81 KiB
% Cabal User Guide: Developing Cabal packages
Quickstart
Lets assume we have created a project directory and already have a Haskell module or two.
Every project needs a name, we'll call this example "proglet".
$ cd proglet/
$ ls
Proglet.hs
It is assumed that (apart from external dependencies) all the files that make up a package live under a common project root directory. This simple example has all the project files in one directory, but most packages will use one or more subdirectories.
To turn this into a Cabal package we need two extra files in the project's root directory:
-
proglet.cabal
: containing package metadata and build information. -
Setup.hs
: usually containing a few standardized lines of code, but can be customized if necessary.
We can create both files manually or we can use cabal init
to create
them for us.
Using "cabal init"
The cabal init
command is interactive. It asks us a number of
questions starting with the package name and version.
$ cabal init
Package name [default "proglet"]?
Package version [default "0.1"]?
...
It also asks questions about various other bits of package metadata. For a package that you never intend to distribute to others, these fields can be left blank.
One of the important questions is whether the package contains a library or an executable. Libraries are collections of Haskell modules that can be re-used by other Haskell libraries and programs, while executables are standalone programs.
What does the package build:
1) Library
2) Executable
Your choice?
For the moment these are the only choices. For more complex packages
(e.g. a library and multiple executables or test suites) the .cabal
file can be edited afterwards.
Finally, cabal init
creates the initial proglet.cabal
and Setup.hs
files, and depending on your choice of license, a LICENSE
file as well.
Generating LICENSE...
Generating Setup.hs...
Generating proglet.cabal...
You may want to edit the .cabal file and add a Description field.
As this stage the proglet.cabal
is not quite complete and before you
are able to build the package you will need to edit the file and add
some build information about the library or executable.
Editing the .cabal file
Load up the .cabal
file in a text editor. The first part of the
.cabal
file has the package metadata and towards the end of the file
you will find the executable
or library
section.
You will see that the fields that have yet to be filled in are commented
out. Cabal files use "--
" Haskell-style comment syntax. (Note that
comments are only allowed on lines on their own. Trailing comments on
other lines are not allowed because they could be confused with program
options.)
If you selected earlier to create a library package then your .cabal
file will have a section that looks like this:
library
exposed-modules: Proglet
-- other-modules:
-- build-depends:
Alternatively, if you selected an executable then there will be a section like:
executable proglet
-- main-is:
-- other-modules:
-- build-depends:
The build information fields listed (but commented out) are just the few most important and common fields. There are many others that are covered later in this chapter.
Most of the build information fields are the same between libraries and
executables. The difference is that libraries have a number of "exposed"
modules that make up the public interface of the library, while
executables have a file containing a Main
module.
The name of a library always matches the name of the package, so it is not specified in the library section. Executables often follow the name of the package too, but this is not required and the name is given explicitly.
Modules included in the package
For a library, cabal init
looks in the project directory for files
that look like Haskell modules and adds all the modules to the
exposed-modules
field. For modules that do not form part of your
package's public interface, you can move those modules to the
other-modules
field. Either way, all modules in the library need to be
listed.
For an executable, cabal init
does not try to guess which file
contains your program's Main
module. You will need to fill in the
main-is
field with the file name of your program's Main
module
(including .hs
or .lhs
extension). Other modules included in the
executable should be listed in the other-modules
field.
Modules imported from other packages
While your library or executable may include a number of modules, it almost certainly also imports a number of external modules from the standard libraries or other pre-packaged libraries. (These other libraries are of course just Cabal packages that contain a library.)
You have to list all of the library packages that your library or executable imports modules from. Or to put it another way: you have to list all the other packages that your package depends on.
For example, suppose the example Proglet
module imports the module
Data.Map
. The Data.Map
module comes from the containers
package,
so we must list it:
library
exposed-modules: Proglet
other-modules:
build-depends: containers, base == 4.*
In addition, almost every package also depends on the base
library
package because it exports the standard Prelude
module plus other
basic modules like Data.List
.
You will notice that we have listed base == 4.*
. This gives a
constraint on the version of the base package that our package will work
with. The most common kinds of constraints are:
pkgname >= n
pkgname >= n && < m
pkgname == n.*
The last is just shorthand, for example base == 4.*
means exactly the
same thing as base >= 4 && < 5
.
Building the package
For simple packages that's it! We can now try configuring and building the package:
cabal configure
cabal build
Assuming those two steps worked then you can also install the package:
cabal install
For libraries this makes them available for use in GHCi or to be used by
other packages. For executables it installs the program so that you can
run it (though you may first need to adjust your system's $PATH
).
Next steps
What we have covered so far should be enough for very simple packages that you use on your own system.
The next few sections cover more details needed for more complex packages and details needed for distributing packages to other people.
The previous chapter covers building and installing packages -- your own packages or ones developed by other people.
Package concepts
Before diving into the details of writing packages it helps to understand a bit about packages in the Haskell world and the particular approach that Cabal takes.
The point of packages
Packages are a mechanism for organising and distributing code. Packages are particularly suited for "programming in the large", that is building big systems by using and re-using code written by different people at different times.
People organise code into packages based on functionality and dependencies. Social factors are also important: most packages have a single author, or a relatively small team of authors.
Packages are also used for distribution: the idea is that a package can be created in one place and be moved to a different computer and be usable in that different environment. There are a surprising number of details that have to be got right for this to work, and a good package system helps to simply this process and make it reliable.
Packages come in two main flavours: libraries of reusable code, and complete programs. Libraries present a code interface, an API, while programs can be run directly. In the Haskell world, library packages expose a set of Haskell modules as their public interface. Cabal packages can contain a library or executables or both.
Some programming languages have packages as a builtin language concept. For example in Java, a package provides a local namespace for types and other definitions. In the Haskell world, packages are not a part of the language itself. Haskell programs consist of a number of modules, and packages just provide a way to partition the modules into sets of related functionality. Thus the choice of module names in Haskell is still important, even when using packages.
Package names and versions
All packages have a name, e.g. "HUnit". Package names are assumed to be unique. Cabal package names can use letters, numbers and hyphens, but not spaces. The namespace for Cabal packages is flat, not hierarchical.
Packages also have a version, e.g "1.1". This matches the typical way in which packages are developed. Strictly speaking, each version of a package is independent, but usually they are very similar. Cabal package versions follow the conventional numeric style, consisting of a sequence of digits such as "1.0.1" or "2.0". There are a range of common conventions for "versioning" packages, that is giving some meaning to the version number in terms of changes in the package. Section [TODO] has some tips on package versioning.
The combination of package name and version is called the package ID and is written with a hyphen to separate the name and version, e.g. "HUnit-1.1".
For Cabal packages, the combination of the package name and version uniquely identifies each package. Or to put it another way: two packages with the same name and version are considered to be the same.
Strictly speaking, the package ID only identifies each Cabal source package; the same Cabal source package can be configured and built in different ways. There is a separate installed package ID that uniquely identifies each installed package instance. Most of the time however, users need not be aware of this detail.
Kinds of package: Cabal vs GHC vs system
It can be slightly confusing at first because there are various different notions of package floating around. Fortunately the details are not very complicated.
- Cabal packages
- Cabal packages are really source packages. That is they contain
Haskell (and sometimes C) source code.
Cabal packages can be compiled to produce GHC packages. They can also be translated into operating system packages.
