mirror of
https://gitlab.com/ita1024/waf.git
synced 2024-11-10 12:19:13 +01:00
175 lines
7.2 KiB
ReStructuredText
175 lines
7.2 KiB
ReStructuredText
Waf tutorial
|
|
============
|
|
|
|
Waf is a piece of software used to help building software projects.
|
|
The goal of this tutorial is to provide a quick overview of how to set up
|
|
the scripts for a project using Waf.
|
|
|
|
Waf scripts and commands
|
|
------------------------
|
|
|
|
A software typically has *source files* which are kept in a version control system (git, subversion, etc),
|
|
and *build scripts* (Makefiles, ..) which describe what to do with those files. A few *build files* are usually
|
|
obtained after transforming the *source files*, but they are optional. The build scripts in Waf are files named 'wscript'.
|
|
|
|
In general, a project will consist of several phases:
|
|
|
|
* configure: configure the project, find the location of the prerequisites
|
|
* build: transform the source files into build files
|
|
* install: install the build files
|
|
* uninstall: uninstall the build files
|
|
* dist: create an archive of the source files
|
|
* clean: remove the build files
|
|
|
|
Each phase is modelled in the wscript file as a python function which takes as argument an instance of :py:class:`waflib.Context.Context`.
|
|
Let's start with a new wscript file in the directory '/tmp/myproject'::
|
|
|
|
def configure(conf):
|
|
print("configure!")
|
|
|
|
def build(bld):
|
|
print("build!")
|
|
|
|
We will also use a Waf binary file, for example waf-2.0.0, which we will copy in the project directory::
|
|
|
|
$ cd /tmp/myproject
|
|
$ wget https://waf.io/waf-2.0.0
|
|
|
|
To execute the project, we will simply call the command as an argument to ``waf``::
|
|
|
|
$ python ./waf-2.0.0 configure build
|
|
configure!
|
|
build!
|
|
|
|
Targets
|
|
-------
|
|
|
|
An important part of the build system is to declare the creation of targets. Here is a very simple example::
|
|
|
|
def build(bld):
|
|
tg = bld(rule='cp ${SRC} ${TGT}', source='wscript', target='foo.txt')
|
|
bld(rule='cp ${SRC} ${TGT}', source='foo.txt', target='bar.txt')
|
|
|
|
The call ``bld(..)`` creates an object called *task generator*, which is used to create *tasks* which will actually
|
|
call the command ``cp``. The commands are not executed unless all the scripts have been read, which is important
|
|
for computing the build order.
|
|
|
|
The expressions *${SRC}* and *${TGT}* are shortcuts to avoid repeating the file names. More shortcuts can be defined
|
|
by using the *${}* symbol, which reads the values from the attribute bld.env::
|
|
|
|
def build(bld):
|
|
bld.env.MESSAGE = 'Hello, world!'
|
|
bld(rule='echo ${MESSAGE}', always=True)
|
|
|
|
The bld object is an instance of :py:class:`waflib.Build.BuildContext`, its *env* attribute is an instance :py:class:`waflib.ConfigSet.ConfigSet`.
|
|
The values are set in this object to be shared/stored/loaded easily. Here is how to do the same thing by sharing data between the configuration and build::
|
|
|
|
def configure(cnf):
|
|
cnf.env.MESSAGE = 'Hello, world!'
|
|
|
|
def build(bld):
|
|
bld(rule='echo ${MESSAGE}', always=True)
|
|
|
|
Scripts and Tools
|
|
-----------------
|
|
|
|
To let a script use a script from a subdirectory, the method :py:meth:`waflib.Context.Context.recurse` has to be used with
|
|
the relative path to the folder containing the wscript file. For example, to call the function *build* in the script ``src/wscript``,
|
|
one should write::
|
|
|
|
def build(bld):
|
|
bld.recurse('src')
|
|
|
|
The support for specific languages and compilers is provided through specific modules called *Waf tools*. The tools are
|
|
similar to wscript files and provide functions such as *configure* or *build*. Here is a simple project for the C programming language::
|
|
|
|
def options(opt):
|
|
opt.load('compiler_c')
|
|
def configure(cnf):
|
|
cnf.load('compiler_c')
|
|
def build(bld):
|
|
bld(features='c cprogram', source='main.c', target='app')
|
|
|
|
The function *options* is another predefined command used for setting command-line options. Its argument is an instance of :py:meth:`waflib.Options.OptionsContext`. The tool *compiler_c* is provided for detecting if a C compiler is present and to set various variables such as ``cnf.env.CFLAGS``.
