C# using namespace directive in nested namespaces

Question

Right, I've usually used 'using' directives as follows

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;

namespace AwesomeLib
{
  //awesome award winning class declarations making use of Linq
}

i've recently seen examples of such as

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;

namespace AwesomeLib
{
  //awesome award winning class declarations making use of Linq

  namespace DataLibrary
  {
    using System.Data;

    //Data access layers and whatnot
  }

}

Granted, i understand that i can put USING inside of my namespace declaration. Such a thing makes sense to me if your namespaces are in the same root (they organized).

System;
namespace 1 {}
namespace 2 
{
  System.data;
}

But what of nested namespaces? Personally, I would leave all USING declarations at the top where you can find them easily. Instead, it looks like they're being spread all over the source file.

Is there benefit to the USING directives being used this way in nested namespaces? Such as memory management or the JIT compiler?

Answer 1

IL Code does Not reflect C# using

Runtime Performance

Is there benefit to the USING directives being used this way in nested namespaces? Such as memory management or the JIT compiler?

Because you're asking about runtime performance, here's a look into what's happening underneath the source code.

If you look at the compiled IL code with Microsoft's IL Diassembler tool (as we're doing here) you will see all class names are fully qualified all the time no matter how the programmer used using in the source code.

In the following sample of compiled IL code notice no "shortcut" mechanism is seen although using was in the original C# source code files. For example IL describes one long extends [System.Web]System.Web.UI.Page whereas C# would have used : Page and also using System.Web.UI; (two separate statements).

// ***** Compiled MSIL CODE ****
// Notice all fully qualified classes throughout.
// 

.class public auto ansi beforefieldinit WebApplication1.process
       extends [System.Web]System.Web.UI.Page
{
  .field family class [System.Web]System.Web.UI.HtmlControls.HtmlForm form1
  .method family hidebysig instance void 
          Page_Load(object sender,
                    class [mscorlib]System.EventArgs e) cil managed
  {
    // Code size       95 (0x5f)
    .maxstack  4
    .locals init ([0] string strName,
             [1] string strTime)
    IL_0000:  nop
    IL_0001:  ldarg.0
    IL_0002:  call       instance class [System.Web]System.Web.HttpRequest [System.Web]System.Web.UI.Page::get_Request()
    IL_0007:  ldstr      "name"
    IL_000c:  callvirt   instance string [System.Web]System.Web.HttpRequest::get_Item(string)

In the compiled IL all classes are fully qualified regardless.

This means there are no performance benefits or drawbacks at runtime based on the design time using statements.

Compile Time

Depending on how you scatter about your usings and namespaces in the source code, there might be more or less of the keywords hanging around. The compiler has to see them all and process them all but overall the compile performance would be negligible for something this trivial, compared to all things a compiler has to do to make the finished product.

Design Time Benefits

namespaces are an organizational technique and using is a way of managing them at the source code level (and to instruct the compiler how you're using them so it can compile the program accordingly). When C# source specifies using System.Web.UI;, nothing is imported and the file size doesn't grow larger (because the assembly is already referenced in); instead using is simply effecting a shorter syntax to the contents of that namespace, from within the scope the using is used in, whether that be in the entire file scope or a declared namespace scope inside the file.

The benefit to the programmer is reduction of ambiguous class name conflicts between multiple namespace usings if they are judiciously used.

Organization of source code namespaces is represented differently in the compiled IL code (as seen in the above example).

Answer 2

When you are in a scope that doesn't need System.Data it is better if the IntelliSense doesn't trigger the System.Data members just to mess with other things you are looking for

About some performance issue - I don't think it matters how you prefer to use it...

Answer 3

There's no benefit to the generated intermediate language. Since the CIL is the input for the JIT compiler, this is also unaffected.

Since you're speaking about best practises: it's best practise is to have only one class per file, so you'd have only one namespace as well.

Answer 4

From the Style Cop rules:

1. Placing using-alias directives within the namespace eliminates compiler confusion between conflicting types.
2. When multiple namespaces are defined within a single file, placing using directives within the namespace elements scopes references and aliases.

Answer 5

The using clauses only help making your code more maintainable, they don't affect the performance of your code AFAIK.

The same applies to namespaces - you could put all your classes in the global namespace and name them Class1, Class2 and so on, and your code would perform just as well. But it would be a nightmare to maintain!

So they are both there to help you, the coder, they don't affect the compiled code.