Boost C++ Libraries

...one of the most highly regarded and expertly designed C++ library projects in the world. Herb Sutter and Andrei Alexandrescu, C++ Coding Standards

This is an old version of boost. Click here for the latest version's documentation home page.
PrevUpHomeNext

Overview

Boost.Jam Language
Configuration
Invocation
Declaring Targets
Projects
The Build Process

This section will provide the information necessary to create your own projects using Boost.Build. The information provided here is relatively high-level, and the section called “Detailed reference” as well as the on-line help system must be used to obtain low-level documentation (see --help).

Boost.Build actually consists of two parts - Boost.Jam, a build engine with its own interpreted language, and Boost.Build itself, implemented in Boost.Jam's language. The chain of events when you type bjam on the command line is:

  1. Boost.Jam tries to find Boost.Build and loads the top-level module. The exact process is described in the section called “Initialization”

  2. The top-level module loads user-defined configuration files, user-config.jam and site-config.jam, which define available toolsets.

  3. The Jamfile in the current directory is read. That in turn might cause reading of further Jamfiles. As a result, a tree of projects is created, with targets inside projects.

  4. Finally, using the build request specified on the command line, Boost.Build decides which targets should be built, and how. That information is passed back to Boost.Jam, which takes care of actually running commands.

So, to be able to successfully use Boost.Build, you need to know only four things:

Boost.Jam Language

This section will describe the basics of the Boost.Jam language—just enough for writing Jamfiles. For more information, please see the Boost.Jam documentation.

Boost.Jam has an interpreted, procedural language. On the lowest level, a Boost.Jam program consists of variables and rules (the Jam term for function). They are grouped in modules—there's one global module and a number of named modules. Besides that, a Boost.Jam program contains classes and class instances.

Syntantically, a Boost.Jam program consists of two kind of elements—keywords (which have a special meaning to Boost.Jam) and literals. Consider this code:

a = b ;

which assigns the value b to the variable a. Here, = and ; are keywords, while a and b are literals.

[Warning] Warning

All syntax elements, even keywords, must be separated by spaces. For example, omitting the space character before ; will lead to a syntax error.

If you want to use a literal value that is the same as some keyword, the value can be quoted:

a = "=" ;

All variables in Boost.Jam have the same type—list of strings. To define a variable one assigns a value to it, like in the previous example. An undefined variable is the same as a variable with an empty value. Variables can be accessed with the $(variable) syntax. For example:

a = $(b) $(c) ;

Rules are defined by specifying the rule name, the parameter names, and the allowed size of the list value for each parameter.

rule example 
     (
         parameter1 : 
         parameter2 ? : 
         parameter3 + :
         parameter4 * 
     )
     {
        // body
     }

When this rule is called, the list passed as the first argument must have exactly one value. The list passed as the second argument can either have one value of be empty. The two remaining arguments can be arbitrary long, but the third argument may not be empty.

The overview of Boost.Jam language statements is given below:

helper 1 : 2 : 3 ; 
x = [ helper 1 : 2 : 3 ] ;

This code calls the named rule with the specified arguments. When the result of the call must be used inside some expression, you need to add brackets around the call, like shown on the second line.

if cond { statements } [ else { statements } ]

This is a regular if-statement. The condition is composed of:

  • Literals (true if at least one string is not empty)

  • Comparisons: a operator b where operator is one of =, !=, <, >, <=, >=. The comparison is done pairwise between each string in the left and the right arguments.

  • Logical operations: ! a, a && b, a || b

  • Grouping: ( cond )

for var in list { statements }

Executes statements for each element in list, setting the variable var to the element value.

while cond { statements }

Repeatedly execute statements while cond remains true upon entry.

return values ;
      

This statement should be used only inside a rule and assigns values to the return value of the rule.

[Warning] Warning

The return statement does not exit the rule. For example:

rule test ( )
{
   if 1 = 1 {
      return "reasonable" ;
   }
   return "strange" ;
}

will return strange, not reasonable.

import module ;
import module : rule ;

The first form imports the specified bjam module. All rules from that module are made available using the qualified name: module.rule. The second form imports the specified rules only, and they can be called using unqualified names.

Sometimes, you'd need to specify the actual command lines to be used when creating targets. In jam language, you use named actions to do this. For example:

actions create-file-from-another
{
    create-file-from-another $(<) $(>)
}

This specifies a named action called create-file-from-another. The text inside braces is the command to invoke. The $(<) variable will be expanded to list of generated files, and the $(>) variable will be expanded to the list of source files.