|
|
|
|
The task generator declared in *bld* does not have a *rule* keyword, but a list of *features* which is used to reference methods that will call the appropriate rules. In this case, a rule is called for compiling the file, and another is used for linking the object files into the binary *app*. Other tool-dependent features exist such as *javac*, *cs*, or *tex*.
|
|
|
|
A C and C++ project
|
|
-------------------
|
|
|
|
Here is a script for a more complicated project::
|
|
|
|
def options(opt):
|
|
opt.load('compiler_c compiler_cxx')
|
|
def configure(cnf):
|
|
cnf.load('compiler_c compiler_cxx')
|
|
cnf.check(features='cxx cxxprogram', lib=['m'], cflags=['-Wall'], defines=['var=foo'], uselib_store='M')
|
|
def build(bld):
|
|
bld(features='c cshlib', source='b.c', target='mylib')
|
|
bld(features='c cxx cxxprogram', source='a.c main.cpp', target='app', use=['M','mylib'], lib=['dl'])
|
|
|
|
The method :py:func:`waflib.Tools.c_config.check` executes a build internally to check if the library ``libm`` is present on the operating system.
|
|
It will then define variables such as:
|
|
|
|
* ``conf.env.LIB_M = ['m']``
|
|
* ``conf.env.CFLAGS_M = ['-Wall']``
|
|
* ``conf.env.DEFINES_M = ['var=foo']``
|
|
|
|
By stating ``use=['M', 'mylib']``, the program *app* is going to inherit all the *M* variables defined
|
|
during the configuration. The program will also use the library *mylib* and both the build order and the dependencies
|
|
will be modified so that *mylib* is linked before *app*.
|
|
|
|
The ``use`` attribute is also working for other languages such as Java (dependencies between jar files) or C# (dependencies between assemblies).
|
|
|
|
Project-specific extensions
|
|
---------------------------
|
|
|
|
The *feature* keyword is a high-level reference to existing Waf methods.
|
|
For example, the **c** feature will add the method :py:func:`waflib.Tools.ccroot.apply_incpaths` for execution.
|
|
To add a new method that will add the task generator path to the include path for all C targets,
|
|
one may use such a declaration::
|
|
|
|
from waflib import Utils
|
|
from waflib.TaskGen import feature, before_method
|
|
@feature('c')
|
|
@before_method('apply_incpaths')
|
|
def add_current_dir_to_includes(self):
|
|
self.includes = Utils.to_list(self.includes)
|
|
self.includes.append(self.path)
|
|
|
|
def build(bld):
|
|
tg = bld(features='c', source='main.c', target='app')
|
|
|
|
The *feature* methods are bound to the :py:class:`waflib.TaskGen.task_gen` class, which is the class of the
|
|
object *tg* in the example. New features can be declared in the same manner::
|
|
|
|
from waflib.TaskGen import feature, after_method
|
|
@feature('debug_tasks')
|
|
@after_method('apply_link')
|
|
def print_debug(self):
|
|
print('tasks created %r' % self.tasks)
|
|
|
|
def build(bld):
|
|
tg = bld(features='c cprogram debug_tasks', source='main.c', target='app')
|
|
|
|
The declaration can be made more user-friendly by binding new methods to the context classes::
|
|
|
|
from waflib.Build import BuildContext
|
|
def enterprise_program(self, *k, **kw):
|
|
kw['features'] = 'c cprogram debug_tasks'
|
|
return self(*k, **kw)
|
|
BuildContext.enterprise_program = enterprise_program
|
|
|
|
def build(bld):
|
|
# no feature line
|
|
bld.enterprise_program(source='main.c', target='app')
|
|
|
|
The support code may be turned into a Waf tool by moving it to a separate file.
|
|
To ease the deployment, the new Waf tool can even be added to the waf file (see https://gitlab.com/ita1024/waf/blob/master/README.md#L20).
|
|
|
|
Conclusion
|
|
----------
|
|
|
|
This concludes the tutorial. For more information consult the apis, the Waf book and the examples.
|
|
|