- GHC packages
- This is GHC's view on packages. GHC only cares about library
packages, not executables. Library packages have to be registered
with GHC for them to be available in GHCi or to be used when
compiling other programs or packages.
The low-level tool
ghc-pkg
is used to register GHC packages and to get information on what packages are currently registered.You never need to make GHC packages manually. When you build and install a Cabal package containing a library then it gets registered with GHC automatically.
Haskell implementations other than GHC have essentially the same concept of registered packages. For the most part, Cabal hides the slight differences.
- Operating system packages
- On operating systems like Linux and Mac OS X, the system has a
specific notion of a package and there are tools for installing and
managing packages.
The Cabal package format is designed to allow Cabal packages to be translated, mostly-automatically, into operating system packages. They are usually translated 1:1, that is a single Cabal package becomes a single system package.
It is also possible to make Windows installers from Cabal packages, though this is typically done for a program together with all of its library dependencies, rather than packaging each library separately.
Unit of distribution
The Cabal package is the unit of distribution. What this means is that each Cabal package can be distributed on its own in source or binary form. Of course there may dependencies between packages, but there is usually a degree of flexibility in which versions of packages can work together so distributing them independently makes sense.
It is perhaps easiest to see what being ``the unit of distribution'' means by contrast to an alternative approach. Many projects are made up of several interdependent packages and during development these might all be kept under one common directory tree and be built and tested together. When it comes to distribution however, rather than distributing them all together in a single tarball, it is required that they each be distributed independently in their own tarballs.
Cabal's approach is to say that if you can specify a dependency on a package then that package should be able to be distributed independently. Or to put it the other way round, if you want to distribute it as a single unit, then it should be a single package.
Explicit dependencies and automatic package management
Cabal takes the approach that all packages dependencies are specified
explicitly and specified in a declarative way. The point is to enable
automatic package management. This means tools like cabal
can resolve
dependencies and install a package plus all of its dependencies
automatically. Alternatively, it is possible to mechanically (or mostly
mechanically) translate Cabal packages into system packages and let the
system package manager install dependencies automatically.
It is important to track dependencies accurately so that packages can
reliably be moved from one system to another system and still be able to
build it there. Cabal is therefore relatively strict about specifying
dependencies. For example Cabal's default build system will not even let
code build if it tries to import a module from a package that isn't
listed in the .cabal
file, even if that package is actually installed.
This helps to ensure that there are no "untracked dependencies" that
could cause the code to fail to build on some other system.
The explicit dependency approach is in contrast to the traditional
"./configure" approach where instead of specifying dependencies
declaratively, the ./configure
script checks if the dependencies are
present on the system. Some manual work is required to transform a
./configure
based package into a Linux distribution package (or
similar). This conversion work is usually done by people other than the
package author(s). The practical effect of this is that only the most
popular packages will benefit from automatic package management. Instead,
Cabal forces the original author to specify the dependencies but the
advantage is that every package can benefit from automatic package
management.
The "./configure" approach tends to encourage packages that adapt themselves to the environment in which they are built, for example by disabling optional features so that they can continue to work when a particular dependency is not available. This approach makes sense in a world where installing additional dependencies is a tiresome manual process and so minimising dependencies is important. The automatic package management view is that packages should just declare what they need and the package manager will take responsibility for ensuring that all the dependencies are installed.
Sometimes of course optional features and optional dependencies do make sense. Cabal packages can have optional features and varying dependencies. These conditional dependencies are still specified in a declarative way however and remain compatible with automatic package management. The need to remain compatible with automatic package management means that Cabal's conditional dependencies system is a bit less flexible than with the "./configure" approach.
Portability
One of the purposes of Cabal is to make it easier to build packages on different platforms (operating systems and CPU architectures), with different compiler versions and indeed even with different Haskell implementations. (Yes, there are Haskell implementations other than GHC!)
Cabal provides abstractions of features present in different Haskell implementations and wherever possible it is best to take advantage of these to increase portability. Where necessary however it is possible to use specific features of specific implementations.
For example a package author can list in the package's .cabal
what
language extensions the code uses. This allows Cabal to figure out if
the language extension is supported by the Haskell implementation that
the user picks. Additionally, certain language extensions such as
Template Haskell require special handling from the build system and by
listing the extension it provides the build system with enough
information to do the right thing.
Another similar example is linking with foreign libraries. Rather than specifying GHC flags directly, the package author can list the libraries that are needed and the build system will take care of using the right flags for the compiler. Additionally this makes it easier for tools to discover what system C libraries a package needs, which is useful for tracking dependencies on system libraries (e.g. when translating into Linux distribution packages).
In fact both of these examples fall into the category of explicitly specifying dependencies. Not all dependencies are other Cabal packages. Foreign libraries are clearly another kind of dependency. It's also possible to think of language extensions as dependencies: the package depends on a Haskell implementation that supports all those extensions.
Where compiler-specific options are needed however, there is an "escape hatch" available. The developer can specify implementation-specific options and more generally there is a configuration mechanism to customise many aspects of how a package is built depending on the Haskell implementation, the operating system, computer architecture and user-specified configuration flags.
Developing packages
The Cabal package is the unit of distribution. When installed, its purpose is to make available:
-
One or more Haskell programs.
-
At most one library, exposing a number of Haskell modules.
However having both a library and executables in a package does not work very well; if the executables depend on the library, they must explicitly list all the modules they directly or indirectly import from that library. Fortunately, starting with Cabal 1.8.0.4, executables can also declare the package that they are in as a dependency, and Cabal will treat them as if they were in another package that depended on the library.
Internally, the package may consist of much more than a bunch of Haskell modules: it may also have C source code and header files, source code meant for preprocessing, documentation, test cases, auxiliary tools etc.
A package is identified by a globally-unique package name, which
consists of one or more alphanumeric words separated by hyphens. To
avoid ambiguity, each of these words should contain at least one letter.
Chaos will result if two distinct packages with the same name are
installed on the same system. A particular version of the package is
distinguished by a version number, consisting of a sequence of one or
more integers separated by dots. These can be combined to form a single
text string called the package ID, using a hyphen to separate the name
from the version, e.g. "HUnit-1.1
".
Note: Packages are not part of the Haskell language; they simply populate the hierarchical space of module names. In GHC 6.6 and later a program may contain multiple modules with the same name if they come from separate packages; in all other current Haskell systems packages may not overlap in the modules they provide, including hidden modules.
Creating a package
Suppose you have a directory hierarchy containing the source files that make up your package. You will need to add two more files to the root directory of the package:
- package
.cabal
-
a Unicode UTF-8 text file containing a package description. For details of the syntax of this file, see the section on package descriptions.
Setup.hs
-
a single-module Haskell program to perform various setup tasks (with the interface described in the section on building and installing packages. This module should import only modules that will be present in all Haskell implementations, including modules of the Cabal library. The content of this file is determined by the
build-type
setting in the.cabal
file. In most cases it will be trivial, calling on the Cabal library to do most of the work.
Once you have these, you can create a source bundle of this directory for distribution. Building of the package is discussed in the section on building and installing packages.
One of the purposes of Cabal is to make it easier to build a package
with different Haskell implementations. So it provides abstractions of
features present in different Haskell implementations and wherever
possible it is best to take advantage of these to increase portability.