To flexibly adjust command line, you can define a rule with the same name as the action, and taking three parameters -- targets, sources and properties. For example:

rule create-file-from-another ( targets * : sources * : properties * )
{
   if <variant>debug in $(properties)
   {
       OPTIONS on $(targets) = --debug ;
   }
}
actions create-file-from-another
{
    create-file-from-another $(OPTIONS) $(<) $(>)
}

In this example, the rule checks if certain build property is specified. If so, it sets variable OPIONS that's used inside action. Note that the variable is set "on targets" -- the value will be only visible inside action, not globally. Were it set globally, using variable named OPTIONS in two unrelated actions would be impossible.

More details can be found in Jam reference, the section called “ Rules”

Configuration

The Boost.Build configuration is specified in the file user-config.jam. You can edit the one in the top-level directory of Boost.Build installation or create a copy in your home directory and edit that. (See Table 25.2, “Search paths for configuration files” for the exact search paths.) The primary function of that file is to declare which compilers and other tools are available. The simplest syntax to configure a tool is:

using tool-name ;        

The using rule is given a name of tool, and will make that tool available to Boost.Build. For example, using gcc ; will make the gcc compiler available.

Since nothing but a tool name is specified, Boost.Build will pick some default settings. For example, it will use the gcc executable found in the PATH, or look in some known installation locations. In most cases, this strategy works automatically. In case you have several versions of a compiler, it's installed in some unusual location, or you need to tweak its configuration, you'll need to pass additional parameters to the using rule. The parameters to using can be different for each tool. You can obtain specific documentation for any tool's configuration parameters by invoking

bjam --help tool-name.init         

That said, for all the compiler toolsets Boost.Build supports out-of-the-box, the list of parameters to using is the same: toolset-name, version, invocation-command, and options.

The version parameter identifies the toolset version, in case you have several installed. It can have any form you like, but it's recommended that you use a numeric identifier like 7.1.

The invocation-command parameter is the command that must be executed to run the compiler. This parameter can usually be omitted if the compiler executable

  • has its “usual name” and is in the PATH, or

  • was installed in a standard “installation directory”, or

  • can be found through a global mechanism like the Windows registry.

For example:

using msvc : 7.1 ;
using gcc ;

If the compiler can be found in the PATH but only by a nonstandard name, you can just supply that name:

using gcc : : g++-3.2 ;

Otherwise, it might be necessary to supply the complete path to the compiler executable:

using msvc : : "Z:/Programs/Microsoft Visual Studio/vc98/bin/cl" ;

Some Boost.Build toolsets will use that path to take additional actions required before invoking the compiler, such as calling vendor-supplied scripts to set up its required environment variables. When compiler executables for C and C++ are different, path to the C++ compiler executable must be specified. The “invocation command” can be any command allowed by the operating system. For example:

using msvc : : echo Compiling && foo/bar/baz/cl ;

will work.

To configure several versions of a toolset, simply invoke the using rule multiple times:

using gcc : 3.3 ;
using gcc : 3.4 : g++-3.4 ;
using gcc : 3.2 : g++-3.2 ;

Note that in the first call to using, the compiler found in the PATH will be used, and there's no need to explicitly specify the command.

As shown above, both the version and invocation-command parameters are optional, but there's an important restriction: if you configure the same toolset more than once, you must pass the version parameter every time. For example, the following is not allowed:

using gcc ;
using gcc : 3.4 : g++-3.4 ;

because the first using call does not specify a version.

The options parameter is used to fine-tune the configuration. All of Boost.Build's standard compiler toolsets accept properties of the four builtin features cflags, cxxflags, compileflags and linkflags as options specifying flags that will be always passed to the corresponding tools. Values of the cflags feature are passed directly to the C compiler, values of the cxxflags feature are passed directly to the C++ compiler, and values of the compileflags feature are passed to both. For example, to configure a gcc toolset so that it always generates 64-bit code you could write:

using gcc : 3.4 : : <compileflags>-m64 <linkflags>-m64 ;

Invocation

This section describes how invoke Boost.Build from the command line

To build all targets defined in Jamfile in the current directory with default properties, run:

bjam

To build specific targets, specify them on the command line:

bjam lib1 subproject//lib2 

To request a certain value for some property, add property=value to the command line:

bjam toolset=gcc variant=debug optimization=space

For often used features, like toolset and variant you can omit the feature name, so the above can be written as:

bjam optimization=space

Boost.Build recognizes the following command line options.

--clean

Cleans all targets in the current directory and in any subprojects. Note that unlike the clean target in make, you can use --clean together with target names to clean specific targets.

