dependencies

[Magnus Hagdorn]

On Sat, 2006-09-09 at 17:24 +0200, Steven Bosscher wrote:
> On Saturday 09 September 2006 12:36, Erik Edelmann wrote:
> > On Sat, Sep 09, 2006 at 12:45:27AM +0300, Janne Blomqvist wrote:
> > > Steve Kargl wrote:
> > > >On Fri, Sep 08, 2006 at 10:51:42AM -0600, Warren Turkal wrote:
> > > >>On Friday 08 September 2006 10:34, Steve Kargl wrote:
> > > >>>I'm not aware of such an option, but then I keep track of
> > > >>>dependencies in my Makefiles as I add new code.
> > > >>
> > > >>Well, it would be nice to have gfortran -M that used cpp -M + the
> > > >> fortran stuff like use and include statements to generate
> > > >> dependencies. I guess I can file an RFE or something.
> > > >
> > > >You might ask on comp.lang.fortran about a tool to generate
> > > >the dependencies.  I vaguely recall that someone once posted
> > > >about a perl script.
> > >
> > > I've been using makedepf90,
> > > http://www.helsinki.fi/~eedelman/makedepf90.html (made by gfortran
> > > maintainer Erik Edelman).
> >
> > I can add that the parser of that tool isn't quite of the quality I
> > would like it to be (it was written many years ago, in my clueless
> > youth).  But for most cases it works pretty well.
> 
> There is also still mkf90.py, which Andy Vaught wrote long ago to
> bulk-compile millions of lines of Fortran to test the g95 parser
> (we're talking 2000, 2001 here).  Someone with a sufficient amount
> of python-fu could turn that script into a Makefile dependencies
> generator.
> 
> You should be able to find mkf90.py with Google.  I haven't tried
> but it used to be pretty easy for Google to find some copies of it.
I've also written a python fortran dependency checker. It works quite
well for me, although there are undoubtedly many wrinkles. You can grab
it from here:
https://forge.nesc.ac.uk/cgi-bin/cvsweb.cgi/glimmer/src/python/f90_dependencies.py?cvsroot=glimmer
cheers
magi