dolfin team mailing list archive

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

On Thu, 2010-08-19 at 10:26 +0200, Anders Logg wrote:
> On Thu, Aug 19, 2010 at 08:28:31AM +0100, Garth N. Wells wrote:
> > On Thu, 2010-08-19 at 09:20 +0200, Anders Logg wrote:
> > > On Thu, Aug 19, 2010 at 08:06:23AM +0100, Garth N. Wells wrote:
> > > > On Wed, 2010-08-18 at 15:23 +0000, noreply@xxxxxxxxxxxxx wrote:
> > > > > ------------------------------------------------------------
> > > > > revno: 5013
> > > > > committer: Anders Logg <logg@xxxxxxxxx>
> > > > > branch nick: dolfin-dev
> > > > > timestamp: Wed 2010-08-18 13:57:25 +0200
> > > > > message:
> > > > >   Improve search for PETSC_PATH by examining LD_LIBRARY_PATH.
> > > > >   No need to set PETSC_PATH or PETSC_ARCH if LD_LIBRARY PATH is set.
> > > >
> > > > Sounds platform dependent to me? Probably better to search the CMake
> > > > list of library locations.
> > >
> > > It's just one of many in a list of places where to look. The list of
> > > Debian install locations is also platform-dependent. We could add
> > > other locations as well.
> > >
> >
> > My point is that CMAke should know the standard library locations better
> > than we do, so let CMake provide the info when possible. When we can't
> > expect CMake to know the location (like something Debian-specific), then
> > we should provide it.
> >
> > Garth
> 
> Does CMake pick up LD_LIBRARY_PATH (or any other standard variables
> like PATH)?
> 
> In that case we could search those.
> 

It has a bunch of different variables for libs, includes, etc.

Garth

> --
> Anders