Where necessary however it is possible to use specific features of
specific implementations. For example one of the pieces of information a
package author can put in the package's .cabal
file is what language
extensions the code uses. This is far preferable to specifying flags for
a specific compiler as it allows Cabal to pick the right flags for the
Haskell implementation that the user picks. It also allows Cabal to
figure out if the language extension is even supported by the Haskell
implementation that the user picks. Where compiler-specific options are
needed however, there is an "escape hatch" available. The developer can
specify implementation-specific options and more generally there is a
configuration mechanism to customise many aspects of how a package is
built depending on the Haskell implementation, the Operating system,
computer architecture and user-specified configuration flags.
name: Foo
version: 1.0
library
build-depends: base
exposed-modules: Foo
extensions: ForeignFunctionInterface
ghc-options: -Wall
if os(windows)
build-depends: Win32
Example: A package containing a simple library
The HUnit package contains a file HUnit.cabal
containing:
name: HUnit
version: 1.1.1
synopsis: A unit testing framework for Haskell
homepage: http://hunit.sourceforge.net/
category: Testing
author: Dean Herington
license: BSD3
license-file: LICENSE
cabal-version: >= 1.10
build-type: Simple
library
build-depends: base >= 2 && < 4
exposed-modules: Test.HUnit.Base, Test.HUnit.Lang,
Test.HUnit.Terminal, Test.HUnit.Text, Test.HUnit
default-extensions: CPP
and the following Setup.hs
:
import Distribution.Simple
main = defaultMain
Example: A package containing executable programs
name: TestPackage
version: 0.0
synopsis: Small package with two programs
author: Angela Author
license: BSD3
build-type: Simple
cabal-version: >= 1.2
executable program1
build-depends: HUnit
main-is: Main.hs
hs-source-dirs: prog1
executable program2
main-is: Main.hs
build-depends: HUnit
hs-source-dirs: prog2
other-modules: Utils
with Setup.hs
the same as above.
Example: A package containing a library and executable programs
name: TestPackage
version: 0.0
synopsis: Package with library and two programs
license: BSD3
author: Angela Author
build-type: Simple
cabal-version: >= 1.2
library
build-depends: HUnit
exposed-modules: A, B, C
executable program1
main-is: Main.hs
hs-source-dirs: prog1
other-modules: A, B
executable program2
main-is: Main.hs
hs-source-dirs: prog2
other-modules: A, C, Utils
with Setup.hs
the same as above. Note that any library modules
required (directly or indirectly) by an executable must be listed again.
The trivial setup script used in these examples uses the simple build infrastructure provided by the Cabal library (see Distribution.Simple). The simplicity lies in its interface rather that its implementation. It automatically handles preprocessing with standard preprocessors, and builds packages for all the Haskell implementations.
The simple build infrastructure can also handle packages where building is governed by system-dependent parameters, if you specify a little more (see the section on system-dependent parameters). A few packages require more elaborate solutions.
Package descriptions
The package description file must have a name ending in ".cabal
". It
must be a Unicode text file encoded using valid UTF-8. There must be
exactly one such file in the directory. The first part of the name is
usually the package name, and some of the tools that operate on Cabal
packages require this.
In the package description file, lines whose first non-whitespace characters
are "--
" are treated as comments and ignored.
This file should contain of a number global property descriptions and several sections.
-
The global properties describe the package as a whole, such as name, license, author, etc.
-
Optionally, a number of configuration flags can be declared. These can be used to enable or disable certain features of a package. (see the section on configurations).
-
The (optional) library section specifies the library properties and relevant build information.
-
Following is an arbitrary number of executable sections which describe an executable program and relevant build information.
Each section consists of a number of property descriptions in the form of field/value pairs, with a syntax roughly like mail message headers.
-
Case is not significant in field names, but is significant in field values.
-
To continue a field value, indent the next line relative to the field name.
-
Field names may be indented, but all field values in the same section must use the same indentation.
-
Tabs are not allowed as indentation characters due to a missing standard interpretation of tab width.
-
To get a blank line in a field value, use an indented "
.
"
The syntax of the value depends on the field. Field types include:
- token, filename, directory
- Either a sequence of one or more non-space non-comma characters, or a quoted string in Haskell 98 lexical syntax. Unless otherwise stated, relative filenames and directories are interpreted from the package root directory.
- freeform, URL, address
- An arbitrary, uninterpreted string.
- identifier
- A letter followed by zero or more alphanumerics or underscores.
- compiler
- A compiler flavor (one of:
GHC
,JHC
,UHC
orLHC
) followed by a version range. For example,GHC ==6.10.3
, orLHC >=0.6 && <0.8
.
Modules and preprocessors
Haskell module names listed in the exposed-modules
and other-modules
fields may correspond to Haskell source files, i.e. with names ending in
".hs
" or ".lhs
", or to inputs for various Haskell preprocessors. The
simple build infrastructure understands the extensions:
When building, Cabal will automatically run the appropriate preprocessor and compile the Haskell module it produces.
Some fields take lists of values, which are optionally separated by commas,
except for the build-depends
field, where the commas are mandatory.
Some fields are marked as required. All others are optional, and unless otherwise specified have empty default values.
Package properties
These fields may occur in the first top-level properties section and describe the package as a whole:
name:
package-name (required)- The unique name of the package, without the version number.
version:
numbers (required)- The package version number, usually consisting of a sequence of natural numbers separated by dots.
cabal-version:
>= x.y- The version of the Cabal specification that this package description uses.
The Cabal specification does slowly evolve, introducing new features and
occasionally changing the meaning of existing features. By specifying
which version of the spec you are using it enables programs which process
the package description to know what syntax to expect and what each part
means.
For historical reasons this is always expressed using >= version range syntax. No other kinds of version range make sense, in particular upper bounds do not make sense. In future this field will specify just a version number, rather than a version range.
The version number you specify will affect both compatibility and behaviour. Most tools (including the Cabal library and cabal program) understand a range of versions of the Cabal specification. Older tools will of course only work with older versions of the Cabal specification. Most of the time, tools that are too old will recognise this fact and produce a suitable error message.
As for behaviour, new versions of the Cabal spec can change the meaning of existing syntax. This means if you want to take advantage of the new meaning or behaviour then you must specify the newer Cabal version. Tools are expected to use the meaning and behaviour appropriate to the version given in the package description.
In particular, the syntax of package descriptions changed significantly with Cabal version 1.2 and the
cabal-version
field is now required. Files written in the old syntax are still recognized, so if you require compatibility with very old Cabal versions then you may write your package description file using the old syntax. Please consult the user's guide of an older Cabal version for a description of that syntax. build-type:
identifier- The type of build used by this package. Build types are the
constructors of the BuildType type, defaulting to
Custom
.If the build type is anything other than
Custom
, then theSetup.hs
file must be exactly the standardized content discussed below. This is because in these cases,cabal
will ignore theSetup.hs
file completely, whereas other methods of package management, such asrunhaskell Setup.hs [CMD]
, still rely on theSetup.hs
file.For build type
Simple
, the contents ofSetup.hs
must be:import Distribution.Simple main = defaultMain
For build type
Configure
(see the section on system-dependent parameters below), the contents ofSetup.hs
must be:import Distribution.Simple main = defaultMainWithHooks autoconfUserHooks
For build type
Make
(see the section on more complex packages below), the contents ofSetup.hs
must be:import Distribution.Make main = defaultMain
For build type
Custom
, the fileSetup.hs
can be customized, and will be used both bycabal
and other tools.For most packages, the build type
Simple
is sufficient. license:
identifier (default:AllRightsReserved
)- The type of license under which this package is distributed. License names are the constants of the License type.
license-file:
filename orlicense-files:
filename list- The name of a file(s) containing the precise copyright license for
this package. The license file(s) will be installed with the package.
