Scroll to navigation

MYMAKE(1) General Commands Manual MYMAKE(1)

NAME

mymake - simple build system for C/C++

SYNOPSIS

mymake [-?] [-f] [-c] [-t] [-e] [-ne] [-p path] [-j processes] [-o output] [-d level] [--default-input input] [files]... [options]... [-- args...]
mymake --config
mymake --target
mymake --project

DESCRIPTION

The mymake program builds C/C++ programs. It aims to be able to build simple programs (referred to as targets) with close to zero configuration, yet be powerful enough to handle larger programs as the initially simple project grows.

Once installed, run

mymake --config

to generate the global configuration file for the current user. This file specifies how to compile C/C++ programs on the current system, and defines the default parameters that can be used in other files as described below. The default contents are typically sufficient.

After the initial configuration, a simple program can be compiled by navigating to the directory the source code of the program and typing

mymake myprogram.cpp

This will compile and link the program using the configuration in the global configuration, and if compilation is successful also execute the program. Command-line parameters may be passed to the program using the -- or the -a parameter

mymake myprogram.cpp -- param1 param2...

The default behavior also works with programs consisting of multiple files. mymake examines which files are included in the source file specified on the command-line and compiles those files as well. This is based on the assumption that if the file utils.h is included, then the file utils.cpp should also be compiled (if it exists). Of course, different extensions than .cpp are also examined. As is typically the case with other build systems, only files changed since the last build are rebuilt.

If the program being compiled does not follow the assumption that header files and implementation files appear pairwise, it is possible to create a file called .mymake in the project directory and add the following to it

[]
input=*

This instructs mymake to include all implementation files in the compilation, regardless of whether or not they are considered to be needed by the program or not. It is also possible to add a particular file to the input statement (e.g. myprogram.cpp). In both cases, it will no longer be necessary to specify a file on the command line.

As of version 2.3.0, mymake tracks the command-lines used to compile each file, and re-compiles source files where the command line has changed. This means that mymake is able to re-compile the relevant parts of the project if the configuration file has changed.

As the project grows, it is also possible to divide the project into separate sub-projects and have mymake manage dependencies automatically. This is done by creating a .myproject file in the root directory of the project. This tells mymake to treat all directories as sub-projects, each which may contain their own .mymake file. See CONFIGURATION for details.

OPTIONS

The usual GNU command line syntax is respected, with long options starting with two dashes (`-'). Long options are equivalent to short options.

-?, --help
Print a message outlining basic usage and command-line options.
Force re-compilation of all required files, even files that would otherwise be considered to be up-to-date. This is useful if the configuration files were changed, for example.
Clean build files by removing the execDir and buildDir indicated by the configuration based on any options specified.
Output information about the wall-clock time used in each step of compilation.
Execute the executable after successfully building a target. The default configuration executes the program, but this behavior may be overridden by project-specific configuration files. The command-line parameters overrides all configuration files.
Do not execute the executable after a successful build. The inverse of -f.
Specify the working directory when running the compiled executable. The default value is the path of the project to which the executable belongs.
Spawn up to the specified number of processes in parallel during compilation. This typically means that compilation may use up to the specified number of hardware threads. This is typically not needed, as the user's preference is specified in the global configuration file.
Specify the name of the output file (note: not the location). This can be used to override the default behavior, but this is better done in one of the configuration files.
Specify the debug level, either as a number or as the associated label. The following levels are available:
0: QUIET
No output except for fatal errors.
1: NORMAL
The default value. Outputs progress information.
2: PEDANTIC
Warns about configuration issues that might be errors. Useful as a first step in debugging configurations.
3: INFO
Prints information about decisions during the build. Good for debugging configurations.
4: VERBOSE
All information you will typically need when debugging configuration issues.
5: DEBUG
Information typically only needed when debugging mymake itself.
Add input as an input file if no other inputs were specified either on the command line or in any configuration files. This option is intended to be used when integrating mymake in an editor. The editor may then always supply the name of the current open file as a --default-input to make mymake do the right thing based on configurations: compiling the current file if nothing else is specified, otherwise follow the other, more precise instructions in the configuration files.
[options]
Zero or more options may also be specified on the command line. These options may correspond to sections in one of the configuration files to enable (or disable). See the CONFIGURATION section below for details.
[files]
Files may also be specified on the command line. Mymake will add all names that correspond to existing files (possibly first appending known file extensions) to the list of input files to process.
Create the global configuration in ~/.mymake. This can also be used to revert the global configuration to the default state. Will ask for the preferred number of threads to use during compilation.
Creates the file .mymake in the current directory. It fills the file with a template configuration that contains common options and settings.
Creates the file .myproject in the current directory. The file contains a template suitable for a project, i.e. a directory consisting of multiple targets.

