Using a library in applications code

In the previous section, we created a library named mylib from three object files called index.o, addtwo.o, and addthree.o. Now we want to use that library to compile an application called six.c:

	int main(void)
	{
		int i, j;

		inc(); inc();
		i = show();
		j = addthree (i, show(), 4);
		return j;
	}

Here, we use the library functions inc(), show(),
and addthree().  These are all contained in the library
mylib, so all we need to do is link that with
six.c to form the executable Six:

	cc -o Six six.c mylib


After compilation, the executable Six will contain the modules
which contain the functions inc(), show(), and
addthree(), but none of the other modules that were archived in
mylib.  Thus, is contains the code for dec(), but
not for addtwo().


Library search paths
By default, the linker will search for a library in one of several directories.
These usually include /lib and /usr/lib, but can include others.  Libraries 
that reside in these directories can be included easily if they conform to a
specific naming convention.


When we compile an executable, we can use the -lname
switch to tell the compiler to look in its search path for the named library.
However, what it will actually look for is the file named
libname.a.  This means that we would have been better
off to name our library libmylib.a instead of just
mylib.  If we change the name we can put our library in a
directory on the search path and use the following command line to link
Six.

	cc -o Six six.c -lmylib


Finally, if we don't have write access to the directories on the search path,
we can add a directory to the search path by using the
-Ldirectory switch:

	cc -o Six six.c -Lsomedirectory -lmylib
Now, the compiler (actually the static linker) will look in the directory
somedirectory for the library archive file libmylib.a.


This page was last modified

.