How do I perform arithmetic in a makefile?

Makefile

Makefile Problem Overview


Is it possible to perform some operations on variables in a makefile? For instance, defining

JPI=4
JPJ=2

is it possible to define in the same makefile a variable JPIJ equal to the expanded value of $(JPI)*$(JPJ)?

Makefile Solutions


Solution 1 - Makefile

Using Bash arithmetic expansion:

SHELL=/bin/bash
JPI=4
JPJ=2
all:
	echo $$(( $(JPI) * $(JPJ) ))

The first line is to choose the Bash shell instead of the default (sh). Typically, sh doesn't support arithmetic expansion. However in Ubuntu, /bin/sh is provided by Dash, which supports this feature. So that line could be skipped.

The double dollar sign is because we want the expansion to be done by the shell. Note: the JPI and JPJ variables are expanded by make first, then the expression is passed to bash like this:

$(( 4 * 2 ))

Solution 2 - Makefile

Answer from @mrkj is great but as @Daniel mentions, not all systems have bc (for example, I don't have it on MSys).

I have found the two following methods, both using shell: $$(( ... )) and expr ...

JPI=4
JPJ=2

#With Double-dollar
JPIJ_1 = $(shell echo $$(( $(JPI) + $(JPJ) )))

#With 'expr'
JPIJ_2 = $(shell expr $(JPI) + $(JPJ) )

$(info Sum with Double-$$: $(JPIJ_1))
$(info Sum with 'expr': $(JPIJ_2))

Note that when using expr, you shall put spaces around the + or it will return 4+2. This is not required when using $$.

.

When you have bc available, you might definitely go with it. I found the following page very interesting: http://www.humbug.in/2010/makefile-tricks-arithmetic-addition-subtraction-multiplication-division-modulo-comparison/

Solution 3 - Makefile

If you're using GNU make and have bc installed on your system, you can use something like this:

JPI=4
JPJ=2
FOO=$(shell echo $(JPI)\*$(JPJ) | bc)
all:
  echo $(FOO)

Solution 4 - Makefile

It's clumsy (or brilliant, depending on your perspective), but you can do arithmetic directly in GNU make. See Learning GNU Make Functions with Arithmetic. Be aware though that this method doesn't scale well. It will work wonderfully for small numbers as you have shown in your question, but it doesn't do well when you're working with numbers with a large magnitude (greater than 10,000,000).

Solution 5 - Makefile

In GNU Make with Guile support (i.e. since version 4.0) it is easy to use call to Scheme language for arithmetic or other calculations. It is done without creating any subshell or child process.

An example

JP-I := 4
JP-J := 2
JP-IJ := $(guile (* $(JP-I) $(JP-J) ))

$(info JP-IJ = $(JP-IJ) )
# prints: JP-IJ = 8

See also the manual for Guile Arithmetic Functions.

A possible check for Guile:

ifeq (,$(filter guile,$(.FEATURES)))
  $(error Your Make version $(MAKE_VERSION) is not built with support for Guile)
endif

Solution 6 - Makefile

The GNU Make Standard Library provides integer arithmetic functions.

include gmsl

JPI = 4
JPJ = 2

JPIJ = $(call plus,$(JPI),$(JPJ))

Solution 7 - Makefile

With makepp it's much easier. You get direct access to the underlying Perl interpreter. In this case the makeperl function does variable expansion before evaluating as Perl, the perl function OTOH would only evaluate:

JPI=4
JPJ=2
JPIJ = $(makeperl $(JPI)*$(JPJ))
&echo result: $(JPIJ)

You can use the builtin &echo command outside of a rule as a statement.

Solution 8 - Makefile

To add a late answer to the pool: The GNUmake table toolkit features many arithmetic functions. You can add, subtract, multiply, divide, take the modulus in base 8,10 and 16. Also there are the usual binary operations and, or, xor and not. Numbers can be around 60 digits but you can adapt this, if you need more. The code is pure GNUmake syntax and therefore portable between Windows and Unix, contrary to shell scripts - in case you want to number crunch, there may be better solutions ;) of course.

Here is an example:

include gmtt/gmtt.mk

NUMBER_A := -12392834798732429827442389
NUMBER_B := 984398723982791273498234
$(info $(call add,$(NUMBER_A),$(NUMBER_B)))
$(info $(call sub,$(NUMBER_A),$(NUMBER_B)))
$(info $(call mul,$(NUMBER_A),$(NUMBER_B)))
$(info $(call div,$(NUMBER_A),$(NUMBER_B)))
$(info $(call mod,$(NUMBER_A),$(NUMBER_B)))

Output:

$ make
-11408436074749638553944155
-13377233522715221100940623
-12199490762401735834920873237276176262117128241026
-12
-580050110938934545463581

Attributions

All content for this solution is sourced from the original question on Stackoverflow.

The content on this page is licensed under the Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) license.

Content TypeOriginal AuthorOriginal Content on Stackoverflow
QuestionmalabocaView Question on Stackoverflow
Solution 1 - MakefileDominicView Answer on Stackoverflow
Solution 2 - MakefileJean-Francois T.View Answer on Stackoverflow
Solution 3 - MakefilemrkjView Answer on Stackoverflow
Solution 4 - MakefileEric MelskiView Answer on Stackoverflow
Solution 5 - MakefileruvimView Answer on Stackoverflow
Solution 6 - MakefileMatthew SimoneauView Answer on Stackoverflow
Solution 7 - MakefileDanielView Answer on Stackoverflow
Solution 8 - MakefileVroomfondelView Answer on Stackoverflow