Nullable reference types

Article
01/17/2024

In a nullable-oblivious context, all reference types were nullable. Nullable reference types refers to a group of features enabled in a nullable aware context that minimize the likelihood that your code causes the runtime to throw System.NullReferenceException. Nullable reference types includes three features that help you avoid these exceptions, including the ability to explicitly mark a reference type as nullable:

Improved static flow analysis that determines if a variable might be null before dereferencing it.
Attributes that annotate APIs so that the flow analysis determines null-state.
Variable annotations that developers use to explicitly declare the intended null-state for a variable.

The compiler tracks the null-state of every expression in your code at compile time. The null-state has one of three values:

not-null: The expression is known to be not-null.
maybe-null: The expression might be null.
oblivious: The compiler can't determine the null-state of the expression.

Variable annotations determine the nullability of a reference type variable:

non-nullable: If you assign a null value or a maybe-null expression to the variable, the compiler issues a warning. Variables that are non-nullable have a default null-state of not-null.
nullable: You can assign a null value or a maybe-null expression to the variable. When the variable's null-state is maybe-null, the compiler issues a warning if you dereference the variable. The default null-state for the variable is maybe-null.
oblivious: You can assign a null value or a maybe-null expression to the variable. The compiler doesn't issue warnings when you dereference the variable, or when you assign a maybe-null expression to the variable.

The oblivious null-state and oblivious nullability match the behavior before nullable reference types were introduced. Those values are useful during migration, or when your app uses a library that hasn't enabled nullable reference types.

Null-state analysis and variable annotations are disabled by default for existing projects—meaning that all reference types continue to be nullable. Starting in .NET 6, they're enabled by default for new projects. For information about enabling these features by declaring a nullable annotation context, see Nullable contexts.

The rest of this article describes how those three feature areas work to produce warnings when your code might be dereferencing a null value. Dereferencing a variable means to access one of its members using the . (dot) operator, as shown in the following example:

string message = "Hello, World!";
int length = message.Length; // dereferencing "message"

When you dereference a variable whose value is null, the runtime throws a System.NullReferenceException.

Similarly warnings can be produced when [] notation is used to access a member of an object when the object is null:

using System;

public class Collection<T>
{
    private T[] array = new T[100];
    public T this[int index]
    {
        get => array[index];
        set => array[index] = value;
    }
}

public static void Main()
{
    Collection<int> c = default;
    c[10] = 1;    // CS8602: Possible derefence of null
}

You'll learn about:

The compiler's null-state analysis: how the compiler determines if an expression is not-null, or maybe-null.
Attributes that are applied to APIs that provide more context for the compiler's null-state analysis.
Nullable variable annotations that provide information about your intent for variables. Annotations are useful for fields to set the default null-state at the beginning of member methods.
The rules governing generic type arguments. New constraints were added because type parameters can be reference types or value types. The ? suffix is implemented differently for nullable value types and nullable reference types.
Nullable contexts help you migrate large projects. You can enable nullable contexts or warnings in parts of your app as you migrate. After you address more warnings, you can enable nullable reference types for the entire project.

Finally, you learn known pitfalls for null-state analysis in struct types and arrays.

You can also explore these concepts in our Learn module on Nullable safety in C#.

null-state analysis

When nullable reference types are enabled, Null-state analysis tracks the null-state of references. An expression is either not-null or maybe-null. The compiler determines that a variable is not-null in two ways:

The variable has been assigned a value that is known to be not-null.
The variable has been checked against null and hasn't been modified since that check.

When nullable reference types aren't enabled, all expressions have the null-state of oblivious. The rest of the section describes the behavior when nullable reference types are enabled.

Any variable that the compiler hasn't determined as not-null is considered maybe-null. The analysis provides warnings in situations where you might accidentally dereference a null value. The compiler produces warnings based on the null-state.

When a variable is not-null, that variable might be dereferenced safely.
When a variable is maybe-null, that variable must be checked to ensure that it isn't null before dereferencing it.

