Tips for code-golfing in C#

Question

What general tips do you have for golfing in C#? I'm looking for ideas that can be applied to code golf problems in general that are at least somewhat specific to C# (e.g. "remove comments" is not an answer). Please post one tip per answer.

-- borrowed from marcog's idea ;)

• undergroundmonorail: Forgive me for the picture of a calendar, it was all I could find on short notice.

Answer 1

Instead of using .ToString() use +"" for numerics and other types that can be natively cast to a string safely.

.ToString() <-- 11 chars
+"" <-- 3 chars

• Cyoce: This also works in JS.
• TheLethalCoder: This is usually actually 5 chars if you need to use the string afterwards for including the braces... (1+"").DoSomethingWith1String();
• jcolebrand: If you need the string you're usually storing it. Just about any other usage can natively infer the ToString()...
• VisualMelon: Note that this actually calls the static String.Concat(object) with the argument, rather than the virtual calling object.ToString(). Concat explicitly converts null to the empty string (see the reference source). There is no 'native casting' going on, you can convert anything like this, it's just that the result might not be very useful in some cases! (but the null behaviour may well be).
• Andriy Tolstoy: Alternative - string interpolation: $"{n}"

Answer 2

I once deliberately placed my program in namespace System so I can shorten access to a specific class. Compare

using System;using M=System.Math;

to

namespace System{using M=Math;

• Nick Larsen: Just remember that it is better to fully qualify classes/functions when a single use solves the problem. This is only useful if you have to call something more than once, and even then only for items in the System namespace.
• ldam: You can also just do using System;class P....
• Joey: @Logan: This wasn't about just using System; but also about having an alias for a class in the same namespace, which is shorter the way I've shown here.
• milk: It's even shorter to do using static System.Math; in C#6 (add you can use any of those functions as if they were truly global--not in a class). The original suggestion may still be shorter than using static if you need to access multiple classes.
• Joey: @milk: the additional static keyword often is longer than any savings from leaving out M. on the method calls, but yes, it's an option, but it comes at a hefty upfront cost that needs lots of calls to amortise.

Answer 3

If using LINQ you can pass a method directly to Select instead of making a lambda.

So, instead of

foo.Select(x=>int.Parse(x))

you can use

foo.Select(int.Parse)

directly.

(Discovered recently when improving on one of Timwi's C# answers.)

• ThreeFx: FWITW this is called η-reduction
• Jonathan Wilson: It's also known as "point free" style
• Joey: To the more pragmatic of us, it's simply shorter :-þ
• David: Interesting! Would love to learn more. I asked a question over at softwareengineering @ThreeFx

Answer 4

Use var for declaring and initializing (single) variables to save characters on the type:

string x="abc";

becomes

var x="abc";

Isn't particulaly necessary for int, of course.

• Ian H.: Remember that var cannot have multiple declarators, for example var x="x",y="y"; is not possible.

Answer 5

Remember that the smallest compilable program in C# is 29 characters:

class P
{
 static void Main()
 { 
 }
}

So start by removing that from your length and judge your answer on how much over that it takes. C# cannot compete with other languages when it comes to printing or reading input, which is the heart of most [code-golf] problems, so don't worry about that. As a C# golfer, you're really competing against the language.

A few other things to keep in mind:

• Reduce all loops and if statements to a single line if possible in order to remove the brackets.
• If given the option between stdin and command line, always use command line!

