doxygen, among other things, creates an on-line documentation browser in HTML for source code. It supports C++, C, Java, IDL and to some extent Objective-C, PHP, C# and D, but this is largely due to the similarity of these languages' syntax. haddock provides most of the functionality of doxygen for the Haskell language, however it does not generate hyper-linked source code listings. A tool that subsumes the functionality of both and allows generation of documents for arbitrary languages would therefore be of great utility to the wider programming community.

This would require the creation of a meta-language for describing the manner in which declarations (be they functions, types, classes etc) are linked to their uses. A typical invocation of the tool would take as inputs a specification written in the aforementioned meta-language and the source code and generate documents accordingly.

Naturally, the tool will need to understand the meta-language. Traditionally, one would implement a lexer, parser and various analysis phases in order to glean meaning from specifications written in the meta-language. However, we suggest embedding the language in Haskell and loading the specifications using Donald Stewart's hsplugin tool.

• Core requirements for successful completion:
  • Develop a language for the specification of relationships between definition and uses in arbitrary source languages.
  • Implement a tool that can read specifications written in this language and generate documents from source files. To show the validity of this approach it would only be necessary to produce the hyper-linked source code listings of doxygen
• Extra tasks for the ambitious:
  • Provide the ability to annotate source code with documentation directives. Since these directives must interact with the commenting mechanism of the language for which you wish to generate documents for a mechanism should be provided in the meta-language for user-definition of the structure of such annotations. Demonstrate implementation by running the tool on C and Haskell source files readable by doxygen and haddock respectively.
  • Implement other features of doxygen such as the generation of class diagrams. Better yet, extend the meta-language to handle arbitrary generation of documents based on the grammatical structure of the source language.

• doxygen: a tool for automatically generating documentation from annotated C++, C, Java and IDL code.
• doxygen manual
• haddock: a tool for automatically generating documentation from annotated Haskell source code.
• lxr: Another tool for the cross referencing of identifiers in C code. Used on the linux kernel.
• An example of a hyper-linked source code listing produce with doxygen
• hs-plugins: Compiler and tool support for compiling and loading Haskell source files as plugins into a Haskell application at runtime. The library provides a convenient interface to GHC's runtime loader and linker, letting you load compiled Haskell code.