Consider the following example:

string message = null;

// warning: dereference null.
Console.WriteLine($"The length of the message is {message.Length}");

var originalMessage = message;
message = "Hello, World!";

// No warning. Analysis determined "message" is not-null.
Console.WriteLine($"The length of the message is {message.Length}");

// warning!
Console.WriteLine(originalMessage.Length);

In the preceding example, the compiler determines that message is maybe-null when the first message is printed. There's no warning for the second message. The final line of code produces a warning because originalMessage might be null. The following example shows a more practical use for traversing a tree of nodes to the root, processing each node during the traversal:

void FindRoot(Node node, Action<Node> processNode)
{
    for (var current = node; current != null; current = current.Parent)
    {
        processNode(current);
    }
}

The previous code doesn't generate any warnings for dereferencing the variable current. Static analysis determines that current is never dereferenced when it's maybe-null. The variable current is checked against null before current.Parent is accessed, and before passing current to the ProcessNode action. The previous examples show how the compiler determines null-state for local variables when initialized, assigned, or compared to null.

The null-state analysis doesn't trace into called methods. As a result, fields initialized in a common helper method called by all constructors generates a warning with the following template:

Non-nullable property 'name' must contain a non-null value when exiting constructor.

You can address these warnings in one of two ways: Constructor chaining, or nullable attributes on the helper method. The following code shows an example of each. The Person class uses a common constructor called by all other constructors. The Student class has a helper method annotated with the System.Diagnostics.CodeAnalysis.MemberNotNullAttribute attribute:


using System.Diagnostics.CodeAnalysis;

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

    public Person(string firstName, string lastName)
    {
        FirstName = firstName;
        LastName = lastName;
    }

    public Person() : this("John", "Doe") { }
}

public class Student : Person
{
    public string Major { get; set; }

    public Student(string firstName, string lastName, string major)
        : base(firstName, lastName)
    {
        SetMajor(major);
    }

    public Student(string firstName, string lastName) :
        base(firstName, lastName)
    {
        SetMajor();
    }

    public Student()
    {
        SetMajor();
    }

    [MemberNotNull(nameof(Major))]
    private void SetMajor(string? major = default)
    {
        Major = major ?? "Undeclared";
    }
}

Note

A number of improvements to definite assignment and null-state analysis were added in C# 10. When you upgrade to C# 10, you may find fewer nullable warnings that are false positives. You can learn more about the improvements in the features specification for definite assignment improvements.

Nullable state analysis and the warnings the compiler generates help you avoid program errors by dereferencing null. The article on resolving nullable warnings provides techniques for correcting the warnings most likely seen in your code.

Attributes on API signatures

The null-state analysis needs hints from developers to understand the semantics of APIs. Some APIs provide null checks, and should change the null-state of a variable from maybe-null to not-null. Other APIs return expressions that are not-null or maybe-null depending on the null-state of the input arguments. For example, consider the following code that displays a message in upper case:

void PrintMessageUpper(string? message)
{
    if (!IsNull(message))
    {
        Console.WriteLine($"{DateTime.Now}: {message.ToUpper()}");
    }
}

bool IsNull(string? s) => s == null;

Based on inspection, any developer would consider this code safe, and shouldn't generate warnings. However the compiler doesn't know that IsNull provides a null check and issues a warning for the message.ToUpper() statement, considering message to be a maybe-null variable. Use the NotNullWhen attribute to fix this warning:

bool IsNull([NotNullWhen(false)] string? s) => s == null;

This attribute informs the compiler, that, if IsNull returns false, the parameter s isn't null. The compiler changes the null-state of message to not-null inside the if (!IsNull(message)) {...} block. No warnings are issued.

Attributes provide detailed information about the null-state of arguments, return values, and members of the object instance used to invoke a member. The details on each attribute can be found in the language reference article on nullable reference attributes. As of .NET 5, all .NET runtime APIs are annotated. You improve the static analysis by annotating your APIs to provide semantic information about the null-state of arguments and return values.

