2212
votes

I have been running StyleCop over some C# code, and it keeps reporting that my using directives should be inside the namespace.

Is there a technical reason for putting the using directives inside instead of outside the namespace?

12
Sometimes it makes difference where you put usings: stackoverflow.com/questions/292535/linq-to-sql-designer-buggius
Just for reference, there are implications beyond just the question of multiple classes per file, so if you're new to this question, please keep reading.Charlie
@user-12506 - this does not work very well in a medium to large development team where some level of code consistency is required. And as noted previously, if you don't understand the different layouts you may find edge cases that don't work as you expect.benPearce
Terminology: Those are not using statements; they are using directives. A using statement, on the other hand, is a language structure that occurs along with other statements inside a method body etc. As an example, using (var e = s.GetEnumerator()) { /* ... */ } is a statement that is loosely the same as var e = s.GetEnumerator(); try { /* ... */ } finally { if (e != null) { e.Dispose(); } }.Jeppe Stig Nielsen
If this was not mentioned already by anyone, actually Microsoft too recommends putting using statements inside the namespace declarations, in their internal coding guidlinesuser1451111

12 Answers

2243
votes

There is actually a (subtle) difference between the two. Imagine you have the following code in File1.cs:

// File1.cs
using System;
namespace Outer.Inner
{
    class Foo
    {
        static void Bar()
        {
            double d = Math.PI;
        }
    }
}

Now imagine that someone adds another file (File2.cs) to the project that looks like this:

// File2.cs
namespace Outer
{
    class Math
    {
    }
}

The compiler searches Outer before looking at those using directives outside the namespace, so it finds Outer.Math instead of System.Math. Unfortunately (or perhaps fortunately?), Outer.Math has no PI member, so File1 is now broken.

This changes if you put the using inside your namespace declaration, as follows:

// File1b.cs
namespace Outer.Inner
{
    using System;
    class Foo
    {
        static void Bar()
        {
            double d = Math.PI;
        }
    }
}

Now the compiler searches System before searching Outer, finds System.Math, and all is well.

Some would argue that Math might be a bad name for a user-defined class, since there's already one in System; the point here is just that there is a difference, and it affects the maintainability of your code.

It's also interesting to note what happens if Foo is in namespace Outer, rather than Outer.Inner. In that case, adding Outer.Math in File2 breaks File1 regardless of where the using goes. This implies that the compiler searches the innermost enclosing namespace before it looks at any using directive.

495
votes

This thread already has some great answers, but I feel I can bring a little more detail with this additional answer.

First, remember that a namespace declaration with periods, like:

namespace MyCorp.TheProduct.SomeModule.Utilities
{
    ...
}

is entirely equivalent to:

namespace MyCorp
{
    namespace TheProduct
    {
        namespace SomeModule
        {
            namespace Utilities
            {
                ...
            }
        }
    }
}

If you wanted to, you could put using directives on all of these levels. (Of course, we want to have usings in only one place, but it would be legal according to the language.)

The rule for resolving which type is implied, can be loosely stated like this: First search the inner-most "scope" for a match, if nothing is found there go out one level to the next scope and search there, and so on, until a match is found. If at some level more than one match is found, if one of the types are from the current assembly, pick that one and issue a compiler warning. Otherwise, give up (compile-time error).

Now, let's be explicit about what this means in a concrete example with the two major conventions.

(1) With usings outside:

using System;
using System.Collections.Generic;
using System.Linq;
//using MyCorp.TheProduct;  <-- uncommenting this would change nothing
using MyCorp.TheProduct.OtherModule;
using MyCorp.TheProduct.OtherModule.Integration;
using ThirdParty;

namespace MyCorp.TheProduct.SomeModule.Utilities
{
    class C
    {
        Ambiguous a;
    }
}

In the above case, to find out what type Ambiguous is, the search goes in this order:

  1. Nested types inside C (including inherited nested types)
  2. Types in the current namespace MyCorp.TheProduct.SomeModule.Utilities
  3. Types in namespace MyCorp.TheProduct.SomeModule
  4. Types in MyCorp.TheProduct
  5. Types in MyCorp
  6. Types in the null namespace (the global namespace)
  7. Types in System, System.Collections.Generic, System.Linq, MyCorp.TheProduct.OtherModule, MyCorp.TheProduct.OtherModule.Integration, and ThirdParty

The other convention:

(2) With usings inside:

namespace MyCorp.TheProduct.SomeModule.Utilities
{
    using System;
    using System.Collections.Generic;
    using System.Linq;
    using MyCorp.TheProduct;                           // MyCorp can be left out; this using is NOT redundant
    using MyCorp.TheProduct.OtherModule;               // MyCorp.TheProduct can be left out
    using MyCorp.TheProduct.OtherModule.Integration;   // MyCorp.TheProduct can be left out
    using ThirdParty;

    class C
    {
        Ambiguous a;
    }
}

Now, search for the type Ambiguous goes in this order:

  1. Nested types inside C (including inherited nested types)
  2. Types in the current namespace MyCorp.TheProduct.SomeModule.Utilities
  3. Types in System, System.Collections.Generic, System.Linq, MyCorp.TheProduct, MyCorp.TheProduct.OtherModule, MyCorp.TheProduct.OtherModule.Integration, and ThirdParty
  4. Types in namespace MyCorp.TheProduct.SomeModule
  5. Types in MyCorp
  6. Types in the null namespace (the global namespace)

(Note that MyCorp.TheProduct was a part of "3." and was therefore not needed between "4." and "5.".)