• jcolebrand: This does usually involve ternary as well ;)
• user8397947: As a C# golfer, you're really competing against the language Unbelievably related
• Mika: Actually, that's not true. It compiles using static int Main() as well, which would be 28 characters.
• user202729: @Metoniem That requires a return something.
• bazzilic: Be advised that starting with C# 9 you can use top-level statements thus making the smallest compilable program exactly 0 bytes (docs.microsoft.com/en-us/dotnet/csharp/fundamentals/…)

Answer 6

Instead of

bool a = true;
bool b = false;

do

var a=0<1;
var b=1<0;

If you need multiple variables, use this (suggested by @VisualMelon)

bool a=0<1,b=!a;

• VisualMelon: Note that if you need multiple variables of the same type it is usually cheaper to declare the type a comma separate the declarations bool a=0<1,b=!a;

Answer 7

When reading each character of a command line argument, rather than looping up to the string's length:

static void Main(string[]a){
 for(int i=0;i<a[0].Length;)Console.Write(a[0][i++]);
}

You can save a character by using a try/catch block to find the end:

static void Main(string[]a){
 try{for(int i=0;;)Console.Write(a[0][i++]);}catch{}
}

This applies to any array within an array such as:

• string[]
• int[][]
• IList<IList<T>>

• Alex Reinking: That is truly horrifying... I love it!
• Gaspa79: holy crap this is genius, I actually just saved a character while looping an array
• GreatAndPowerfulOz: This is truly evil!

Answer 8

If you need to use Console.ReadLine() multiple times in your code (min 3 times), you could do:

Func<string>r=Console.ReadLine;

and then just use

r()

instead

• mellamokb: I think you need to remove () from the first line.
• Claudiu: Can't you do auto r=Console.ReadLine;?
• Cristian Lupascu: @claudiu no, unfortunally not ideone.com/jFsVPX
• GreatAndPowerfulOz: @Claudiu, auto is a C++ verb. var is for C#. The reason this can't be done is because Console.ReadLine is overloaded so the function signature needs to be specified in order to tell the compiler which overload is wanted.
• btnlq: You can do var r=Console.ReadLine; and var s=(int x)=>x*x; since C# 10

Answer 9

Favor the ternary operator over if..else blocks where appropriate.

For example:

if(i<1)
 j=1;
else
 j=0;

is more efficiently:

j=i<1?1:0;

• jcolebrand: Am I the only one that feels that the second case is inherently more readable for things like this in general? I do that routinely. In addition, if I need to avoid a null condition (like on a string) I do something like var x = input ?? ""; (I loves my coalesces)
• Nick Larsen: There are times when it is far from being the more readable option, particularly when i < 1 is a complex statement or when the name of j is long. IMO, it also fails to convey side effects very well. In the case where if (i < 1) is something like if (SendEmail(recipient)) which returns true/false depending on the success of the side effects, I prefer the if/then notation.
• Danko Durbić: No need for parentheses in the second case - j=i<1?1:0; is enough.
• Peter Taylor: The question asks for tips which are somewhat specific to C#. This is one included in the tips for all languages.
• Nellius: @PeterTaylor I answered this question over 3 years ago, well before the thread you linked was created

Answer 10

Effective use of using

You can replace float (which is an alias for System.Single) with z using z=System.Single;

Then replace z=System.Single; with z=Single; by placing the program in the namespace System. (As with Joey's answer)

This can be applied for other value types (use what they are an alias for), structs and classes

Answer 11

LINQ

Instead of using:

Enumerable.Range(0,y).Select(i=>f(i))

to get an Enumerable with the result of function f for every int in [0,y] you can use

new int[y].Select((_,i)=>f(i))

if you need string or anything that implements Enumerable in your program you can use them too

var s="I need this anyway";
s.Select((_,i)=>f(i))

• aloisdg: I use this trick in my answer for the Shamir's Secret Sharing challenge.
• Gaspa79: I don't think that the string part will execute if you don't iterate the ienumerable with optimization on. Just failed for me until I did .ToArray();. Other than that, amazing tip!
• raggy: Yes enumerables are lazy but that is true for all three examples not just the one with the string.

Answer 12

In C#, we are not allowed to do if(n%2) to check if n is a even number. If we do, we get a cannot implicity convert int to bool. A naive handling would be to do:

if(n%2==0)

A better way is to use:

if(n%2<1)

I used this to gain one byte here.

note that this only works for positive numbers, as -1%2==-1, it is considered even with this method.