Nullable variable annotations

The null-state analysis provides robust analysis for local variables. The compiler needs more information from you for member variables. The compiler needs more information to set the null-state of all fields at the opening bracket of a member. Any of the accessible constructors could be used to initialize the object. If a member field might ever be set to null, the compiler must assume its null-state is maybe-null at the start of each method.

You use annotations that can declare whether a variable is a nullable reference type or a non-nullable reference type. These annotations make important statements about the null-state for variables:

A reference isn't supposed to be null. The default state of a non-nullable reference variable is not-null. The compiler enforces rules that ensure it's safe to dereference these variables without first checking that it isn't null:
- The variable must be initialized to a non-null value.
- The variable can never be assigned the value null. The compiler issues a warning when code assigns a maybe-null expression to a variable that shouldn't be null.
A reference might be null. The default state of a nullable reference variable is maybe-null. The compiler enforces rules to ensure that you correctly check for a null reference:
- The variable might only be dereferenced when the compiler can guarantee that the value isn't null.
- These variables might be initialized with the default null value and might be assigned the value null in other code.
- The compiler doesn't issue warnings when code assigns a maybe-null expression to a variable that might be null.

Any non-nullable reference variable has a default null-state of not-null. Any nullable reference variable has the initial null-state of maybe-null.

A nullable reference type is noted using the same syntax as nullable value types: a ? is appended to the type of the variable. For example, the following variable declaration represents a nullable string variable, name:

string? name;

When nullable reference types are enabled, any variable where the ? isn't appended to the type name is a non-nullable reference type. That includes all reference type variables in existing code once you enable this feature. However, any implicitly typed local variables (declared using var) are nullable reference types. As the preceding sections showed, static analysis determines the null-state of local variables to determine if they're maybe-null before dereferencing it.

Sometimes you must override a warning when you know a variable isn't null, but the compiler determines its null-state is maybe-null. You use the null-forgiving operator ! following a variable name to force the null-state to be not-null. For example, if you know the name variable isn't null but the compiler issues a warning, you can write the following code to override the compiler's analysis:

name!.Length;

Nullable reference types and nullable value types provide a similar semantic concept: A variable can represent a value or object, or that variable might be null. However, nullable reference types and nullable value types are implemented differently: nullable value types are implemented using System.Nullable<T>, and nullable reference types are implemented by attributes read by the compiler. For example, string? and string are both represented by the same type: System.String. However, int? and int are represented by System.Nullable<System.Int32> and System.Int32, respectively.

Nullable reference types are a compile time feature. That means it's possible for callers to ignore warnings, intentionally use null as an argument to a method expecting a non nullable reference. Library authors should include run-time checks against null argument values. The ArgumentNullException.ThrowIfNull is the preferred option for checking a parameter against null at run time.

Important

Enabling nullable annotations can change how Entity Framework Core determines if a data member is required. You can learn more details in the article on Entity Framework Core Fundamentals: Working with Nullable Reference Types.

Generics

Generics require detailed rules to handle T? for any type parameter T. The rules are necessarily detailed because of history and the different implementation for a nullable value type and a nullable reference type. Nullable value types are implemented using the System.Nullable<T> struct. Nullable reference types are implemented as type annotations that provide semantic rules to the compiler.

If the type argument for T is a reference type, T? references the corresponding nullable reference type. For example, if T is a string, then T? is a string?.
If the type argument for T is a value type, T? references the same value type, T. For example, if T is an int, the T? is also an int.
If the type argument for T is a nullable reference type, T? references that same nullable reference type. For example, if T is a string?, then T? is also a string?.
If the type argument for T is a nullable value type, T? references that same nullable value type. For example, if T is a int?, then T? is also a int?.

For return values, T? is equivalent to [MaybeNull]T; for argument values, T? is equivalent to [AllowNull]T. For more information, see the article on Attributes for null-state analysis in the language reference.