CONFIGURATION

The configuration in mymake consists of a set of named variables, each of which contain an array of strings. The value of these variables may originate from one (or more) of four locations: the global configuration (~/.mymake), the project configuration (.myproject, if present), the target configuration (.mymake, if present), or the command-line. The project and target configurations may reside in the current directory, or any parent directories. If a project configuration is found, target configurations are expected to be located in sub-directories to the project configuration. Options are always applied in the order specified above. This means that options in the global configuration may be overridden by other configurations, and that the command-line parameters are always applied last. The exception is when compiling a project. Then, command line parameters are only applied when resolving the project configuration, not when resolving the configuration for individual targets.

All configuration files follow the same format. They all consist of a sequence of assignments to variables, optionally grouped into zero or more sections. Lines starting with `#' are comments. Assignments have one of the following two forms:

name=value
name+=value

The first form replaces the entire contents of the variable with the string value, while the second adds value as the last element of the array. As a special case, if the first form is used, and value is empty, the variable is assigned the empty array. Assignments may be grouped into sections by adding section headers before them. A section header has the following form:

[option1,option2,...]

The meaning of the section header is: only consider the following assignments if all options that appear in the header are present in the context in which the configuration is evaluated (e.g. specified on the command line). Thus, the header [] is always evaluated, [release] is only evaluated if the release option has been specified, and [release,unix] only if both release and unix have been specified. Additionally, an option may be prefixed with an `!' to mean that the particular option has to be absent, for example: [!release,unix].

Each configuration file is then evaluated by mymake in turn to provide the set of variables to use during compilation. This is done by evaluating each assignment in each file in the order they are specified, ignoring any sections that should be skipped according to the available options. Typically, each file is only evaluated once, with a context consisting of the options specified on the command-line. The exception is .myproject-files, which are covered in the PROJECTS section below.

PRE-DEFINED OPTIONS

The following options are pre-defined by mymake or the default configuration, and can be used by default:

Produce a release version of the program. This typically means turning on more aggressive optimizations.
Produces a static library. Typically used in projects when some targets are static libraries used by other targets in the project.
Produces a dynamic library. Typically used in projects, like lib.
Defined by mymake when compiling for a UNIX-like system.
Defined by mymake when compiling for a Windows-like system.
Defined automatically when evaluating the .myproject file in the project context.
Defined automatically when evaluating the .myproject file to find options for the targets in a project.
Defined automatically when evaluating the .myproject file to find explicit dependencies between projects.

VARIABLES

The following variables are used by mymake to define what should be done. Some of these variables are treated specially by mymake itself, others are just defined by the global configuration. It is possible to define and use other variables in configuration files.

Array of the file extensions you want to compile. Whenever mymake realizes you have included x.h, looks for all extensions in ext and tries them to find the corresponding implementation file.
File extension of executable files. Added to the output filename automatically.
File extension of intermediate files. Typically .o on UNIX systems.
String containing the directory used to store all temporary files when building your program. Relative to the root directory of the target (i.e. where the .mymake file is).
String containing the directory used to store the final output (the executable) of all targets. Relative to the root directory of the target.
Array of wildcard patterns (like in the shell) that determines if a certain file should be ignored. Useful when working with templates sometimes, or when parts of the source code should not be compiled.
Array of wildcard patterns (like in the shell) that determines if a certain path should not be scanned for headers. Useful when you want to parts of the code that is not C/C++, where it is not meaningful to look for #include.
Array of file names to use as roots when looking for files that needs to be compiled. Anything that is not an option that is specified on the command line is appended to this variable. The special value * can be used to indicate that all files with an extension in the ext variable should be compiled. This is usually what you want when you are compiling a library of some kind.
String specifying the name of the output file. If not specified, the name of the first input file is used instead.
Append the original extension of the original source file to the intermediate file when compiling. This allows mymake to compile projects where there are multiple files with the same name, e.g. foo.cpp and foo.c without both trying to create foo.o and thereby causing compilation to fail. Mymake warns you if you might need to add use this option.
Array of paths that should be added to the include path of the compilation.
Flag to prepend all elements in include.
Generated automatically by mymake, equivalent to adding the contents of includeCl before each element in include.
Array of system libraries that should be linked to your executable.
Flag to prepend all elements in library.
Array of local libraries that should be linked to your executable (usually used in a project).
Flag to prepend all elements in localLibrary.
Automatically generated by mymake, equivalent to adding libraryCl before all elements of library, also including local libraries.
Preprocessor defines.
Preprocessor define flag.
Yes or no, telling if mymake should execute the program after a successful compilation. This can be overridden on the command line using -e or -ne.
The precompiled header file name that should be used. If you are using the default configuration, you only need to set this variable to use precompiled headers. If you are using #pragma once in gcc, you will sadly get a warning that seems impossible to disable (it is not a problem when precompiling headers).
The name of the compiled version of the file in pch.
Command line for compiling the precompiled header file.
If set to yes, pchCompile is expected to generate both the pch-file and compile a .cpp-file.
Array of command-lines that should be executed before the build is started. Expands variables.
Array of files created by the pre-build step which should also be included in the compilation. These are expected not to introduce any additional dependencies into the project, as they are not available at the point where mymake resolves dependencies between files and targets.
Array of command-lines that should be executed after the build is completed. Expands variables.
Array of command lines to use when compiling files. Each command line starts with a pattern (ending in :) that is matched against the file name to be compiled. The command line added last is checked first, and the first matching command-line is used. Therefore it is useful to first add the general command-line (starting with *:), and then add more specific ones. Here, you can use <file> for the input file and <output>.
Command line used when linking the intermediate files. Use <files> for all input files and <output> for the output file-name.
Link the output of one target to any target that are dependent on that target. See projects for more information.
Forward any of this target's dependencies to any target that is dependent on this target.
Set environment variables. Each of the elements in env are expected to be of the form: variable=value or variable<=value or variable=>value. The first form replaces the environment variable variable with value, the second form prepends value to variable using the system's separator (: on unix and ; on windows), the third form appends value to variable. The second and third forms are convenient when working with PATH for example.
In projects: ignore any potential targets that do not have their own .mymake-file.
In projects, this indicates if projects that have all dependencies satisfied may be built in parallel. The default value is yes, so projects not tolerating parallel builds may set it to no. In targets, this indicates if files in targets may be built in parallel. If so, all input files, except precompiled headers, are built in parallel using up to maxThreads threads globally. If specific targets do not tolerate this, set parallel to no, and mymake will build those targets in serial.
Limits the global number of threads (actually processes) used to build the project/target globally.
When building in parallel, add a prefix to the output corresponding to different targets. Defaults to either vc or gnu (depending on your system). If you set it to no, no prefix is added. vc adds n> before output, gnu adds pn: before output. This is so that Emacs recognizes the error messages from the vc and the gnu compiler, respectively.
Send absolute paths to the compiler, this helps emacs find proper source files in projects with multiple targets.
(defaults to yes), if set, mymake tries to figure out dependencies between targets by looking at includes. Sometimes, this results in unneeded circular dependencies, causing compilation to fail, so sometimes it is neccessary to set this to no.

