Re: What is wrong with this simple rule in makefile?

[John Graham-Cumming]

Lin George wrote:
> I want to use g++ $(CFLAGS) to compile any .cpp files
> into .o files. For example, using g++ $(CFLAGS)
> foo.cpp to generate foo.o, using g++ $(CFLAGS) goo.cpp
> to generate goo.o. CFLAGS is a variale which I defined
> before. I write a rule in Makefile as,
>
> %.o :
>         g++ $(CFLAGS) %.cpp
>
> But when executing the Makefile, it reports error
> message "g++: %.cpp: No such file or directory".

There are two important things wrong with this rule:

1. You've specified no prerequisite so this rule means "in order to 
build any file ending in .o do this".  That could be incorrect if there 
were .o's that were not built from .cpp files as you intend.

2. You can't say %.cpp in the rule body.  That's the wrong syntax.  What 
you need to do is use the automatic variable $<.

Here's the correct rule:

    %.o : %.cpp
        g++ $(CFLAGS) $<

You probably also want to set a bunch of other g++ options (such as -o), 
but that's a bit out of the range of what you asked.

The other question I would have is why you need to define this rule at 
all.  Don't GNU Make's built-in rules do what you need? GNU Make has the 
following as a built-in:

    %.o: %.cpp
        $(COMPILE.cpp) $(OUTPUT_OPTION) $<

where COMPILE.cpp = $(COMPILE.cc) and COMPILE.cc = $(CXX) $(CXXFLAGS) 
$(CPPFLAGS) $(TARGET_ARCH) -c and CXX = g++.

So you could use GNU Make's built-in rule and just do CXXFLAGS += 
$(CFLAGS) somewhere in your Makefile and you'd be golden.

John.