--clean-all

Cleans all targets, no matter where they are defined. In particular, it will clean targets in parent Jamfiles, and targets defined under other project roots.

--build-dir

Changes build directories for all project roots being built. When this option is specified, all Jamroot files should declare project name. The build directory for the project root will be computed by concatanating the value of the --build-dir option, the project name specified in Jamroot, and the build dir specified in Jamroot (or bin, if none is specified).

The option is primarily useful when building from read-only media, when you can't modify Jamroot.

--version

Prints information on Boost.Build and Boost.Jam versions.

--help

Invokes the online help system. This prints general information on how to use the help system with additional --help* options.

--debug-configuration

Produces debug information about loading of Boost.Build and toolset files.

--debug-building

Prints what targets are being built and with what properties.

--debug-generators

Produces debug output from generator search process. Useful for debugging custom generators.

--ignore-config

Do not load site-config.jam and user-config.jam configuration files.

--debug

Enables internal checks.

For complete specification of command line syntax, see the section called “Command line arguments”

Declaring Targets

A Main target is a user-defined named entity that can be built, for example an executable file. Declaring a main target is usually done using one of the main target rules described in the section called “Builtin rules”. The user can also declare custom main target rules as shown in the section called “Main target rules”.

Most main target rules in Boost.Build have the same common signature:

rule rule-name (
     main-target-name :
     sources + :
     requirements * :
     default-build * :
     usage-requirements * )
  • main-target-name is the name used to request the target on command line and to use it from other main targets. A main target name may contain alphanumeric characters, dashes (‘-’), and underscores (‘_’).
  • sources is the list of source files and other main targets that must be combined.
  • requirements is the list of properties that must always be present when this main target is built.
  • default-build is the list of properties that will be used unless some other value of the same feature is already specified, e.g. on the command line or by propagation from a dependent target.
  • usage-requirements is the list of properties that will be propagated to all main targets that use this one, i.e. to all its dependents.

Some main target rules have a different list of parameters as explicitly stated in their documentation.

The actual requirements for a target are obtained by refining requirements of the project where a target is declared with the explicitly specified requirements. The same is true for usage-requirements. More details can be found in the section called “Property refinement”

Name

The name of main target has two purposes. First, it's used to refer to this target from other targets and from command line. Second, it's used to compute the names of the generated files. Typically, filenames are obtained from main target name by appending system-dependent suffixes and prefixes.

The name of a main target can contain alphanumeric characters, dashes, undescores and dots. The entire name is significant when resolving references from other targets. For determining filenames, only the part before the first dot is taken. For example:

obj test.release : test.cpp : <variant>release ;
obj test.debug : test.cpp : <variant>debug ;

will generate two files named test.obj (in two different directories), not two files named test.release.obj and test.debug.obj.

Sources

The list of sources specifies what should be processed to get the resulting targets. Most of the time, it's just a list of files. Sometimes, you'll want to automatically construct the list of source files rather than having to spell it out manually, in which case you can use the glob rule. Here are two examples:

exe a : a.cpp ;           # a.cpp is the only source file
exe b : [ glob *.cpp ] ;  # all .cpp files in this directory are sources 

Unless you specify a file with an absolute path, the name is considered relative to the source directory — which is typically the directory where the Jamfile is located, but can be changed as described in the section called “Projects”.

The list of sources can also refer to other main targets. Targets in the same project can be referred to by name, while targets in other projects must be qualified with a directory or a symbolic project name. The directory/project name is separated from the target name by a double forward slash. There's no special syntax to distinguish the directory name from the project name—the part before the double slash is first looked up as project name, and then as directory name. For example:

lib helper : helper.cpp ;
exe a : a.cpp helper ;
# Since all project ids start with slash, ".." is directory name.
exe b : b.cpp ..//utils ;
exe c : c.cpp /boost/program_options//program_options ;

The first exe uses the library defined in the same project. The second one uses some target (most likely library) defined by Jamfile one level higher. Finally, the third target uses some C++ Boost library, referring to it by absolute symbolic name. More information about target references can be found in the section called “Dependent Targets” and the section called “Target identifiers and references”.