Concluding remarks

No matter if you put the usings inside or outside the namespace declaration, there's always the possibility that someone later adds a new type with identical name to one of the namespaces which have higher priority.

Also, if a nested namespace has the same name as a type, it can cause problems.

It is always dangerous to move the usings from one location to another because the search hierarchy changes, and another type may be found. Therefore, choose one convention and stick to it, so that you won't have to ever move usings.

Visual Studio's templates, by default, put the usings outside of the namespace (for example if you make VS generate a new class in a new file).

One (tiny) advantage of having usings outside is that you can then utilize the using directives for a global attribute, for example [assembly: ComVisible(false)] instead of [assembly: System.Runtime.InteropServices.ComVisible(false)].

199
votes

Putting it inside the namespaces makes the declarations local to that namespace for the file (in case you have multiple namespaces in the file) but if you only have one namespace per file then it doesn't make much of a difference whether they go outside or inside the namespace.

using ThisNamespace.IsImported.InAllNamespaces.Here;

namespace Namespace1
{ 
   using ThisNamespace.IsImported.InNamespace1.AndNamespace2;

   namespace Namespace2
   { 
      using ThisNamespace.IsImported.InJustNamespace2;
   }       
}

namespace Namespace3
{ 
   using ThisNamespace.IsImported.InJustNamespace3;
}
61
votes

According to Hanselman - Using Directive and Assembly Loading... and other such articles there is technically no difference.

My preference is to put them outside of namespaces.

55
votes

According the to StyleCop Documentation:

SA1200: UsingDirectivesMustBePlacedWithinNamespace

Cause A C# using directive is placed outside of a namespace element.

Rule Description A violation of this rule occurs when a using directive or a using-alias directive is placed outside of a namespace element, unless the file does not contain any namespace elements.

For example, the following code would result in two violations of this rule.

using System;
using Guid = System.Guid;

namespace Microsoft.Sample
{
    public class Program
    {
    }
}

The following code, however, would not result in any violations of this rule:

namespace Microsoft.Sample
{
    using System;
    using Guid = System.Guid;

    public class Program
    {
    }
}

This code will compile cleanly, without any compiler errors. However, it is unclear which version of the Guid type is being allocated. If the using directive is moved inside of the namespace, as shown below, a compiler error will occur:

namespace Microsoft.Sample
{
    using Guid = System.Guid;
    public class Guid
    {
        public Guid(string s)
        {
        }
    }

    public class Program
    {
        public static void Main(string[] args)
        {
            Guid g = new Guid("hello");
        }
    }
}

The code fails on the following compiler error, found on the line containing Guid g = new Guid("hello");

CS0576: Namespace 'Microsoft.Sample' contains a definition conflicting with alias 'Guid'

The code creates an alias to the System.Guid type called Guid, and also creates its own type called Guid with a matching constructor interface. Later, the code creates an instance of the type Guid. To create this instance, the compiler must choose between the two different definitions of Guid. When the using-alias directive is placed outside of the namespace element, the compiler will choose the local definition of Guid defined within the local namespace, and completely ignore the using-alias directive defined outside of the namespace. This, unfortunately, is not obvious when reading the code.

When the using-alias directive is positioned within the namespace, however, the compiler has to choose between two different, conflicting Guid types both defined within the same namespace. Both of these types provide a matching constructor. The compiler is unable to make a decision, so it flags the compiler error.

Placing the using-alias directive outside of the namespace is a bad practice because it can lead to confusion in situations such as this, where it is not obvious which version of the type is actually being used. This can potentially lead to a bug which might be difficult to diagnose.

Placing using-alias directives within the namespace element eliminates this as a source of bugs.

  1. Multiple Namespaces

Placing multiple namespace elements within a single file is generally a bad idea, but if and when this is done, it is a good idea to place all using directives within each of the namespace elements, rather than globally at the top of the file. This will scope the namespaces tightly, and will also help to avoid the kind of behavior described above.

It is important to note that when code has been written with using directives placed outside of the namespace, care should be taken when moving these directives within the namespace, to ensure that this is not changing the semantics of the code. As explained above, placing using-alias directives within the namespace element allows the compiler to choose between conflicting types in ways that will not happen when the directives are placed outside of the namespace.

How to Fix Violations To fix a violation of this rule, move all using directives and using-alias directives within the namespace element.

35
votes

There is an issue with placing using statements inside the namespace when you wish to use aliases. The alias doesn't benefit from the earlier using statements and has to be fully qualified.

