(s, out int i) ? i : -1;

C# 6 Preview: using static

Update 31st January 2015: The syntax for using static has changed as from VS2015 CTP5. Please see C# 6 Preview: Changes in VS2015 CTP 5 for the latest syntax and examples. This article remains available due to historical significance.

Take a look at this little VB .NET program:

Module Module1

    Sub Main()

        Console.WriteLine("Hello world!")
        Console.ReadLine()

    End Sub

End Module

In VB .NET, you can take the Console static class name out of there and put it in the imports:

Imports System.Console

Module Module1

    Sub Main()

        WriteLine("Hello world!")
        ReadLine()

    End Sub

End Module

In fact, that feature works for any static class. And now, this feature is being added to C# starting from version 6. Let’s see the same example in C#. This is as you would write it in C# 5:

using System;

namespace UsingStaticHelloWorld
{
    class Program
    {
        static void Main(string[] args)
        {
            Console.WriteLine("Hello world!");
            Console.ReadLine();
        }
    }
}

And this is how it looks like with the new static using feature in C# 6:

using System.Console;

namespace UsingStaticHelloWorld
{
    class Program
    {
        static void Main(string[] args)
        {
            WriteLine("Hello world!");
            ReadLine();
        }
    }
}

Although most examples I’ve seen use this kind of scenario, I don’t think it’s a very good example of where this feature is useful. In fact, by extracting the static class name from the code that uses it, we’re reducing readability, and we’re also introducing the risk of ambiguity with other static classes that define the same method names. In this case we can’t add a static using for System.Diagnostics.Trace (which also has a WriteLine() method), but it’s quite easy to write a static Logger class that also defines a WriteLine() method, and that can be problematic.

However, the static using feature can be very useful in code that uses static methods heavily. A good example is mathematical calculations. Consider this code:

using System;

namespace UsingStaticMath
{
    class Program
    {
        static double CalculateSomething(double angle)
        {
            var numerator = Math.Sin(angle) * Math.Sin(angle) + Math.Cos(angle);
            var denominator = 1 + Math.Sqrt(Math.Sin(angle) + Math.Pow(Math.Cos(angle), 3));

            return Math.Pow(numerator / denominator, 2);
        }

        static void Main(string[] args)
        {
            var result = CalculateSomething(Math.PI / 2);

            Console.WriteLine(result);
            Console.ReadLine();
        }
    }
}

We can make it a whole lot more concise by introducing the using static:

using System;
using System.Math;

namespace UsingStaticMath
{
    class Program
    {
        static double CalculateSomething(double angle)
        {
            var numerator = Sin(angle) * Sin(angle) + Cos(angle);
            var denominator = 1 + Sqrt(Sin(angle) + Pow(Cos(angle), 3));

            return Pow(numerator / denominator, 2);
        }

        static void Main(string[] args)
        {
            var result = CalculateSomething(PI / 2);

            Console.WriteLine(result);
            Console.ReadLine();
        }
    }
}

In this case, when you read the code, it’s pretty obvious that Sin(), Cos(), Pow() and Sqrt() are mathematical functions, and it’s also very unlikely that there will be an ambiguity.

So in summary, the using static feature can be useful in some scenarios and disruptive in others. Use it wisely.

C# 6 Preview: nameof expressions

One of the new features in C# 6 is the new nameof operator. This will give you the name of a variable or type name, for instance nameof(id) gives you “id”. It’s a very simple feature. Let’s see a couple of examples where it is useful.

The simplest example is when checking whether arguments are null. I’ve got this simple Person class:

    public class Person
    {
        public string FirstName { get; set; }
        public string LastName { get; set; }

        public Person(string firstName, string lastName)
        {
            if (firstName == null)
                throw new ArgumentNullException("firstName");
            if (lastName == null)
                throw new ArgumentNullException("lastName");

            this.FirstName = firstName;
            this.LastName = lastName;
        }
    }

Notice how I’m passing magic strings into the ArgumentNullExceptions. If I rename the arguments, then I must remember to update the magic strings as well.

Let us now take advantage of the new nameof() operator:

    public class Person
    {
        public string FirstName { get; set; }
        public string LastName { get; set; }

        public Person(string firstName, string lastName)
        {
            if (firstName == null)
                throw new ArgumentNullException(nameof(firstName));
            if (lastName == null)
                throw new ArgumentNullException(nameof(lastName));

            this.FirstName = firstName;
            this.LastName = lastName;
        }
    }

It’s evident that with this change, if you rename the arguments, you will have to rename them throughout, or you will get a compiler error. No more magic strings.

Another scenario where this is really useful is when you are dealing with NotifyPropertyChanged in a WPF application. After implementing INotifyPropertyChanged in my ViewModel, I added the following property with a backing field to facilitate data binding:

        private string currentTime;

        public string CurrentTime
        {
            get
            {
                return this.currentTime;
            }
            set
            {
                this.currentTime = value;
                this.OnPropertyChanged("CurrentTime");
            }
        }

Here we have the same problem as before: we are using magic strings, and that can be problematic especially if we refactor our property and forget to update the property name in the string. Over the years, MVVM library developers have come up with various ways to mitigate this unpleasant scenario, including passing backing fields as ref parameters when raising the PropertyChanged event, using CallerMemberName, or using expression trees to pass in the property itself. nameof() again makes things a lot easier, as shown below. Even better, it is evaluated at compile-time, so there is no performance penalty associated with its use.

        private string currentTime;

        public string CurrentTime
        {
            get
            {
                return this.currentTime;
            }
            set
            {
                this.currentTime = value;
                this.OnPropertyChanged(nameof(CurrentTime));
            }
        }

"You don't learn to walk by following rules. You learn by doing, and by falling over." — Richard Branson