Back to files

README file of the MetaC Compiler "metacc"

What is MetaC?

MetaC is a metalanguage that is based on the C language. Programs written in MetaC have two parts: a meta-program and a regular program. The meta-program is executed at compile-time and the regular program provides the code for execution at run-time.

With MetaC you can analyze and modify the regular program.

Getting started:

To get started, first read this file, then go to the examples directory and take a look and compile the provided MetaC programs. First file to start with is hello.mc (01.mc) and follow the sequence of the dot-mc files.

To build a MetaC program, you have to go through the following steps: - Compile the MetaC program:
$ metacc -c example.mc
This will produce a file example.i
- Compile the generated C source code with a normal C compiler: $ cc example.i -o example
- Execute the program:
$ ./example

Some MetaC files are based on the 1999 version of the C standard. As support for this standard is incomplete up to now, you must manually enable it with "-std=c99". Additionally, certain extensions of the GNU compiler collection are supported. These must be enabled with the "-xgnu" flag on the command line. "metacc" provides even more options that can be set on the command line. For a full list of available switches and arguments type:
$ metacc -help

During the execution of metacc the meta-program is run. After this all meta-functions and meta variables are stripped off the source code and the source code is printed to the standard output, if you didn't provide the -c flag or set explicitly an output file with the -o flag.

The MetaC language:

The core of the language is based on two concepts: - Special meta-data-types for referencing certain source code structures, based on their syntax.
- A methodology to define code-structure-patterns that can be used to search for certain code structures in the sources and to instantiate new code.

The following meta-data-types are available: - zed: The keyword zed stands for the letter "Z" that is the name of the set of whole numbers. It is the type of a variable that can hold an integer value that can be used for calculations within meta-programs and to instantiate integer-literals of arbitrary type. - real: The same as zed, but for floating point values. - strg: Refers to a string literal, including the enclosing quotation marks.
- ident: Holds a lexical token of type identifier. I.e. it is a character sequence that can be used to give variables a name. In accordance to the C standard an identifier may use any combination of letters (A-Z, a-z), numbers (0-9), and the underscore (_), provided that the first character is not a number. - expr: Refers to an expression. I.e. binary expression such as (+,|), unary expressions (++,--), postfix expressions (function calls, operator ->), and so on.
- stmt: Refers to a statement. Valid statements are e.g. for-, while-, do-, expression-, if-, switch-statement. - type: Encapsulates the type of an object or function. E.g. int *, void *(*)(int, float, struct st *).
- symb: Refers to the definition of a variable and (indirectly) to its uses. - func: Is a special form of symb that is a symbol that is known to be function that is defined in the current translation-unit. - scope: A metavariable of type scope refers the area of code in which a certain symb is visible. Additionally, it provides access to all definitions visible in this area.

Derived data-types (i.e. functions, arrays, pointers, and structs) can be constructed, too. Union types are unavailable, as the meta-data-types have unspecified layout and size in memory. Arrays are enhanced in a way that they can be resized dynamically, by accessing the element after the last valid element. The newly added element is initialized with a null-equivalent value for all meta-data types. Pointers are implemented as references to objects or to objects within arrays as base reference plus offset. In consequence pointer are valid after a resizing an array with a possibly required relocation of the array-data in memory.

The dynamic semantics of statements of meta-programs written in MetaC are equivalent to their logical base. I.e. an if-statement in a MetaC meta-program has the same runtime behavior as an if-statement in C. The semantics of C sources within MetaC source code is unchanged.

Feedback and bugs:

Please read the file FEEDBACK concerning these topics.