wiki:Commentary/CodingStyle

Version 1 (modified by simonpj, 8 years ago) (diff)

--

The GHC Commentary - Coding Style Guidelines

This is a rough description of some of the coding practices and style that we use for Haskell code inside compiler.

The general rule is to stick to the same coding style as is already used in the file you're editing. If you must make stylistic changes, commit them separately from functional changes, so that someone looking back through the change logs can easily distinguish them.

To literate or not to literate?

In GHC we use a mixture of literate (.lhs) and non-literate (.hs) source. I (Simon M.) prefer to use non-literate style, because I think the \begin{code}..\end{code} clutter up the source too much, and I like to use Haddock-style comments (we haven't tried processing the whole of GHC with Haddock yet, though).

To CPP or not to CPP?

We pass all the compiler sources through CPP. The -cpp flag is always added by the build system. The following CPP symbols are used throughout the compiler:

DEBUG
Used to enables extra checks and debugging output in the compiler. The ASSERT macro (see HsVersions?.h) provides assertions which disappear when DEBUG is not defined.

All debugging output should be placed inside #ifdef DEBUG; we generally use this to provide warnings about strange cases and things that might warrant investigation. When DEBUG is off, the compiler should normally be silent unless something goes wrong (exception when the verbosity level is greater than zero).

A good rule of thumb is that DEBUG shouldn't add more than about 10-20% to the compilation time. This is the case at the moment. If it gets too expensive, we won't use it. For more expensive runtime checks, consider adding a flag - see for example -dcore-lint.

GHCI
Enables GHCi support, including the byte code generator and interactive user interface. This isn't the default, because the compiler needs to be bootstrapped with itself in order for GHCi to work properly. The reason is that the byte-code compiler and linker are quite closely tied to the runtime system, so it is essential that GHCi is linked with the most up-to-date RTS. Another reason is that the representation of certain datatypes must be consistent between GHCi and its libraries, and if these were inconsistent then disaster could follow.
Platform tests
There are three platforms of interest to GHC:
  • The Build platform: This is the platform on which we are building GHC.
  • The Host platform: This is the platform on which we are going to run this GHC binary, and associated tools.
  • The Target platform: This is the platform for which this GHC binary will generate code. At the moment, there is very limited support for having different values for buil, host, and target. In particular:

The build platform is currently always the same as the host platform. The build process needs to use some of the tools in the source tree, for example ghc-pkg and hsc2hs.

If the target platform differs from the host platform, then this is generally for the purpose of building .hc files from Haskell source for porting GHC to the target platform. Full cross-compilation isn't supported (yet). In the compiler's source code, you may make use of the following CPP symbols:

xxx_TARGET_ARCH 
xxx_TARGET_VENDOR 
xxx_TARGET_OS 
xxx_HOST_ARCH 
xxx_HOST_VENDOR 
xxx_HOST_OS 

where xxx is the appropriate value: eg. i386_TARGET_ARCH.

Compiler versions

GHC must be compilable by every major version of GHC from 5.02 onwards, and itself. It isn't necessary for it to be compilable by every intermediate development version (that includes last week's CVS sources).

To maintain compatibility, use HsVersions.h (see below) where possible, and try to avoid using #ifdef in the source itself.

The source file

We now describe a typical source file, annotating stylistic choices as we go.

The OPTIONS pragma

An {-# OPTIONS ... #-} pragma is optional, but if present it should go right at the top of the file. Things you might want to put in OPTIONS include:

  • #include options to bring into scope prototypes for FFI declarations
  • -fvia-C if you know that this module won't compile with the native code generator.

Don't bother putting -cpp or -fglasgow-exts in the OPTIONS pragma; these are already added to the command line by the build system.

Exports

module Foo (
   T(..),
   foo,	     -- :: T -> T
 ) where

We usually (99% of the time) include an export list. The only exceptions are perhaps where the export list would list absolutely everything in the module, and even then sometimes we do it anyway.

It's helpful to give type signatures inside comments in the export list, but hard to keep them consistent, so we don't always do that.

HsVersions.h

HsVersions.h is a CPP header file containing a number of macros that help smooth out the differences between compiler versions. It defines, for example, macros for library module names which have moved between versions. Take a look.

#include "HsVersions.h"

Imports

List imports in the following order:

  • Local to this subsystem (or directory) first
  • Compiler imports, generally ordered from specific to generic (ie. modules from utils/ and basicTypes/ usually come last)
  • Library imports
  • Standard Haskell 98 imports last
    -- friends
    import SimplMonad
    
    -- GHC
    import CoreSyn
    import Id           ( idName, idType )
    import BasicTypes
    
    -- libraries
    import DATA_IOREF   ( newIORef, readIORef )
    
    -- std
    import List         ( partition )
    import Maybe        ( fromJust )
    

Import library modules from the base and haskell98 packages only. Use #defines in HsVersions?.h when the modules names differ between versions of GHC (eg. DATA_IOREF in the example above). For code inside #ifdef GHCI, don't need to worry about GHC versioning (because we are bootstrapped).

We usually use import specs to give an explicit list of the entities imported from a module. The main reason for doing this is so that you can search the file for an entity and find which module it comes from. However, huge import lists can be a pain to maintain, so we often omit the import specs when they start to get long (actually I start omitting them when they don't fit on one line --Simon M.). Tip: use GHC's -fwarn-unused-imports flag so that you get notified when an import isn't being used any more.

If the module can be compiled multiple ways (eg. GHCI vs. non-GHCI), make sure the imports are properly #ifdefed too, so as to avoid spurious unused import warnings.

ToDo?: finish this

Attachments (1)

Download all attachments as: .zip