LearnHub

Automatic Variables

• $@ - The rule's target
• $< - The rule's first prerequisite
• $? - All the out of date prerequisites
• $^ - All prerequisites

all : hydroxyl_422.csv methyl_422.csv

hydroxyl_422.csv : hydroxyl_422.dat
        @dat2csv $< > $@

methyl_422.csv : methyl_422.dat
       @dat2csv $< > $@

clean :
       @rm -f *.csv

While he is at it he renames his file from hello.mk to Makefile which is the default name make expects for its files which means make all is what he now types to convert his files. Hurray for less typing!

Pattern Matching

Nigel's Makefile is getting increasingly powerful, but increasingly complex but it lets him convert his files by typing just two words in the console. But what if he added another compound to his experiment? hydroxyl_480.csv for example. He would have to edit the Makefile to create a new rule.

He can however avoid this by using Make's pattern matching capabilities. Based upon the fact that most project manipulate similar types of files in a similar manner, Make has a single wildcard, %, which matches the stem portion of a file.
all : hydroxyl_422.csv methyl_422.csv hydroxyl_480.csv

%.csv : %.dat
  dat2csv $< > $@

clean :
        rm -f *.csv
Here we see a target that will match all csv files, and which depends on all the dat files.

Somethings to note here are:

• We have sacrificed another bit of clarity (individual rule) for power (a more powerful single rule)
• Nigel still has to add new csv files to the all rule, but that is easier than writing a whole new rule
• When using pattern matching, you must use automatic variables in the action. This is because Make does not know the actual name of either the target or prerequisite files until runtime.

More Dependencies

all : hydroxyl_all.csv methyl_all.csv

%_all.csv : %_422.csv %_480.csv
 summarize $^ > $@

%.csv : %.dat
   dat2csv $< > $@

clean :
        @rm -f *.csv

• the %_all.csv rule is checked for any dependencies
• the %.cvs rule is triggered to run the dat2csv file and create new csv files
• the %_all.csv action runs the summarize command producing both hydroxyl_all.csv and methyl_all.csv
• it then deletes all the csv files it created while producing the ones asked for (in this case hydroxyl_422.csv, methyl_422.csv and hydroxyl_480.csv)

Macros

INPUT_DIR = /lab/gamma2100
OUTPUT_DIR = /tmp

all : ${OUTPUT_DIR}/hydroxyl_all.csv ${OUTPUT_DIR}/methyl_all.csv

${OUTPUT_DIR}/%_all.csv : ${OUTPUT_DIR}/%_422.csv ${OUTPUT_DIR}/%_480.csv
    @summarize $^ > $@

${OUTPUT_DIR}/%.csv : ${INPUT_DIR}/%.dat
       @dat2csv $< > $@

clean :
       @rm -f *.csv

• To set a value, you assign it with an =
• To access a value, you reference the variable with a $ and put {} around it. If you forget the {}, you will be accessing $O followed by the characters UTPUT_DIR after it instead of ${OUTPUT_DIR}
• By convention, macro names are in all uppercase

Passing Values to Make

Functions

all : ${SUMMARIES}

${OUTPUT_DIR}/%_all.csv : ${OUTPUT_DIR}/%_422.csv ${OUTPUT_DIR}/%_480.csv
       @summarize $^ > $@

${OUTPUT_DIR}/%.csv : ${INPUT_DIR}/%.dat
       @dat2csv $< > $@

clean :
       @rm -f *.csv

• $(addprefix prefix,filenames) - Add a prefix to each filename in a list
• $(addsuffix suffix,filenames) - Add a suffix to each filename in a list
• $(dir filenames) - Extract the directory name portion of each filename in a list
• $(filter pattern,text) - Keep words in text that match pattern
• $(filter-out pattern,text) - Keep words in text that don't match pattern
• $(patsubst pattern,replacement,text) - Replace everything that matches pattern in text
• $(sort text) - Sort the words in text, removing duplicates
• $(strip text) - Remove leading and trailing whitespace from text
• $(subst from,to,text) - Replace from with to in text
• $(wildcard pattern) - Create a list of filenames that match a pattern

Summary

• Automate repetitive tasks
• Remove duplication within your Makefile, and
• Build you Makefile incrementally from least to greatest complexity