If you have multiple license files then use the
license-files
field instead of (or in addition to) thelicense-file
field. copyright:
freeform- The content of a copyright notice, typically the name of the holder
of the copyright on the package and the year(s) from which copyright
is claimed. For example:
Copyright: (c) 2006-2007 Joe Bloggs
author:
freeform- The original author of the package.
Remember that
.cabal
files are Unicode, using the UTF-8 encoding. maintainer:
address- The current maintainer or maintainers of the package. This is an e-mail address to which users should send bug reports, feature requests and patches.
stability:
freeform- The stability level of the package, e.g.
alpha
,experimental
,provisional
,stable
. homepage:
URL- The package homepage.
bug-reports:
URL- The URL where users should direct bug reports. This would normally be either:
-
A
mailto:
URL, e.g. for a person or a mailing list. -
An
http:
(orhttps:
) URL for an online bug tracking system.
For example Cabal itself uses a web-based bug tracking system
bug-reports: http://hackage.haskell.org/trac/hackage/
-
package-url:
URL- The location of a source bundle for the package. The distribution should be a Cabal package.
synopsis:
freeform- A very short description of the package, for use in a table of packages. This is your headline, so keep it short (one line) but as informative as possible. Save space by not including the package name or saying it's written in Haskell.
description:
freeform- Description of the package. This may be several paragraphs, and
should be aimed at a Haskell programmer who has never heard of your
package before.
For library packages, this field is used as prologue text by
setup haddock
, and thus may contain the same markup as haddock documentation comments. category:
freeform- A classification category for future use by the package catalogue Hackage. These categories have not yet been specified, but the upper levels of the module hierarchy make a good start.
tested-with:
compiler list- A list of compilers and versions against which the package has been tested (or at least built).
data-files:
filename list- A list of files to be installed for run-time use by the package.
This is useful for packages that use a large amount of static data,
such as tables of values or code templates. Cabal provides a way to
find these files at
run-time.
A limited form of
*
wildcards in file names, for exampledata-files: images/*.png
matches all the.png
files in theimages
directory.The limitation is that
*
wildcards are only allowed in place of the file name, not in the directory name or file extension. In particular, wildcards do not include directories contents recursively. Furthermore, if a wildcard is used it must be used with an extension, sodata-files: data/*
is not allowed. When matching a wildcard plus extension, a file's full extension must match exactly, so*.gz
matchesfoo.gz
but notfoo.tar.gz
. A wildcard that does not match any files is an error.The reason for providing only a very limited form of wildcard is to concisely express the common case of a large number of related files of the same file type without making it too easy to accidentally include unwanted files.
data-dir:
directory- The directory where Cabal looks for data files to install, relative to the source directory. By default, Cabal will look in the source directory itself.
extra-source-files:
filename list- A list of additional files to be included in source distributions
built with
setup sdist
. As withdata-files
it can use a limited form of*
wildcards in file names. extra-doc-files:
filename list- A list of additional files to be included in source distributions,
and also copied to the html directory when Haddock documentation is
generated. As with
data-files
it can use a limited form of*
wildcards in file names. extra-tmp-files:
filename list- A list of additional files or directories to be removed by
setup clean
. These would typically be additional files created by additional hooks, such as the scheme described in the section on system-dependent parameters.
Library
The library section should contain the following fields:
exposed-modules:
identifier list (required if this package contains a library)- A list of modules added by this package.
exposed:
boolean (default:True
)- Some Haskell compilers (notably GHC) support the notion of packages
being "exposed" or "hidden" which means the modules they provide can
be easily imported without always having to specify which package
they come from. However this only works effectively if the modules
provided by all exposed packages do not overlap (otherwise a module
import would be ambiguous).
Almost all new libraries use hierarchical module names that do not clash, so it is very uncommon to have to use this field. However it may be necessary to set
exposed: False
for some old libraries that use a flat module namespace or where it is known that the exposed modules would clash with other common modules. reexported-modules:
_exportlist _- Supported only in GHC 7.10 and later. A list of modules to reexport from
this package. The syntax of this field is
orig-pkg:Name as NewName
to reexport moduleName
fromorig-pkg
with the new nameNewName
. We also support abbreviated versions of the syntax: if you omitas NewName
, we'll reexport without renaming; if you omitorig-pkg
, then we will automatically figure out which package to reexport from, if it's unambiguous.Reexported modules are useful for compatibility shims when a package has been split into multiple packages, and they have the useful property that if a package provides a module, and another package reexports it under the same name, these are not considered a conflict (as would be the case with a stub module.) They can also be used to resolve name conflicts.
The library section may also contain build information fields (see the section on build information).
Opening an interpreter session
While developing a package, it is often useful to make its code available inside
an interpreter session. This can be done with the repl
command:
cabal repl
The name comes from the acronym REPL, which stands for
"read-eval-print-loop". By default cabal repl
loads the first component in a
package. If the package contains several named components, the name can be given
as an argument to repl
. The name can be also optionally prefixed with the
component's type for disambiguation purposes. Example:
cabal repl foo
cabal repl exe:foo
cabal repl test:bar
cabal repl bench:baz
Freezing dependency versions
If a package is built in several different environments, such as a development
environment, a staging environment and a production environment, it may be
necessary or desirable to ensure that the same dependency versions are
selected in each environment. This can be done with the freeze
command:
cabal freeze
The command writes the selected version for all dependencies to the
cabal.config
file. All environments which share this file will use the
dependency versions specified in it.
Executables
Executable sections (if present) describe executable programs contained in the package and must have an argument after the section label, which defines the name of the executable. This is a freeform argument but may not contain spaces.
The executable may be described using the following fields, as well as build information fields (see the section on build information).
main-is:
filename (required)- The name of the
.hs
or.lhs
file containing theMain
module. Note that it is the.hs
filename that must be listed, even if that file is generated using a preprocessor. The source file must be relative to one of the directories listed inhs-source-dirs
.
Running executables
You can have Cabal build and run your executables by using the run
command:
$ cabal run EXECUTABLE [-- EXECUTABLE_FLAGS]
This command will configure, build and run the executable EXECUTABLE
. The
double dash separator is required to distinguish executable flags from run
's
own flags. If there is only one executable defined in the whole package, the
executable's name can be omitted. See the output of cabal help run
for a list
of options you can pass to cabal run
.
Test suites
Test suite sections (if present) describe package test suites and must have an argument after the section label, which defines the name of the test suite. This is a freeform argument, but may not contain spaces. It should be unique among the names of the package's other test suites, the package's executables, and the package itself. Using test suite sections requires at least Cabal version 1.9.2.
The test suite may be described using the following fields, as well as build information fields (see the section on build information).
type:
interface (required)- The interface type and version of the test suite. Cabal supports two test
suite interfaces, called
exitcode-stdio-1.0
anddetailed-0.9
. Each of these types may require or disallow other fields as described below.