Consider:

namespace MyNamespace
{
    using System;
    using MyAlias = System.DateTime;

    class MyClass
    {
    }
}

versus:

using System;

namespace MyNamespace
{
    using MyAlias = DateTime;

    class MyClass
    {
    }
}

This can be particularly pronounced if you have a long-winded alias such as the following (which is how I found the problem):

using MyAlias = Tuple<Expression<Func<DateTime, object>>, Expression<Func<TimeSpan, object>>>;

With using statements inside the namespace, it suddenly becomes:

using MyAlias = System.Tuple<System.Linq.Expressions.Expression<System.Func<System.DateTime, object>>, System.Linq.Expressions.Expression<System.Func<System.TimeSpan, object>>>;

Not pretty.

8
votes

One wrinkle I ran into (that isn't covered in other answers):

Suppose you have these namespaces:

  • Something.Other
  • Parent.Something.Other

When you use using Something.Other outside of a namespace Parent, it refers to the first one (Something.Other).

However if you use it inside of that namespace declaration, it refers to the second one (Parent.Something.Other)!

There is a simple solution: add the "global::" prefix: docs

namespace Parent
{
   using global::Something.Other;
   // etc
}
6
votes

As Jeppe Stig Nielsen said, this thread already has great answers, but I thought this rather obvious subtlety was worth mentioning too.

using directives specified inside namespaces can make for shorter code since they don't need to be fully qualified as when they're specified on the outside.

The following example works because the types Foo and Bar are both in the same global namespace, Outer.

Presume the code file Foo.cs:

namespace Outer.Inner
{
    class Foo { }
}

And Bar.cs:

namespace Outer
{
    using Outer.Inner;

    class Bar
    {
        public Foo foo;
    }
}

That may omit the outer namespace in the using directive, for short:

namespace Outer
{
    using Inner;

    class Bar
    {
        public Foo foo;
    }
}
5
votes

Another subtlety that I don't believe has been covered by the other answers is for when you have a class and namespace with the same name.

When you have the import inside the namespace then it will find the class. If the import is outside the namespace then the import will be ignored and the class and namespace have to be fully qualified.

//file1.cs
namespace Foo
{
    class Foo
    {
    }
}

//file2.cs
namespace ConsoleApp3
{
    using Foo;
    class Program
    {
        static void Main(string[] args)
        {
            //This will allow you to use the class
            Foo test = new Foo();
        }
    }
}

//file2.cs
using Foo; //Unused and redundant    
namespace Bar
{
    class Bar
    {
        Bar()
        {
            Foo.Foo test = new Foo.Foo();
            Foo test = new Foo(); //will give you an error that a namespace is being used like a class.
        }
    }
}
2
votes

The technical reasons are discussed in the answers and I think that it comes to the personal preferences in the end since the difference is not that big and there are tradeoffs for both of them. Visual Studio's default template for creating .cs files use using directives outside of namespaces e.g.

One can adjust stylecop to check using directives outside of namespaces through adding stylecop.json file in the root of the project file with the following:

{
  "$schema": "https://raw.githubusercontent.com/DotNetAnalyzers/StyleCopAnalyzers/master/StyleCop.Analyzers/StyleCop.Analyzers/Settings/stylecop.schema.json",
    "orderingRules": {
      "usingDirectivesPlacement": "outsideNamespace"
    }
  }
}

You can create this config file in solution level and add it to your projects as 'Existing Link File' to share the config across all of your projects too.

0
votes

When citing Microsoft's internal guidelines, keep in mind that they're written by somebody who probably has less than ten years experience coding. In other words, they're likely based on nothing more solid than personal preference. Especially in something like C# that's so new.

As a rule, external using directives (System and Microsoft namespaces for example) should be placed outside the namespace directive. They are defaults that should be applied in all cases unless otherwise specified. This should include any of your own organization's internal libraries that are not part of the current project, or using directives that reference other primary namespaces in the same project. Any using directives that reference other modules in the current project and namespace should be placed inside the namespace directive. This serves two specific functions:

  • It provides a visual distinction between local modules and 'other' modules, meaning everything else.
  • It scopes the local directives to be applied preferentially over global directives.

The latter reason is significant. It means that it's harder to introduce an ambiguous reference issue that can be introduced by a change no more significant than refactoring code. That is to say, you move a method from one file to another and suddenly a bug shows up that wasn't there before. Colloquially, a 'heisenbug' - historically fiendishly difficult to track down.

As an even more general rule, a good one to follow is this. If you see something intrinsic to a language that seems to be a useless option, assume that it's NOT. In fact, the harder it is to see why the option exists, the more important you should assume it is. Do the research about the specific differences between the two options and then think long and hard about the implications. You'll usually find an amazingly insightful and clever solution to an obscure problem that the language designer put in there specifically to make your life easier. Be appropriately grateful and take advantage of it.

-7
votes

It is a better practice if those default using i.e. "references" used in your source solution should be outside the namespaces and those that are "new added reference" is a good practice is you should put it inside the namespace. This is to distinguish what references are being added.