[Dune] last changes of dgfparser.hh and gridtype.hh

Andreas Dedner andreas.dedner at mathematik.uni-freiburg.de
Wed Aug 22 15:07:46 CEST 2007


Hi,
I still do not see the advantage of including
all grid implementations found during configure
in gridtype.hh - only one typedef GridType
can be set in any case. If more than one grid
implementation
is to be used in a code than the user must hard wire them
into his code in any case, e.g., including gridtype.hh
and dgfoned.hh to use one grid defined in the Makefile
and a OneDGrid...
As Christoph said including all grid implementation just
needlessly prolongs compilation time - which takes a long
time anyway.
Greeting Andreas



On Wed, 22 Aug 2007 14:52:44 +0200
 Christoph Gersbacher <gersbach at mathematik.uni-freiburg.de>
wrote:
> Hi Andreas and Christian,
> 
> as I wrote Christian before, after the last updates of
> dgfparser.hh and gridtype.hh the grid tests don't compile
> any longer. It seems that there are recursive include
> commands in dgfparser.hh. Moreover
> <grid/utility/gridtype.hh> includes ALL grids found by
> configure - not only the specified one, unnecessarily
> consuming compile time and possible errors. Some errors
> can be avoided by using the $(ALL_PKG_CPPFLAGS) - but if
> you try to compile with GRIDDIM=1, alberta.h is included
> producing errors! I attach a quick workaround that works
> for me, but is not really nice, since dgfparser.hh just
> mimicks the structure of gridtype.hh. 
> 
> Unfortunately I don't have permission to commit these
> changes. Do you have a better idea?
> 
> Cheers,
> 
> Christoph.
> 
> ______________________________________
> Christoph Gersbacher
> Section of Applied Mathematics
> email: gersbach at mathematik.uni-freiburg.de
> phone.: +49 - (0)761 - 203 5635
> 




More information about the Dune mailing list