You can specify different behavior using constraints:

The class constraint means that T must be a non-nullable reference type (for example string). The compiler produces a warning if you use a nullable reference type, such as string? for T.
The class? constraint means that T must be a reference type, either non-nullable (string) or a nullable reference type (for example string?). When the type parameter is a nullable reference type, such as string?, an expression of T? references that same nullable reference type, such as string?.
The notnull constraint means that T must be a non-nullable reference type, or a non-nullable value type. If you use a nullable reference type or a nullable value type for the type parameter, the compiler produces a warning. Furthermore, when T is a value type, the return value is that value type, not the corresponding nullable value type.

These constraints help provide more information to the compiler on how T is used. That helps when developers choose the type for T and provides better null-state analysis when an instance of the generic type is used.

Nullable contexts

For small projects, you can enable nullable reference types, fix warnings, and continue. However, for larger projects and multi-project solutions, that might generate a large number of warnings. You can use pragmas to enable nullable reference types file-by-file as you begin using nullable reference types. The new features that protect against throwing a System.NullReferenceException can be disruptive when turned on in an existing codebase:

All explicitly typed reference variables are interpreted as non-nullable reference types.
The meaning of the class constraint in generics changed to mean a non-nullable reference type.
New warnings are generated because of these new rules.

The nullable annotation context determines the compiler's behavior. There are four values for the nullable annotation context:

disable: The code is nullable-oblivious. Disable matches the behavior before nullable reference types were enabled, except the new syntax produces warnings instead of errors.
- Nullable warnings are disabled.
- All reference type variables are nullable reference types.
- Use of the ? suffix to declare a nullable reference type produces a warning.
- You can use the null forgiving operator, !, but it has no effect.
enable: The compiler enables all null reference analysis and all language features.
- All new nullable warnings are enabled.
- You can use the ? suffix to declare a nullable reference type.
- Reference type variables without the ? suffix are non-nullable reference types.
- The null forgiving operator suppresses warnings for a possible dereference of null.
warnings: The compiler performs all null analysis and emits warnings when code might dereference null.
- All new nullable warnings are enabled.
- Use of the ? suffix to declare a nullable reference type produces a warning.
- All reference type variables are allowed to be null. However, members have the null-state of not-null at the opening brace of all methods unless declared with the ? suffix.
- You can use the null forgiving operator, !.
annotations: The compiler doesn't emit warnings when code might dereference null, or when you assign a maybe-null expression to a non-nullable variable.
- All new nullable warnings are disabled.
- You can use the ? suffix to declare a nullable reference type.
- Reference type variables without the ? suffix are non-nullable reference types.
- You can use the null forgiving operator, !, but it has no effect.

The nullable annotation context and nullable warning context can be set for a project using the <Nullable> element in your .csproj file. This element configures how the compiler interprets the nullability of types and what warnings are emitted. The following table shows the allowable values and summarizes the contexts they specify.

Context	Dereference warnings	Assignment warnings	Reference types	`?` suffix	`!` operator
`disable`	Disabled	Disabled	All are nullable	Produces a warning	Has no effect
`enable`	Enabled	Enabled	Non-nullable unless declared with `?`	Declares nullable type	Suppresses warnings for possible `null` assignment
`warnings`	Enabled	Not applicable	All are nullable, but members are considered not-null at opening brace of methods	Produces a warning	Suppresses warnings for possible `null` assignment
`annotations`	Disabled	Disabled	Non-nullable unless declared with `?`	Declares nullable type	Has no effect

Reference type variables in code compiled in a disabled context are nullable-oblivious. You can assign a null literal or a maybe-null variable to a variable that is nullable-oblivious. However, the default state of a nullable-oblivious variable is not-null.

You can choose which setting is best for your project:

Choose disable for legacy projects that you don't want to update based on diagnostics or new features.
Choose warnings to determine where your code might throw System.NullReferenceExceptions. You can address those warnings before modifying code to enable non-nullable reference types.
Choose annotations to express your design intent before enabling warnings.
Choose enable for new projects and active projects where you want to protect against null reference exceptions.