Test suites using the exitcode-stdio-1.0
interface are executables
that indicate test failure with a non-zero exit code when run; they may provide
human-readable log information through the standard output and error channels.
This interface is provided primarily for compatibility with existing test
suites; it is preferred that new test suites be written for the detailed-0.9
interface. The exitcode-stdio-1.0
type requires the main-is
field.
main-is:
filename (required:exitcode-stdio-1.0
, disallowed:detailed-0.9
)- The name of the
.hs
or.lhs
file containing theMain
module. Note that it is the.hs
filename that must be listed, even if that file is generated using a preprocessor. The source file must be relative to one of the directories listed inhs-source-dirs
. This field is analogous to themain-is
field of an executable section.
Test suites using the detailed-0.9
interface are modules exporting the symbol
tests :: IO [Test]
. The Test
type is exported by the module
Distribution.TestSuite
provided by Cabal. For more details, see the example below.
The detailed-0.9
interface allows Cabal and other test agents to inspect a
test suite's results case by case, producing detailed human- and
machine-readable log files. The detailed-0.9
interface requires the
test-module
field.
test-module:
identifier (required:detailed-0.9
, disallowed:exitcode-stdio-1.0
)- The module exporting the
tests
symbol.
Example: Package using exitcode-stdio-1.0
interface
The example package description and executable source file below demonstrate
the use of the exitcode-stdio-1.0
interface. For brevity, the example package
does not include a library or any normal executables, but a real package would
be required to have at least one library or executable.
foo.cabal:
Name: foo
Version: 1.0
License: BSD3
Cabal-Version: >= 1.9.2
Build-Type: Simple
Test-Suite test-foo
type: exitcode-stdio-1.0
main-is: test-foo.hs
build-depends: base
test-foo.hs:
module Main where
import System.Exit (exitFailure)
main = do
putStrLn "This test always fails!"
exitFailure
Example: Package using detailed-0.9
interface
The example package description and test module source file below demonstrate
the use of the detailed-0.9
interface. For brevity, the example package does
note include a library or any normal executables, but a real package would be
required to have at least one library or executable. The test module below
also develops a simple implementation of the interface set by
Distribution.TestSuite
, but in actual usage the implementation would be
provided by the library that provides the testing facility.
bar.cabal:
Name: bar
Version: 1.0
License: BSD3
Cabal-Version: >= 1.9.2
Build-Type: Simple
Test-Suite test-bar
type: detailed-0.9
test-module: Bar
build-depends: base, Cabal >= 1.9.2
Bar.hs:
module Bar ( tests ) where
import Distribution.TestSuite
tests :: IO [Test]
tests = return [ Test succeeds, Test fails ]
where
succeeds = TestInstance
{ run = return $ Finished Pass
, name = "succeeds"
, tags = []
, options = []
, setOption = \_ _ -> Right succeeds
}
fails = TestInstance
{ run = return $ Finished $ Fail "Always fails!"
, name = "fails"
, tags = []
, options = []
, setOption = \_ _ -> Right fails
}
Running test suites
You can have Cabal run your test suites using its built-in test runner:
$ cabal configure --enable-tests
$ cabal build
$ cabal test
See the output of cabal help test
for a list of options you can pass
to cabal test
.
Benchmarks
Benchmark sections (if present) describe benchmarks contained in the package and must have an argument after the section label, which defines the name of the benchmark. This is a freeform argument, but may not contain spaces. It should be unique among the names of the package's other benchmarks, the package's test suites, the package's executables, and the package itself. Using benchmark sections requires at least Cabal version 1.9.2.
The benchmark may be described using the following fields, as well as build information fields (see the section on build information).
type:
interface (required)- The interface type and version of the benchmark. At the moment Cabal only
support one benchmark interface, called
exitcode-stdio-1.0
.
Benchmarks using the exitcode-stdio-1.0
interface are executables that
indicate failure to run the benchmark with a non-zero exit code when run; they
may provide human-readable information through the standard output and error
channels.
main-is:
filename (required:exitcode-stdio-1.0
)- The name of the
.hs
or.lhs
file containing theMain
module. Note that it is the.hs
filename that must be listed, even if that file is generated using a preprocessor. The source file must be relative to one of the directories listed inhs-source-dirs
. This field is analogous to themain-is
field of an executable section.
Example: Package using exitcode-stdio-1.0
interface
The example package description and executable source file below demonstrate
the use of the exitcode-stdio-1.0
interface. For brevity, the example package
does not include a library or any normal executables, but a real package would
be required to have at least one library or executable.
foo.cabal:
Name: foo
Version: 1.0
License: BSD3
Cabal-Version: >= 1.9.2
Build-Type: Simple
Benchmark bench-foo
type: exitcode-stdio-1.0
main-is: bench-foo.hs
build-depends: base, time
bench-foo.hs:
{-# LANGUAGE BangPatterns #-}
module Main where
import Data.Time.Clock
fib 0 = 1
fib 1 = 1
fib n = fib (n-1) + fib (n-2)
main = do
start <- getCurrentTime
let !r = fib 20
end <- getCurrentTime
putStrLn $ "fib 20 took " ++ show (diffUTCTime end start)
Running benchmarks
You can have Cabal run your benchmark using its built-in benchmark runner:
$ cabal configure --enable-benchmarks
$ cabal build
$ cabal bench
See the output of cabal help bench
for a list of options you can
pass to cabal bench
.
Build information
The following fields may be optionally present in a library or executable section, and give information for the building of the corresponding library or executable. See also the sections on system-dependent parameters and configurations for a way to supply system-dependent values for these fields.
build-depends:
package list- A list of packages needed to build this one. Each package can be
annotated with a version constraint.
Version constraints use the operators
==, >=, >, <, <=
and a version number. Multiple constraints can be combined using&&
or||
. If no version constraint is specified, any version is assumed to be acceptable. For example:library build-depends: base >= 2, foo >= 1.2 && < 1.3, bar
Dependencies like
foo >= 1.2 && < 1.3
turn out to be very common because it is recommended practise for package versions to correspond to API versions. As of Cabal 1.6, there is a special syntax to support this use:build-depends: foo ==1.2.*
It is only syntactic sugar. It is exactly equivalent to
foo >= 1.2 && < 1.3
.With Cabal 1.20 and GHC 7.10,
build-depends
also supports module thinning and renaming, which allows you to selectively decide what modules become visible from a package dependency. For example:build-depends: containers (Data.Set, Data.IntMap as Map)
This results in only the modules
Data.Set
andMap
being visible to the user from containers, hiding all other modules. To add additional names for modules without hiding the others, you can use thewith
keyword:build-depends: containers with (Data.IntMap as Map)
Note: Prior to Cabal 1.8, build-depends specified in each section were global to all sections. This was unintentional, but some packages were written to depend on it, so if you need your build-depends to be local to each section, you must specify at least
Cabal-Version: >= 1.8
in your.cabal
file. other-modules:
identifier list- A list of modules used by the component but not exposed to users.
For a library component, these would be hidden modules of the
library. For an executable, these would be auxiliary modules to be
linked with the file named in the
main-is
field.Note: Every module in the package must be listed in one of
other-modules
,exposed-modules
ormain-is
fields. hs-source-dirs:
directory list (default: ".
")- Root directories for the module hierarchy.
