Questions tagged [compilation]

Compilation is the transformation of source text into some other form or representation. The software tool used is called a compiler. Most compilers process the source text to generate some sort of machine code for a target hardware machine. Some compilers generate the "machine code" for a target virtual machine (e.g. bytecode for a Java Virtual Machine).

In all of these cases the compiler creates new files of the transformed source text and these new files are used in some other processing step up to and including executing directly on hardware or virtual hardware.

Interpretation is the processing of source text by a software tool called an interpreter. Interpreters immediately execute the sense of the source text without generating a new, externally visible form of the source text. Compilers generate a new, externally visible form of the source text which is then executed by some other device whether actual hardware or virtual machine.

The lines between interpreters and compilers have blurred somewhat with the introduction of languages whose tools generate an intermediate language which is then executed on an internal virtual machine. The traditional compiler generates machine code files which are then further processed into an application to execute. The traditional interpreter parses the source text line by line performing some action indicated by the line of source text at the time the line of source is read.

A C or C++ compiler generates object files containing binary code which are then processed by a linker into an application and executed. A Java compiler generates class files containing Java Virtual Machine byte code which are then combined into a Java application and executed.

Engines for scripting languages such as Php and JavaScript may use an internal compiler to generate an intermediate form of the source text which is then executed by an internal virtual machine. For some types of applications the intermediate form is temporarily stored or cached so that if the same script is being used by multiple threads or multiple repetions in a short time span, the overhead of rescaning the source text is reduced to improve efficiency. However these are not considered to be compilers.

Compilation usually involves the following steps:

• Scanning - The scanner is responsible of tokenizing the source code into the smallest chunks of information (keywords, operators, brackets, variables, literals etc.).
• Parsing - The parser is responsible with creating the Abstract Syntax Tree (AST) which is a tree representation of the code according to the source language definition.
• Optimization - The AST representing the code is sent through various optimizers in order to optimize for speed or space (this part is optional).
• Code generation - The code generator creates a linear translated document from the AST and the output language definition.

In many languages and compilers there are additional steps added to the process (like pre-processing), but these are language and compiler specific.

In most cases, where compilation is a part of the build/make/publish process, the output of the compiler will be sent to the linker which will produce the ready-to-use files.

Questions using this tag should be about the compilation process, not about how to write compilers for example (use the compiler tag for that).