Example:

<Nullable>enable</Nullable>

You can also use directives to set these same contexts anywhere in your source code. These directives are most useful when you're migrating a large codebase.

#nullable enable: Sets the nullable annotation context and nullable warning context to enable.
#nullable disable: Sets the nullable annotation context and nullable warning context to disable.
#nullable restore: Restores the nullable annotation context and nullable warning context to the project settings.
#nullable disable warnings: Set the nullable warning context to disable.
#nullable enable warnings: Set the nullable warning context to enable.
#nullable restore warnings: Restores the nullable warning context to the project settings.
#nullable disable annotations: Set the nullable annotation context to disable.
#nullable enable annotations: Set the nullable annotation context to enable.
#nullable restore annotations: Restores the nullable annotation context to the project settings.

For any line of code, you can set any of the following combinations:

Warning context	Annotation context	Use
project default	project default	Default
enable	disable	Fix analysis warnings
enable	project default	Fix analysis warnings
project default	enable	Add type annotations
enable	enable	Code already migrated
disable	enable	Annotate code before fixing warnings
disable	disable	Adding legacy code to migrated project
project default	disable	Rarely
disable	project default	Rarely

Those nine combinations provide you with fine-grained control over the diagnostics the compiler emits for your code. You can enable more features in any area you're updating, without seeing more warnings you aren't ready to address yet.

Important

The global nullable context does not apply for generated code files. Under either strategy, the nullable context is disabled for any source file marked as generated. This means any APIs in generated files are not annotated. There are four ways a file is marked as generated:

In the .editorconfig, specify generated_code = true in a section that applies to that file.
Put <auto-generated> or <auto-generated/> in a comment at the top of the file. It can be on any line in that comment, but the comment block must be the first element in the file.
Start the file name with TemporaryGeneratedFile_
End the file name with .designer.cs, .generated.cs, .g.cs, or .g.i.cs.

Generators can opt-in using the #nullable preprocessor directive.

By default, nullable annotation and warning contexts are disabled. That means that your existing code compiles without changes and without generating any new warnings. Beginning with .NET 6, new projects include the <Nullable>enable</Nullable> element in all project templates.

These options provide two distinct strategies to update an existing codebase to use nullable reference types.

Known pitfalls

Arrays and structs that contain reference types are known pitfalls in nullable references and the static analysis that determines null safety. In both situations, a non-nullable reference might be initialized to null, without generating warnings.

Structs

A struct that contains non-nullable reference types allows assigning default for it without any warnings. Consider the following example:

using System;

#nullable enable

public struct Student
{
    public string FirstName;
    public string? MiddleName;
    public string LastName;
}

public static class Program
{
    public static void PrintStudent(Student student)
    {
        Console.WriteLine($"First name: {student.FirstName.ToUpper()}");
        Console.WriteLine($"Middle name: {student.MiddleName?.ToUpper()}");
        Console.WriteLine($"Last name: {student.LastName.ToUpper()}");
    }

    public static void Main() => PrintStudent(default);
}

In the preceding example, there's no warning in PrintStudent(default) while the non-nullable reference types FirstName and LastName are null.

Another more common case is when you deal with generic structs. Consider the following example:

#nullable enable

public struct S<T>
{
    public T Prop { get; set; }
}

public static class Program
{
    public static void Main()
    {
        string s = default(S<string>).Prop;
    }
}

In the preceding example, the property Prop is null at run time. It's assigned to non-nullable string without any warnings.

Arrays

Arrays are also a known pitfall in nullable reference types. Consider the following example that doesn't produce any warnings:

using System;

#nullable enable

public static class Program
{
    public static void Main()
    {
        string[] values = new string[10];
        string s = values[0];
        Console.WriteLine(s.ToUpper());
    }
}

In the preceding example, the declaration of the array shows it holds non-nullable strings, while its elements are all initialized to null. Then, the variable s is assigned a null value (the first element of the array). Finally, the variable s is dereferenced causing a runtime exception.

Share via