Yes and that is the better way to do it. Compiling an enormous C file is slow. Many compilers have low limits (around 50 MB or so) but gcc has a huge limit. But even on gcc, a large file slows down the build process significantly.

More importantly gcc's memory requirements balloon quickly as the size of the file grows. There are projects that can't be built on platforms like a Raspberry Pi simply because some of the source files (typically heavily templated C++ code) blow out the memory. A good example is Firefox.

I like objcopy for this. Here is an example command:

    objcopy --input binary --output elf64-x86-64 myfile.txt myfile.o

you could get a look at https://csl.name/post/embedding-binary-data/

What is the clang or macos equivalent to objcopy?

Pass -sectcreate to the linker, e.g. -sectcreate __DATA __foo foo.txt. Then you can reference the data like this:

    extern char foo_start[] asm("section$start$__DATA$__foo");
    extern char foo_end[] asm("section$end$__DATA$__foo");

You are just dealing with ELF files, so objcopy works fine in those scenarios as long as the architecture matches for the linker. This trick works many other compilers that use ELF such as clang, MSVC, Green Hills, etc.

Except that MacOS does not use ELF, it uses Mach-O.

Clang handles intermediate ELF object files fine and can link into a Mach-O final executable.

Windows under MSVC doesn't use ELF either (it uses PE for the .EXE) and it links in intermediate ELF object files fine. I've made an EXE that was a mix of files compiled by MSVC and GCC.