dolfin team mailing list archive

["Garth N. Wells" <gnw20@xxxxxxxxx>]

On Sun, 2010-08-22 at 01:17 +0200, Anders Logg wrote:
> On Sun, Aug 22, 2010 at 01:11:50AM +0200, Anders Logg wrote:
> > On Sat, Aug 21, 2010 at 11:49:06PM +0100, Garth N. Wells wrote:
> > > On Sat, 2010-08-21 at 20:52 +0000, noreply@xxxxxxxxxxxxx wrote:
> > > > ------------------------------------------------------------
> > > > revno: 5046
> > > > committer: Anders Logg <logg@xxxxxxxxx>
> > > > branch nick: dolfin-dev
> > > > timestamp: Sat 2010-08-21 22:45:13 +0200
> > > > message:
> > > >   Add FindCGAL.cmake.dolfin for finding CGAL config. Includes compile/run
> > > >   test that passes on Lucid.
> > >
> > > I don't think that this is the right approach (plus it no longer works
> > > on my computer).
> > >
> > > CGAL uses CMake, and generates a config file with loads of build
> > > information, like libraries on which it depends. We should use the CMake
> > > config file.
> > >
> > > Garth
> >
> > It didn't work for me before and I found the CGAL CMake file difficult
> > to understand. Works fine for me now on Lucid.
> 
> I take that back. It still breaks on Lucid. It also breaks on Snow
> Leopard for the same reason but in different places. On Lucid, it
> breaks when I try to build a demo and on Snow Leopard it breaks
> earlier when it links something.
> 
> In both cases, the problem is missing references to mpfr (which I have
> installed).
> 
> What flags should be added?
> 
> Perhaps it's best to rely on ${CGAL_DIR}/CGAL/CGALConfig.cmake to get
> the variables but I don't think ${CGAL_DIR}/CGAL/FindCGAL.cmake looks
> very good.
> 

I've never looked at it - I just look at CGALConfig.cmake which is easy
to follow. I guess we need to add 

    CGAL_3RD_PARTY_LIBRARIES_DIRS
    CGAL_3RD_PARTY_LIBRARIES

Garth


> --
> Anders
> 
> 
> > Another problem with using the installed CMake file was that if CGAL
> > was not installed then it would just assume things work.
> >
> > Anoher improvement with the new config script is that it actually
> > tries to build a CGAL test program (which uses essentially the stuff
> > that we need in DOLFIN).