If you are trying to understand dynamic linking, this question is likely to be of interest.
One of the answers to that question provides a wonderful example of creating and using a dynamic library. Based on it, I some simple files:
main.c:
extern void someFunction (int x);
int main (int argc, char** argv ) {
someFunction(666);
}
mylibrary.c:
#include <stdio.h>
void someFunction (int x) {
printf ("\nsomeFunction called with x=%d\n", x);
}
makefile:
main: mylibrary.c main.c
gcc -c mylibrary.c
gcc -dynamiclib -current_version 1.0 mylibrary.o -o libmylibrary.dylib
gcc -c main.c
gcc -v main.o ./libmylibrary.dylib -o main
clean:
rm *.o
rm main
rm *.dylib
So far, everything works great. If I make and then enter ./main at the command prompt, I see the expected output:
someFunction called with x=666
Now, I want to mix things up a little. I've created a directory hidelib, which is a subdirectory of my main directory. And I'm adding one line to my makefile:
main: mylibrary.c main.c
gcc -c mylibrary.c
gcc -dynamiclib -current_version 1.0 mylibrary.o -o libmylibrary.dylib
gcc -c main.c
mv libmylibrary.dylib hidelib # this is the new line
clean:
rm *.o
rm main
rm hidelib/*.*
Now, I want to add another line to the makefile so it will find libmylibrary.dylib in the hidelib subdirectory. I want to be able to run ./main in the same way. How can I do that?
EDIT: Thanks for the response. Having lots of options is wonderful, but a beginner just wants one concrete option that works. Here is what I am trying for the last line, but clearly I don't understand something. The makefile executes without errors, but at runtime it says "library not found."
gcc main.o -rpath,'$$ORIGIN/hidelib' -lmylibrary -o main
One concrete option that works would be to set the install_name
flag when linking the .dylib
.
gcc -dynamiclib -install_name '$(CURDIR)/hidelib/libmylibrary.dylib' -current_version 1.0 mylibrary.o -o libmylibrary.dylib
Then you can just link to the library normally:
gcc main.o -L '$(CURDIR)/hidelib' -lmylibrary -o main