For backwards compatibility, the old variant
hs-source-dir
is also recognized. extensions:
identifier list- A list of Haskell extensions used by every module. Extension names
are the constructors of the Extension type. These
determine corresponding compiler options. In particular,
CPP
specifies that Haskell source files are to be preprocessed with a C preprocessor.Extensions used only by one module may be specified by placing a
LANGUAGE
pragma in the source file affected, e.g.:{-# LANGUAGE CPP, MultiParamTypeClasses #-}
Note: GHC versions prior to 6.6 do not support the
LANGUAGE
pragma. build-tools:
program list- A list of programs, possibly annotated with versions, needed to
build this package, e.g.
c2hs >= 0.15, cpphs
.If no version constraint is specified, any version is assumed to be acceptable. buildable:
boolean (default:True
)- Is the component buildable? Like some of the other fields below, this field is more useful with the slightly more elaborate form of the simple build infrastructure described in the section on system-dependent parameters.
ghc-options:
token list- Additional options for GHC. You can often achieve the same effect
using the
extensions
field, which is preferred.Options required only by one module may be specified by placing an
OPTIONS_GHC
pragma in the source file affected. ghc-prof-options:
token list- Additional options for GHC when the package is built with profiling enabled.
ghc-shared-options:
token list- Additional options for GHC when the package is built as shared library.
includes:
filename list- A list of header files to be included in any compilations via C. This field applies to both header files that are already installed on the system and to those coming with the package to be installed. These files typically contain function prototypes for foreign imports used by the package.
install-includes:
filename list- A list of header files from this package to be installed into
$libdir/includes
when the package is installed. Files listed ininstall-includes:
should be found in relative to the top of the source tree or relative to one of the directories listed ininclude-dirs
.install-includes
is typically used to name header files that contain prototypes for foreign imports used in Haskell code in this package, for which the C implementations are also provided with the package. Note that to include them when compiling the package itself, they need to be listed in theincludes:
field as well. include-dirs:
directory list- A list of directories to search for header files, when preprocessing
with
c2hs
,hsc2hs
,cpphs
or the C preprocessor, and also when compiling via C. c-sources:
filename list- A list of C source files to be compiled and linked with the Haskell files.
js-sources:
filename list- A list of JavaScript source files to be linked with the Haskell files (only for JavaScript targets).
extra-libraries:
token list- A list of extra libraries to link with.
extra-ghci-libraries:
token list- A list of extra libraries to be used instead of 'extra-libraries' when the package is loaded with GHCi.
extra-lib-dirs:
directory list- A list of directories to search for libraries.
cc-options:
token list- Command-line arguments to be passed to the C compiler. Since the arguments are compiler-dependent, this field is more useful with the setup described in the section on system-dependent parameters.
cpp-options:
token list- Command-line arguments for pre-processing Haskell code. Applies to haskell source and other pre-processed Haskell source like .hsc .chs. Does not apply to C code, that's what cc-options is for.
ld-options:
token list- Command-line arguments to be passed to the linker. Since the arguments are compiler-dependent, this field is more useful with the setup described in the section on system-dependent parameters>.
pkgconfig-depends:
package list- A list of pkg-config packages, needed to build this package.
They can be annotated with versions, e.g. `gtk+-2.0 >= 2.10, cairo
= 1.0
. If no version constraint is specified, any version is assumed to be acceptable. Cabal uses
pkg-config` to find if the packages are available on the system and to find the extra compilation and linker options needed to use the packages.If you need to bind to a C library that supports
pkg-config
(usepkg-config --list-all
to find out if it is supported) then it is much preferable to use this field rather than hard code options into the other fields. frameworks:
token list- On Darwin/MacOS X, a list of frameworks to link to. See Apple's developer documentation for more details on frameworks. This entry is ignored on all other platforms.
Configurations
Library and executable sections may include conditional blocks, which test for various system parameters and configuration flags. The flags mechanism is rather generic, but most of the time a flag represents certain feature, that can be switched on or off by the package user. Here is an example package description file using configurations:
Example: A package containing a library and executable programs
Name: Test1
Version: 0.0.1
Cabal-Version: >= 1.2
License: BSD3
Author: Jane Doe
Synopsis: Test package to test configurations
Category: Example
Flag Debug
Description: Enable debug support
Default: False
Flag WebFrontend
Description: Include API for web frontend.
-- Cabal checks if the configuration is possible, first
-- with this flag set to True and if not it tries with False
Library
Build-Depends: base
Exposed-Modules: Testing.Test1
Extensions: CPP
if flag(debug)
GHC-Options: -DDEBUG
if !os(windows)
CC-Options: "-DDEBUG"
else
CC-Options: "-DNDEBUG"
if flag(webfrontend)
Build-Depends: cgi > 0.42
Other-Modules: Testing.WebStuff
Executable test1
Main-is: T1.hs
Other-Modules: Testing.Test1
Build-Depends: base
if flag(debug)
CC-Options: "-DDEBUG"
GHC-Options: -DDEBUG
Layout
Flags, conditionals, library and executable sections use layout to indicate structure. This is very similar to the Haskell layout rule. Entries in a section have to all be indented to the same level which must be more than the section header. Tabs are not allowed to be used for indentation.
As an alternative to using layout you can also use explicit braces {}
.
In this case the indentation of entries in a section does not matter,
though different fields within a block must be on different lines. Here
is a bit of the above example again, using braces:
Example: Using explicit braces rather than indentation for layout
Name: Test1
Version: 0.0.1
Cabal-Version: >= 1.2
License: BSD3
Author: Jane Doe
Synopsis: Test package to test configurations
Category: Example
Flag Debug {
Description: Enable debug support
Default: False
}
Library {
Build-Depends: base
Exposed-Modules: Testing.Test1
Extensions: CPP
if flag(debug) {
GHC-Options: -DDEBUG
if !os(windows) {
CC-Options: "-DDEBUG"
} else {
CC-Options: "-DNDEBUG"
}
}
}
Configuration Flags
A flag section takes the flag name as an argument and may contain the following fields.
description:
freeform- The description of this flag.
default:
boolean (default:True
)- The default value of this flag.
Note that this value may be overridden in several ways. The rationale for having flags default to True is that users usually want new features as soon as they are available. Flags representing features that are not (yet) recommended for most users (such as experimental features or debugging support) should therefore explicitly override the default to False.
manual:
boolean (default:False
)- By default, Cabal will first try to satisfy dependencies with the default flag value and then, if that is not possible, with the negated value. However, if the flag is manual, then the default value (which can be overridden by commandline flags) will be used.
Conditional Blocks
Conditional blocks may appear anywhere inside a library or executable section. They have to follow rather strict formatting rules. Conditional blocks must always be of the shape
`if `_condition_
_property-descriptions-or-conditionals*_
or
`if `_condition_
_property-descriptions-or-conditionals*_
`else`
_property-descriptions-or-conditionals*_
Note that the if
and the condition have to be all on the same line.
Conditions
Conditions can be formed using boolean tests and the boolean operators
||
(disjunction / logical "or"), &&
(conjunction / logical "and"),
or !
(negation / logical "not"). The unary !
takes highest
precedence, ||
takes lowest. Precedence levels may be overridden
through the use of parentheses. For example, os(darwin) && !arch(i386) || os(freebsd)
is equivalent to (os(darwin) && !(arch(i386))) || os(freebsd)
.