Answer 13

Use lambdas to define a function in C# 6

In C# 6, you can use a lambda to define a function:

int s(int a,int b)=>a+b;

This is shorter than defining a function like this:

int s(int a,int b){return a+b;}

• jcolebrand: C#6 gives a whole new range of ability to code-golf
• TehPers: In C#7, this can be done inside another function to create local functions. I doubt this will help while golfing, but it's still just a neat trick to know.
• recursive: That's not formally a lambda. It's an expression bodied member.

Answer 14

Looping:

Variable declarations:

int max;
for(int i=1;i<max;i++){
}

become:

int max,i=1;
for(;i<max;i++){
}

And if you have a need to or work with the i variable only once, you could start at -1 (or 0 depending on the loop circumstance) and increment inline:

int max,i=1;
for(;i<max;i++){
 Console.WriteLine(i);
}

to

int max,i=1;
for(;i<max;){
 Console.WriteLine(++i);
}

And that reduces by one character, and slightly obfuscates the code as well. Only do that to the FIRST i reference, like thus: (granted one character optimizations aren't much, but they can help)

int max,i=1;
for(;i<max;i++){
 Console.WriteLine(i + " " + i);
}

to

int max,i=1;
for(;i<max;){
 Console.WriteLine(++i + " " + i);
}

---

when the loop does not have to increment i (reverse order loop):

for(int i=MAX;--i>0;){
 Console.WriteLine(i);
}

• Joey: I usually put the ++ in such cases directly into the loop header: for(;++i<max;) which is both easier to follow and harder to get wrong.
• ICR: @Joey In those cases I tend to switch to while(++i<max) which is the same length but easier to read.
• Joey: ICR: depends on whether you can put another (earlier) statement into the for header as well, which would then save a character again.
• recursive: You can move both declarations back into the for clause for a1 byte savings.
• Zuabros: If you won't need the variable again, you can use: while(max-->0) { foo;}

Answer 15

If you need to use a generic Dictionary<TKey, TValue> at least two times in your code, you could declare a dictionary class, like in this example:

class D:Dictionary<int,string>{}

and then just use

D d=new D{{1,"something"},{2,"something else"}};

instead of repeating Dictionary<int,string> for every instantiation.

I have used this technique in this answer

• Zukki: And also "D d" instead of "var d"
• Cristian Lupascu: @Zukki Obviously! What was I thinking? :)
• Andriy Tolstoy: Alternative: using D = System.Collections.Generic.Dictionary<int,string>;

Answer 16

For one-line lambda expressions, you can skip the brackets and semicolon. For one-parameter expressions, you can skip the parentheses.

Instead of

SomeCall((x)=>{DoSomething();});

Use

SomeCall(x=>DoSomething);

• R. Martinho Fernandes: I never write the parentheses for one-parameter lambdas, even on production code.
• Juliana Peña: I always use the brackets because I like to split the lambda into multiple lines for readability.
• GreatAndPowerfulOz: SomeCall(DoSomething) is even better

Answer 17

You can use float and double literals to save a few bytes.

var x=2.0;
var y=2d; // saves 1 byte

When you need some int arithmetic to return a float or double you can use the literals to force the conversion.

((float)a+b)/2; // this is no good
(a+b)/2.0; // better
(a+b)/2f; // best 

If you ever run into a situation where you have to to cast you can save a few bytes by using multiplication instead.

((double)x-y)/(x*y);
(x*1d-y)/(x*y); // saves 5 bytes

• recursive: Even shorter: (x-y)*1d/x/y;

Answer 18

Avoid single-statement foreach loops

If the loop's statement returns a non-int (including void!) "value", it can be replaced with LINQ:

foreach(var x in a)Console.WriteLine(F(x));
a.Any(x=>Console.WriteLine(F(x))is int);

If the value happens to be an int, you can use a condition that will always be true or always be false (for example, >0 or <n), a different type and/or All instead of Any.