Requirements

Requirements are the properties that should always be present when building a target. Typically, they are includes and defines:

exe hello : hello.cpp : <include>/opt/boost <define>MY_DEBUG ;

There is a number of other features, listed in the section called “Builtin features”. For example if a library can only be built statically, or a file can't be compiled with optimization due to a compiler bug, one can use

lib util : util.cpp : <link>static ;
obj main : main.cpp : <optimization>off ;

Sometimes, particular relationships need to be maintained among a target's build properties. This can be achieved with conditional requirements. For example, you might want to set specific #defines when a library is built as shared, or when a target's release variant is built in release mode.

lib network : network.cpp 
    : <link>shared:<define>NEWORK_LIB_SHARED
     <variant>release:<define>EXTRA_FAST
    ;

In the example above, whenever network is built with <link>shared, <define>NEWORK_LIB_SHARED will be in its properties, too.

You can use several properties in the condition, for example:

lib network : network.cpp
    : <toolset>gcc,<optimization>speed:<define>USE_INLINE_ASSEMBLER
    ;

A more powerful variant of conditional requirements is indirect conditional requirements. You can provide a rule that will be called with the current build properties and can compute additional properties to be added. For example:

lib network : network.cpp 
    : <conditional>@my-rule
    ;
rule my-rule ( properties * )
{
    local result ;
    if <toolset>gcc <optimization>speed in $(properties)
    {
        result += <define>USE_INLINE_ASSEMBLER ;
    }
    return $(result) ;
}  

This example is equivalent to the previous one, but for complex cases, indirect conditional requirements can be easier to write and understand.

Requirements explicitly specified for a target are usually combined with the requirements specified for the containing project. You can cause a target to completely ignore specific project's requirement using the syntax by adding a minus sign before a property, for example:

exe main : main.cpp : -<define>UNNECESSARY_DEFINE ;

This syntax is the only way to ignore free properties from a parent, such as defines. It can be also useful for ordinary properties. Consider this example:

project test : requirements <threading;>multi ;
exe test1 : test1.cpp ;
exe test2 : test2.cpp : <threading;>single ;
exe test3 : test3.cpp : -<threading;>multi ;

Here, test1 inherits project requirements and will always be built in multi-threaded mode. The test2 target overrides project's requirements and will always be built in single-threaded mode. In contrast, the test3 target removes a property from project requirements and will be built either in single-threaded or multi-threaded mode depending on which variant is requested by the user.

Note that the removal of requirements is completely textual: you need to specify exactly the same property to remove it.

Default Build

The default-build parameter is a set of properties to be used if the build request does not otherwise specify a value for features in the set. For example:

exe hello : hello.cpp : : <threading>multi ;

would build a multi-threaded target unless the user explicitly requests a single-threaded version. The difference between requirements and default-build is that requirements cannot be overridden in any way.

Additional Information

The ways a target is built can be so different that describing them using conditional requirements would be hard. For example, imagine that a library actually uses different source files depending on the toolset used to build it. We can express this situation using target alternatives:

lib demangler : dummy_demangler.cpp ;                      # alternative 1
lib demangler : demangler_gcc.cpp : <toolset>gcc ;   # alternative 2
lib demangler : demangler_msvc.cpp : <toolset>msvc ; # alternative 3

In the example above, when built with gcc or msvc, demangler will use a source file specific to the toolset. Otherwise, it will use a generic source file, dummy_demangler.cpp.

It is possible to declare a target inline, i.e. the "sources" parameter may include calls to other main rules. For example:

exe hello : hello.cpp 
    [ obj helpers : helpers.cpp : <optimization>off ] ;

Will cause "helpers.cpp" to be always compiled without optimization. When referring to an inline main target, its declared name must be prefixed by its parent target's name and two dots. In the example above, to build only helpers, one should run bjam hello..helpers.

When no target is requested on the command line, all targets in the current project will be built. If a target should be built only by explicit request, this can be expressed by the explicit rule:

explicit install_programs ;

Projects

As mentioned before, targets are grouped into projects, and each Jamfile is a separate project. Projects are useful because they allow us to group related targets together, define properties common to all those targets, and assign a symbolic name to the project that can be used in referring to its targets.

Projects are named using the project rule, which has the following syntax:

project id : attributes ;

Here, attributes is a sequence of rule arguments, each of which begins with an attribute-name and is followed by any number of build properties. The list of attribute names along with its handling is also shown in the table below. For example, it is possible to write:

project tennis 
    : requirements <threading>multi 
    : default-build release
    ;

The possible attributes are listed below.

Project id is a short way to denote a project, as opposed to the Jamfile's pathname. It is a hierarchical path, unrelated to filesystem, such as "boost/thread". Target references make use of project ids to specify a target.

Source location specifies the directory where sources for the project are located.

Project requirements are requirements that apply to all the targets in the projects as well as all subprojects.

Default build is the build request that should be used when no build request is specified explicitly.

The default values for those attributes are given in the table below.