The following tests are currently supported.
os(
name)
- Tests if the current operating system is name. The argument is
tested against
System.Info.os
on the target system. There is unfortunately some disagreement between Haskell implementations about the standard values ofSystem.Info.os
. Cabal canonicalises it so that in particularos(windows)
works on all implementations. If the canonicalised os names match, this test evaluates to true, otherwise false. The match is case-insensitive. arch(
name)
- Tests if the current architecture is name. The argument is
matched against
System.Info.arch
on the target system. If the arch names match, this test evaluates to true, otherwise false. The match is case-insensitive. impl(
compiler)
- Tests for the configured Haskell implementation. An optional version
constraint may be specified (for example
impl(ghc >= 6.6.1)
). If the configured implementation is of the right type and matches the version constraint, then this evaluates to true, otherwise false. The match is case-insensitive. flag(
name)
- Evaluates to the current assignment of the flag of the given name. Flag names are case insensitive. Testing for flags that have not been introduced with a flag section is an error.
true
- Constant value true.
false
- Constant value false.
Resolution of Conditions and Flags
If a package descriptions specifies configuration flags the package user can control these in several ways. If the user does not fix the value of a flag, Cabal will try to find a flag assignment in the following way.
-
For each flag specified, it will assign its default value, evaluate all conditions with this flag assignment, and check if all dependencies can be satisfied. If this check succeeded, the package will be configured with those flag assignments.
-
If dependencies were missing, the last flag (as by the order in which the flags were introduced in the package description) is tried with its alternative value and so on. This continues until either an assignment is found where all dependencies can be satisfied, or all possible flag assignments have been tried.
To put it another way, Cabal does a complete backtracking search to find
a satisfiable package configuration. It is only the dependencies
specified in the build-depends
field in conditional blocks that
determine if a particular flag assignment is satisfiable (build-tools
are not considered). The order of the declaration and the default value
of the flags determines the search order. Flags overridden on the
command line fix the assignment of that flag, so no backtracking will be
tried for that flag.
If no suitable flag assignment could be found, the configuration phase will fail and a list of missing dependencies will be printed. Note that this resolution process is exponential in the worst case (i.e., in the case where dependencies cannot be satisfied). There are some optimizations applied internally, but the overall complexity remains unchanged.
Meaning of field values when using conditionals
During the configuration phase, a flag assignment is chosen, all conditionals are evaluated, and the package description is combined into a flat package descriptions. If the same field both inside a conditional and outside then they are combined using the following rules.
-
Boolean fields are combined using conjunction (logical "and").
-
List fields are combined by appending the inner items to the outer items, for example
Extensions: CPP if impl(ghc) Extensions: MultiParamTypeClasses
when compiled using GHC will be combined to
Extensions: CPP, MultiParamTypeClasses
Similarly, if two conditional sections appear at the same nesting level, properties specified in the latter will come after properties specified in the former.
-
All other fields must not be specified in ambiguous ways. For example
Main-is: Main.hs if flag(useothermain) Main-is: OtherMain.hs
will lead to an error. Instead use
if flag(useothermain) Main-is: OtherMain.hs else Main-is: Main.hs
Source Repositories
It is often useful to be able to specify a source revision control repository for a package. Cabal lets you specifying this information in a relatively structured form which enables other tools to interpret and make effective use of the information. For example the information should be sufficient for an automatic tool to checkout the sources.
Cabal supports specifying different information for various common source control systems. Obviously not all automated tools will support all source control systems.
Cabal supports specifying repositories for different use cases. By declaring which case we mean automated tools can be more useful. There are currently two kinds defined:
-
The
head
kind refers to the latest development branch of the package. This may be used for example to track activity of a project or as an indication to outside developers what sources to get for making new contributions. -
The
this
kind refers to the branch and tag of a repository that contains the sources for this version or release of a package. For most source control systems this involves specifying a tag, id or hash of some form and perhaps a branch. The purpose is to be able to reconstruct the sources corresponding to a particular package version. This might be used to indicate what sources to get if someone needs to fix a bug in an older branch that is no longer an active head branch.
You can specify one kind or the other or both. As an example here are
the repositories for the Cabal library. Note that the this
kind of
repository specifies a tag.
source-repository head
type: darcs
location: http://darcs.haskell.org/cabal/
source-repository this
type: darcs
location: http://darcs.haskell.org/cabal-branches/cabal-1.6/
tag: 1.6.1
The exact fields are as follows:
type:
token- The name of the source control system used for this repository. The
currently recognised types are:
darcs
git
svn
cvs
mercurial
(or aliashg
)bazaar
(or aliasbzr
)arch
monotone
This field is required.
location:
URL- The location of the repository. The exact form of this field depends
on the repository type. For example:
- for darcs:
http://code.haskell.org/foo/
- for git:
git://github.com/foo/bar.git
- for CVS:
anoncvs@cvs.foo.org:/cvs
This field is required.
- for darcs:
module:
token- CVS requires a named module, as each CVS server can host multiple
named repositories.
This field is required for the CVS repository type and should not be used otherwise.
branch:
token- Many source control systems support the notion of a branch, as a
distinct concept from having repositories in separate locations. For
example CVS, SVN and git use branches while for darcs uses different
locations for different branches. If you need to specify a branch to
identify a your repository then specify it in this field.
This field is optional.
tag:
token- A tag identifies a particular state of a source repository. The tag
can be used with a
this
repository kind to identify the state of a repository corresponding to a particular package version or release. The exact form of the tag depends on the repository type.This field is required for the
this
repository kind. subdir:
directory- Some projects put the sources for multiple packages under a single
source repository. This field lets you specify the relative path
from the root of the repository to the top directory for the
package, i.e. the directory containing the package's
.cabal
file.This field is optional. It default to empty which corresponds to the root directory of the repository.
Downloading a package's source
The cabal get
command allows to access a package's source code - either by
unpacking a tarball downloaded from Hackage (the default) or by checking out a
working copy from the package's source repository.
$ cabal get [FLAGS] PACKAGES
The get
command supports the following options:
-d --destdir
PATH- Where to place the package source, defaults to (a subdirectory of) the current directory.
-s --source-repository
[head|this|...]- Fork the package's source repository using the appropriate version control system. The optional argument allows to choose a specific repository kind.
Accessing data files from package code
The placement on the target system of files listed in the data-files
field varies between systems, and in some cases one can even move
packages around after installation (see prefix
independence). To enable
packages to find these files in a portable way, Cabal generates a module
called Paths_
pkgname (with any hyphens in pkgname replaced by
underscores) during building, so that it may be imported by modules of
the package. This module defines a function
getDataFileName :: FilePath -> IO FilePath
If the argument is a filename listed in the data-files
field, the
result is the name of the corresponding file on the system on which the
program is running.
Note: If you decide to import the Paths_
pkgname module then it
must be listed in the other-modules
field just like any other module
in your package.
The Paths_
pkgname module is not platform independent so it does not
get included in the source tarballs generated by sdist
.
Accessing the package version
The aforementioned auto generated Paths_
pkgname module also
exports the constant version ::
Version which is
defined as the version of your package as specified in the version
field.
System-dependent parameters
For some packages, especially those interfacing with C libraries,
implementation details and the build procedure depend on the build
environment. The build-type
Configure
can be used to handle many
such situations. In this case, Setup.hs
should be:
import Distribution.Simple
main = defaultMainWithHooks autoconfUserHooks
Most packages, however, would probably do better using the Simple
build type and configurations.