• btnlq: is GC saves 1 byte

Answer 19

Remember where private or public are inherent, such as the following:

class Default{static void Main()

as compared to

public class Default { public static void Main()

• Joey: And always make the class one letter only :-)
• Joey: Oh, and another nice thing, implied here: Main does not need any arguments in contrast to Java, for example.
• R. Martinho Fernandes: @Joey: and neither does it need to be public.
• jcolebrand: @martinho ~ did you read my answer? ;) no public on main
• jcolebrand: @Joey ~ I was trying to keep it to one per post ;) ... figured someone else would post taht about main or classes only being one letter. Seeing as how nobody else has, I'll go ahead and add that one too.

Answer 20

There are circumstances when an output parameter can save characters. Here's a slightly contrived example, a 10 pin bowling score algorithm.

With a return statement:

........10........20........30........40........50........60........70........80........90.......100.......110.......120.......130.......140.......150..
public double c(int[]b){int n,v,i=0,X=10;double t=0;while(i<19){n=b[i]+b[i+1];v=b[i+2];t+=(n<X)?n:X+v;if(b[i]>9)t+=b[i+(i>16|v!=X?3:4)];i+=2;}return t;}

And with an output parameter:

........10........20........30........40........50........60........70........80........90.......100.......110.......120.......130.......140.......
public void d(int[]b,out double t){int n,v,i=0,X=10;t=0;while(i<19){n=b[i]+b[i+1];v=b[i+2];t+=(n<X)?n:X+v;if(b[i]>9)t+=b[i+(i>16|v!=X?3:4)];i+=2;}}

The output parameter here saves a total of 5 characters.

Answer 21

Use dynamic to group declarations

dynamic is a forgotten feature that literally performs dynamic typing in C#! It has limitations (doesn't support extension methods, is bad at inferring that you want to use it, ...), but can often save bytes by merging declarations of incompatible types. Compare the following:

var a=[something of type IEnumerable<int>];var b=[something of type string];int i=x+~y^z
dynamic a=[something of type IEnumerable<int>],b=[something of type string],i=x+~y^z

That's 4 bytes of savings!

• primo: One of the best tips on the page.
• nwellnhof: With C# 7, you can sometimes save another byte with tuples: var(a,i)=([...],[...]).

Answer 22

String Interpolation

A really simple space-saving improvement is interpolation. Instead of:

string.Format("The value is ({0})", (method >> 4) + 8)

just use $ to inline expressions:

$"The value is ({(method >> 4) + 8})"

This, together with the new expression bodies in C#6.0 should make any simple string-calculation challenge pretty golfable in C#.

• aloisdg: Also note that i+$" bottles of beer"; is shorter than $"{i} bottles of beer".
• Mika: @aloisdg In that first case you should leave $ out, though.
• aloisdg: @Metoniem Indeed! I let it because in my original case I had two {i} one in front and one in the middle ;)
• Mika: @aloisdg Ahh, I see. Yeah, shame comments can't be edited :(

Answer 23

Use the weird kind of the is operator

a is var b always defines the variable b equal to a and returns true. It is unclear how have the humans who design C# come up with this (inline variable declarations that return true for some reason seem to be beyond Javascript to me), but it works. It can sometimes be used to shorten code (this is an oversimplified example; savings can often be smaller or zero):

a=>{var b=f(a);return g(b,b);}
a=>g(f(a)is var b?b:b,b)

Answer 24

If you need to include multiple usings that all fall off of the same hierarchy it is often shorter to use the longest one as the namespace:

using System;
using System.Linq;
//Some code

vs:

namespace System.Linq
{
 //Some code
}

Answer 25

The Compute instance method of System.Data.DataTable, allows to evaluate a simple string expression, e.g. :

C# (Visual C# Compiler), 166 bytes

namespace System.Data
{
 class P
 {
 static void Main()
 {
 Console.Write(new DataTable().Compute("30*2+50*5/4",""));
 }
 }
}

Try it online!