All variables that contain commands to execute can be prefixed with an integer followed by a colon (for example 1:gcc). If prefixed with the integer N, the first N lines of output from that command are removed from the output. As such, this can be used to ignore status information from commands during the build, without ignoring useful information. For example, the C++ compiler on Windows outputs a line containing the filename being compiled by default. This is unnecessary as mymake already does this. It can thus be suppressed by issuing the command 1:cl ....

This feature is possible to use in the variables compile, pchCompile, and link.

VARIABLES IN STRINGS

When mymake uses some variables (most notably, the compilation and link command lines) it looks at each string and recursively replaces any variables that appear there. Note that this is not done when the configuration is evaluated, only when the variables are actually used.

Any occurrences of <variable> are replaced with the contents of variable. It is also possible to prepend a string to each element in another variable using the syntax <prefix*variable>, which means that the string in the variable prefix is prepended to each element in the variable variable.

The special variable <env:X> (where X is any string) can be used to extract the value of environment variables. This is useful to, for example, inspect the value of CFLAGS during compilation. The benefit of using this syntax over relying on shell expansion is that it allows mymake to track changes in the environment variables.

It is also possible to perform an operation on each element in the array using the syntax <op|variable>. It is also possible to both perform an operation and prepend data using <prefix*op|variable>. Supported operations are:

Treat the element as a path and extract the file or directory name (e.g. src/foo.txt gives foo.txt).
Same as title, but the file extension is removed as well.
Remove the file extension from a path.
Format the element as a path for the current operating system. For example src/foo.txt evaluates to src\foo.txt on Windows.
Make the element into a path inside the build path.
Make the element into a path inside the executable path.
Evaluates to the parent directory of the path. If no parent is given (e.g. only a file name), the element is removed from the array.
Make all elements empty. This can be used to test if a variable contains a value and then include some other text. For example <usePch*if|pchFile> to add the flag inside usePch if a file is specified in pchFile.

PROJECTS

A project is a collection of targets linked together by a .myproject-file. The .myproject file is evaluated multiple times with different options present to extract information about the project:

build: This option is specified during one evaluation to extract a list of options to apply to each of the sub-projects. Thus, a project file typically contains a section as follows:

[build]
main+=debug
libfoo+=lib
libfoo+=debug

In this case, we instruct mymake to build the main target with the debug option present, and the target libfoo with the options debug and lib. There is also a special target, all, which options will apply to all targets in the project.

deps: This option is specified during one evaluation to extract explicit dependencies between projects. By default, mymake finds dependences between projects automatically by examining includes across projects. In certain cases it is, however, useful to introduce extra dependencies to ensure that some dynamically loaded parts are also built. This section is very similar to the build section. Mymake expects to find variables corresponding to each target, and that these variables contains names of other targets.

If one target results in a library, it is convenient to set the variable linkOutput to yes for that target. Mymake will then add the output of the library target to the library variable of any targets that depend on it.

After mymake has extracted the necessary project information, the .myproject-file is also evaluated once for each target. Thus, it is possible to specify additional variables or options that apply to all targets in the project in the .myproject file.

June 22 2021