C::B itself is a good testing workspace, though probably too complicated. It shows the problem on Windows (On Linux, it seems to work as I think it should).
Re-build totally the workspace and you'll see that only the last created .depend file has something else than the title line.
If you rebuild only one of the sub-project, the corresponding .depend file seems to be correct.
On a C++ workspace, I must admit, as you said, that it's only bother me, because it seems to work without these .depend files. (if I say "seems" it's because I'm not totally sure of this in all cases: sometimes I need to build 2 or 3 consecutively times the C::B build workspace because it claims that some header files are missing though they are here. It looks this is essentially with .gch files. Is it the same problem or another, I don't know).
On a fortran project, it's not the same. As you suggested me in an other post (
here), I first tried to understand how desplib works to be able to eventually modify or adapt it to fortran project. But, until now, I don't really understand how it works, and more, if it's not really used (or useful), it's not here that I have to search how to get dependencies work in fortran included files. That's the main reason of this help request.
gd_on