Table 25.1. 

Attribute Name Default value Handling by the project rule
Project id none none Assigned from the first parameter of the 'project' rule. It is assumed to denote absolute project id.
Source location source-location The location of jamfile for the project Sets to the passed value
Requirements requirements The parent's requirements The parent's requirements are refined with the passed requirement and the result is used as the project requirements.
Default build default-build none Sets to the passed value
Build directory build-dir Empty if the parent has no build directory set. Otherwise, the parent's build directory with the relative path from parent to the current project appended to it. Sets to the passed value, interpreted as relative to the project's location.

Besides defining projects and main targets, Jamfiles often invoke various utility rules. For the full list of rules that can be directly used in Jamfile see the section called “Builtin rules”.

Each subproject inherits attributes, constants and rules from its parent project, which is defined by the nearest Jamfile in an ancestor directory above the subproject. The top-level project is declared in a file called Jamroot rather than Jamfile. When loading a project, Boost.Build looks for either Jamroot or Jamfile. They are handled identically, except that if the file is called Jamroot, the search for a parent project is not performed.

Even when building in a subproject directory, parent project files are always loaded before those of their subprojects, so that every definition made in a parent project is always available to its children. The loading order of any other projects is unspecified. Even if one project refers to another via the use-project or a target reference, no specific order should be assumed.

[Note] Note

Giving the root project the special name “Jamroot” ensures that Boost.Build won't misinterpret a directory above it as the project root just because the directory contains a Jamfile.

The Build Process

When you've described your targets, you want Boost.Build to run the right tools and create the needed targets. This section will describe two things: how you specify what to build, and how the main targets are actually constructed.

The most important thing to note is that in Boost.Build, unlike other build tools, the targets you declare do not correspond to specific files. What you declare in a Jamfile is more like a “metatarget.” Depending on the properties you specify on the command line, each metatarget will produce a set of real targets corresponding to the requested properties. It is quite possible that the same metatarget is built several times with different properties, producing different files.

[Tip] Tip

This means that for Boost.Build, you cannot directly obtain a build variant from a Jamfile. There could be several variants requested by the user, and each target can be built with different properties.

Build Request

The command line specifies which targets to build and with which properties. For example:

bjam app1 lib1//lib1 toolset=gcc variant=debug optimization=full

would build two targets, "app1" and "lib1//lib1" with the specified properties. You can refer to any targets, using target id and specify arbitrary properties. Some of the properties are very common, and for them the name of the property can be omitted. For example, the above can be written as:

bjam app1 lib1//lib1 gcc debug optimization=full

The complete syntax, which has some additional shortcuts, is described in the section called “Command line”.

Building a main target

When you request, directly or indirectly, a build of a main target with specific requirements, the following steps are done. Some brief explanation is provided, and more details are given in the section called “Build process”.

  1. Applying default build. If the default-build property of a target specifies a value of a feature that is not present in the build request, that value is added.

  2. Selecting the main target alternative to use. For each alternative we look how many properties are present both in alternative's requirements, and in build request. The alternative with large number of matching properties is selected.

  3. Determining "common" properties. The build request is refined with target's requirements. The conditional properties in requirements are handled as well. Finally, default values of features are added.

  4. Building targets referred by the sources list and dependency properties. The list of sources and the properties can refer to other target using target references. For each reference, we take all propagated properties, refine them by explicit properties specified in the target reference, and pass the resulting properties as build request to the other target.

  5. Adding the usage requirements produced when building dependencies to the "common" properties. When dependencies are built in the previous step, they return both the set of created "real" targets, and usage requirements. The usage requirements are added to the common properties and the resulting property set will be used for building the current target.

  6. Building the target using generators. To convert the sources to the desired type, Boost.Build uses "generators" --- objects that correspond to tools like compilers and linkers. Each generator declares what type of targets it can produce and what type of sources it requires. Using this information, Boost.Build determines which generators must be run to produce a specific target from specific sources. When generators are run, they return the "real" targets.

  7. Computing the usage requirements to be returned. The conditional properties in usage requirements are expanded and the result is returned.

Building a Project

Often, a user builds a complete project, not just one main target. In fact, invoking bjam without arguments builds the project defined in the current directory.

When a project is built, the build request is passed without modification to all main targets in that project. It's is possible to prevent implicit building of a target in a project with the explicit rule:

explicit hello_test ;

would cause the hello_test target to be built only if explicitly requested by the user or by some other target.

The Jamfile for a project can include a number of build-project rule calls that specify additional projects to be built.


PrevUpHomeNext