Mixins

Mixins (not to be confused with template mixins) enable string constants to be compiled as regular D code and inserted into the program. Combining this with compile time manipulation of strings enables the creation of domain-specific languages.

For example, here we can create a template that generates a struct with the named members:

template GenStruct(string Name, string M1)
{
    const char[] GenStruct = "struct " ~ Name ~ "{ int " ~ M1 ~ "; }";
}

mixin(GenStruct!("Foo", "bar"));

which generates:

struct Foo { int bar; }

Superficially, since D mixins can manipulate text and compile the result, it has some similar properties to the C preprocessor. But there are major, fundamental differences:

The C preprocessing step occurs before lexical analysis. This makes it impossible to lex or parse C without access to all of the context, including all #include'd files, paths and all relevant compiler switches. Mixins occur during semantic analysis, and do not affect the lexing or parsing process. Lexing and parsing can still occur without semantic analysis.
The C preprocessor can be used to create what appears to be different syntax:
```
#define BEGIN {
#define END   }

BEGIN
  int x = 3;
  foo(x);
END
```
This monkey business is impossible with mixins. Mixed in text must form complete declarations, statements, or expressions.
C macros will affect everything following that has the same name, even if they are in nested scopes. C macros cut across all scopes. This problem is called being not "coding hygenic". Mixins follow the usual scoping rules, and is hygenic.
C preprocessing expressions follow a different syntax and have different semantic rules than the C language. The C preprocessor is technically a different language. Mixins are in the same language.
C const declarations and C++ templates are invisible to C preprocessing. Mixins can be manipulated using templates and const declarations.

Articles

Slides

Tools

Forums

Community

Appendices

Books

Mixins