GHC Commentary: The Compiler

The compiler itself is written entirely in Haskell, and lives in the many sub-directories of the compiler directory. Here is a block diagram of its top-level structure:

Contents

• Compiler Module Dependencies (deals with the arcane mutual recursions among GHC's many data types)
• Coding guidelines

• Compiling one module: HscMain
• Key data types (Simon PJ's diagram is attached at the bottom of this document):
  • The source language: HsSyn
  • RdrNames, Modules, and OccNames
  • Names
  • Entities: variables, type constructors, data constructors, and classes.
  • Types: Type and Kind, equality types and coercions
  • The core language
  • The STG language
  • The Cmm language
  • ModIface?, ModDetails?, ModGuts?
• Passes:
  • Renamer
  • Typechecker
  • Desugarer
  • Core->core
    • Strictness analysis
  • Core->CorePrep?
  • CorePrep->Stg?
  • The code generator: Stg->Cmm

• The GHC API
• Symbol names and the Z-encoding
• Template Haskell?
• Wired-in and known-key things
• Packages
• The Finder?
• Backends:
  • C code generator
  • Native code generator

The GHC API is the interface exported by compiler/main/GHC.hs. To compile a Haskell module that uses the GHC API, use the flag -package ghc (in GHC 6.6 and later). GHC itself contains a few front-ends:

• The "one-shot" mode, where GHC compiles each file on the command line separately (eg. ghc -c Foo.hs). This mode is implemented directly on top of HscMain, since it compiles only one file at a time. In fact, this is all that GHC consisted of prior to version 5.00 when GHCi and --make were introduced.

• GHCi, the interactive environment, is implemented in compiler/ghci/InteractiveUI.hs and sits squarely on top of the GHC API.

• --make is almost a trivial client of the GHC API, and is implemented in compiler/main/Main.hs.

• -M, the Makefile dependency generator, is also a client of the GHC API and is implemented in compiler/main/DriverMkDepend.hs.

Note that since GHC is packaged as a single binary, all of these front-ends are present, and there is a single command-line interface implemented in compiler/main/Main.hs.