Automatic variables

Automatic variables are an advance feature you may use inside your rules. These variables expand to very specific values regarding data available on rule’s header.

Note D and F variations of the variable are actually splitting every word in Directory and File parts:

$ cat Makefile
/tmp/file:
    @echo "The directory is $(@D)"
    @echo "The file is $(@F)"
$ make
The directory is /tmp
The file is file

Let’s see the those in action with a complete example. Imagine you want to build a C program which needs one source file and at least 2 header files. $ tree
.
├── header1.h
├── header2.h
└── program.c

So to build the program you need to generate a couple of implicit rules (You may use the built-in rules but you are stubborn and a DIY guy) so: $ cat Makefile

headers:= header1.h header2.h

program: program.o
    $(CC) $(CFLAGS) -o $@ $<

program.o: program.c $(headers)
    $(CC) -c $(CFLAGS) -o $@ $<

$ make program
cc -c  -o program.o program.c
cc  -o program program.o

This example is a little more complex so I’m going to take a closer look to every part of the line. – We decided to store all header file names in a variable so we can use it later. This is an easy task. headers:= header1.h header2.h
– Now we define our first target: program which has a prerequisite on it’s object. program: program.o
$(CC) $(CFLAGS) -o $@ $<;

Here we used 2 automatic variables. And the expansion of the recipe looks like: cc -o program program.o

As you can see the automatic variables has done their work pretty well!
– Last we can see the rule for making the object file and it’s output fits what was described.

program.o: program.c $(headers)
    $(CC) -c $(CFLAGS) -o $@ $<

cc -c  -o program.o program.c

Make is mixing that information together an executing the proper commands in the proper order. But for a better explanation about how rules works see the section dedicated to rules.