GNU Make is a build automation tool typically used on Linux systems1. Below I will summarize a list of FAQs, which explain how Make works and how to use Make.

General questions

Why do we need Make?

Suppose that we want to build an executable or a library with multiple source and header files, to build it, we often need to run quite a few commands. During the development cycle, we need to repeatedly run these commands to build the project. Of course we can do it manually, but it would be time-consuming and error-prone.

With Make, we can run repeating build tasks easily without having to type the build commands manually.

How is Make different from CMake?

Make is build tool that uses Makefile to build a project. CMake is a meta-build tools, in that it can produce a Makefile or other build files such as build.ninja for Ninja.

What name should we use for Makefile?

Make usually looks for makefile or Makefile in the current directory. The convention is to use Makefile. However, makefile also works fine.

What is a target, recipe?

A rule is a bunch of commands we want to run together to build a target, e.g., building a certain object file. The format is like:

Target_name: dependency
# <--> indicates a TAB character.
<-->command1
<-->command2
...

These commands are called recipe for this target. Each command must start with a TAB, though it can be changed in newer versions of make.

A target is the result of executing the rule.

How does make work generally?

Make is smart. It will not blindly build all the target each time you run it. Since each target may have some dependencies, and each dependency in turn will have their own dependencies. In the end, when we make a target, there is a dependency tree. If you change a source file, make will only compile its parent target and other targets that depend on its parent target recursively. Specifically, make will check the last-modify time of a target’s dependency, if the time stamp is newer than the target itself, it means that this target should be rebuilt to make it up-to-date.

In summary, make is smart enough to only build a target that needs to be rebuilt, reducing compilation time significantly.

Ref:

What is the default target for Make?

When we simply type make, it can build some target. By default, if we do not specify the target to build, make will pick the first target that does not start with . in its name. So we may use a default target (the name does not really matter) as the first target to do the default build action if the user does not specify one.

Ref:

Usage question

Difference between @echo and echo

By default, make will print a command before executing it. Adding @ before a command will suppress its printing before execution.

Ref:

Difference between := and = for assignment?

Variables defined using := get its value when it is defined. Variables defined using = get its value when it is used (so there is a risk that the variable value may change if it depends on the value of other variables).

Ref:

What is a .PHONY target and why is it needed?

PHONY target is a target that does not produce a file. Typically, a target name is the corresponding file to generate. However, for certain target, we only want to run some command and do not want to generate a target of the same name.

If you do not mark the target as phony, make will first check if there is file of the same name that has been updated. If there is a file of the same name that has been updated, make will not build that target, which is not what we want.

For example, if you have the following Makefile:

foo:
	@echo "hello, this is target foo"

If there is a file named foo in the directory, when you run make foo, you get the following message:

make: 'foo' is up to date

So we mark this target as phony (i.e., we do not want to generate file foo):

.PHONY: target_name

By doing this, we make sure that this target will always be build, ignoring a file with the same name. Another example of phony target is the clean target where we want to remove the generated executable and object files:

.PHONY: clean
clean:
	foo bar.o baz.o

Ref:

Special variables used in Makefile

We can use special variable to simplify the writing of Makefile:

  • $@: the target name
  • $<: the first dependency
  • $^: all the dependencies
main: foo.o bar.o
	g++ -o main foo.o bar.o

In the above example, $@ would be main. $< would be foo.o and $^ would be foo.o bar.o.

Ref:

Assign command output to a variable

Sometimes, we want to assign the output of some command to a variable. We can use the shell builtin to get the shell command output:

VIM_PATH := $(shell which vim)

Ref:

References


  1. Technically, Make can also be used on Windows, see here. ↩︎