The build-type
Configure
differs from Simple
in two ways:
-
The package root directory must contain a shell script called
configure
. The configure step will run the script. Thisconfigure
script may be produced by autoconf or may be hand-written. Theconfigure
script typically discovers information about the system and records it for later steps, e.g. by generating system-dependent header files for inclusion in C source files and preprocessed Haskell source files. (Clearly this won't work for Windows without MSYS or Cygwin: other ideas are needed.) -
If the package root directory contains a file called package
.buildinfo
after the configuration step, subsequent steps will read it to obtain additional settings for build information fields,to be merged with the ones given in the.cabal
file. In particular, this file may be generated by theconfigure
script mentioned above, allowing these settings to vary depending on the build environment.The build information file should have the following structure:
buildinfo
executable:
name buildinfoexecutable:
name buildinfo ...where each buildinfo consists of settings of fields listed in the section on build information. The first one (if present) relates to the library, while each of the others relate to the named executable. (The names must match the package description, but you don't have to have entries for all of them.)
Neither of these files is required. If they are absent, this setup
script is equivalent to defaultMain
.
Example: Using autoconf
This example is for people familiar with the autoconf tools.
In the X11 package, the file configure.ac
contains:
AC_INIT([Haskell X11 package], [1.1], [libraries@haskell.org], [X11])
# Safety check: Ensure that we are in the correct source directory.
AC_CONFIG_SRCDIR([X11.cabal])
# Header file to place defines in
AC_CONFIG_HEADERS([include/HsX11Config.h])
# Check for X11 include paths and libraries
AC_PATH_XTRA
AC_TRY_CPP([#include <X11/Xlib.h>],,[no_x=yes])
# Build the package if we found X11 stuff
if test "$no_x" = yes
then BUILD_PACKAGE_BOOL=False
else BUILD_PACKAGE_BOOL=True
fi
AC_SUBST([BUILD_PACKAGE_BOOL])
AC_CONFIG_FILES([X11.buildinfo])
AC_OUTPUT
Then the setup script will run the configure
script, which checks for
the presence of the X11 libraries and substitutes for variables in the
file X11.buildinfo.in
:
buildable: @BUILD_PACKAGE_BOOL@
cc-options: @X_CFLAGS@
ld-options: @X_LIBS@
This generates a file X11.buildinfo
supplying the parameters needed by
later stages:
buildable: True
cc-options: -I/usr/X11R6/include
ld-options: -L/usr/X11R6/lib
The configure
script also generates a header file include/HsX11Config.h
containing C preprocessor defines recording the results of various tests. This
file may be included by C source files and preprocessed Haskell source files in
the package.
Note: Packages using these features will also need to list additional files such
as configure
, templates for .buildinfo
files, files named only in
.buildinfo
files, header files and so on in the extra-source-files
field to
ensure that they are included in source distributions. They should also list
files and directories generated by configure
in the extra-tmp-files
field to
ensure that they are removed by setup clean
.
Quite often the files generated by configure
need to be listed somewhere in
the package description (for example, in the install-includes
field). However,
we usually don't want generated files to be included in the source tarball. The
solution is again provided by the .buildinfo
file. In the above example, the
following line should be added to X11.buildinfo
:
install-includes: HsX11Config.h
In this way, the generated HsX11Config.h
file won't be included in the source
tarball in addition to HsX11Config.h.in
, but it will be copied to the right
location during the install process. Packages that use custom Setup.hs
scripts
can update the necessary fields programmatically instead of using the
.buildinfo
file.
Conditional compilation
Sometimes you want to write code that works with more than one version
of a dependency. You can specify a range of versions for the dependency
in the build-depends
, but how do you then write the code that can use
different versions of the API?
Haskell lets you preprocess your code using the C preprocessor (either
the real C preprocessor, or cpphs
). To enable this, add extensions: CPP
to your package description. When using CPP, Cabal provides some
pre-defined macros to let you test the version of dependent packages;
for example, suppose your package works with either version 3 or version
4 of the base
package, you could select the available version in your
Haskell modules like this:
#if MIN_VERSION_base(4,0,0)
... code that works with base-4 ...
#else
... code that works with base-3 ...
#endif
In general, Cabal supplies a macro MIN_VERSION_
package
_(A,B,C)
for each package depended on via build-depends
. This macro is true if
the actual version of the package in use is greater than or equal to
A.B.C
(using the conventional ordering on version numbers, which is
lexicographic on the sequence, but numeric on each component, so for
example 1.2.0 is greater than 1.0.3).
Since version 1.20, there is also the MIN_TOOL_VERSION_
tool
family of
macros for conditioning on the version of build tools used to build the program
(e.g. hsc2hs
).
Cabal places the definitions of these macros into an automatically-generated header file, which is included when preprocessing Haskell source code by passing options to the C preprocessor.
Cabal also allows to detect when the source code is being used for generating
documentation. The __HADDOCK_VERSION__
macro is defined only when compiling
via haddock instead of a normal Haskell compiler. The value of the
__HADDOCK_VERSION__
macro is defined as A*1000 + B*10 + C
, where A.B.C
is
the Haddock version. This can be useful for working around bugs in Haddock or
generating prettier documentation in some special cases.
More complex packages
For packages that don't fit the simple schemes described above, you have a few options:
-
By using the
build-type
Custom
, you can supply your ownSetup.hs
file, and customize the simple build infrastructure using hooks. These allow you to perform additional actions before and after each command is run, and also to specify additional preprocessors. A typicalSetup.hs
may look like this:import Distribution.Simple main = defaultMainWithHooks simpleUserHooks { postHaddock = posthaddock } posthaddock args flags desc info = ....
See
UserHooks
in Distribution.Simple for the details, but note that this interface is experimental, and likely to change in future releases. -
You could delegate all the work to
make
, though this is unlikely to be very portable. Cabal supports this with thebuild-type
Make
and a trivial setup library Distribution.Make, which simply parses the command line arguments and invokesmake
. HereSetup.hs
should look like this:import Distribution.Make main = defaultMain
The root directory of the package should contain a
configure
script, and, after that has run, aMakefile
with a default target that builds the package, plus targetsinstall
,register
,unregister
,clean
,dist
anddocs
. Some options to commands are passed through as follows:-
The
--with-hc-pkg
,--prefix
,--bindir
,--libdir
,--datadir
,--libexecdir
and--sysconfdir
options to theconfigure
command are passed on to theconfigure
script. In addition the value of the--with-compiler
option is passed in a--with-hc
option and all options specified with--configure-option=
are passed on. -
The
--destdir
option to thecopy
command becomes a setting of adestdir
variable on the invocation ofmake copy
. The suppliedMakefile
should provide acopy
target, which will probably look like this:copy : $(MAKE) install prefix=$(destdir)/$(prefix) \ bindir=$(destdir)/$(bindir) \ libdir=$(destdir)/$(libdir) \ datadir=$(destdir)/$(datadir) \ libexecdir=$(destdir)/$(libexecdir) \ sysconfdir=$(destdir)/$(sysconfdir) \
-
-
Finally, with the
build-type
Custom
, you can also write your own setup script from scratch. It must conform to the interface described in the section on building and installing packages, and you may use the Cabal library for all or part of the work. One option is to copy the source ofDistribution.Simple
, and alter it for your needs. Good luck.