Not very "golfy" per se, but sometimes might be useful.

Answer 26

Swapping two variables

Normally, to swap two variables, you have to declare a temporary variable to store the value. It would look like something along the lines of this:

var c=a;a=b;b=c;

That's 16 bytes! There are some other methods of swapping that are better.

//Using tuples
(a,b)=(b,a);
//Bitwise xoring 
a=a^b^(b=a);
//Addition and subtraction
a=a+b-(b=a);
//Multiplication and division
a=a*b/(b=a);

The last three only work for numeric values, and as ASCII-only pointed out, the last two might result in an ArithmeticOverflow exception. All of the above are 12 bytes, a 4 byte saving compared to the first example.

• ASCII-only: Only applies for numbers though, and even then anything other than tuples and xor risk running into integer limits of you're applying this to integers. Occasionally, other number types will run into limits too

Answer 27

Use C# lambda. Since PPCG allows lambda for input/output we should use them.

A classic C# methods looks like this:

bool Has(string s, char c)
{
 return s.Contains(c);
}

As a lambda, we will write

Func<string, char, bool> Has = (s, c) => s.Contains(c);

Anonymous lambda are allowed too:

(s, c) => s.Contains(c)

Remove all the noise and focus!

Update:

We can improve one step more with currying as @TheLethalCoder comment:

s => c => s.Contains(c);

Example of curring by @Felix Palmen: How compute WPA key?

It will be helpful when you have exactly 2 parameters, then a empty unused variable _ will be better. See meta post about this. I use this trick here. You will have to change a bit the function. Example: Try it online!

• TheLethalCoder: Not sure if it's anywhere else in the tips but for this example you can use currying too... s=>c=>...
• aloisdg: @TheLethalCoder Indeed we can! I will update the answer thank you!
• corvus_192: Can you use eta-reduction in this case? Something like this: s=>s.Contains.
• VisualMelon: Note that C# and Java answers of the 'untyped lambda' variety are falling out of favour, you might wish to join the discussion on this meta post. The suggested alternative is (string s,char c)=>s.Contains(c)
• aloisdg: Discussion about unamed functions

Answer 28

Use .NET 6 Top Level Statements, File-Scoped Namespaces, and target-typed new()

Upgrade to .net 6 and you can use single file programs with no boilerplate. You can also use file-scoped namespaces like so:

namespace whatever;

Also, when you instantiate an object, if it's type can be inferred by the compiler, you can use a target-typed new() constructor. e.g.,

class C {}
C c=new();

See (https://docs.microsoft.com/en-us/dotnet/csharp/whats-new/tutorials/top-level-statements)

See Also (https://docs.microsoft.com/en-us/dotnet/csharp/language-reference/proposals/csharp-10.0/file-scoped-namespaces)

• rydwolf: Welcome to Code Golf, and nice tip!

Answer 29

Make classnames only one letter. Enhancing on Tips for code-golfing in C# we go from

class Default{static void Main()

to

class D{static void Main()

which knocks out another 6 chars in this case.

• Nellius: ditto with variable names
• Peter Taylor: How is this "at least somewhat specific to C#"?

Answer 30

Group statements via tuples

Let's assume the simplest case: you have two statements that return a value and would like to group them into one that returns the second value (perhaps the first one has useful side effects):

(s1,o:s2).o

This can be used along with inline variable declarations via is:

(s1 is var x,o:f(x,g(x))).o

It also works with the void -> value transformation via is int (although it doesn't always help):

(Print(x)is int,o:f(x)).o

• recursive: (s1 is var x,o:f(x,g(x))).o can usually be shortened to (s1 is{}x,o:f(x,g(x))).o now.