[Octopus-devel] [Octopus-notify] svn commit: r5253 - in trunk/src: basic hamiltonian ions states by acastro

[Micael Oliveira]

Hi,

  I think we should be careful with this. Some time ago I had problems 
with this, as some compilers did not nullify the pointers in data types 
when you declare them to be components of another data type. I don't 
know is if this is a bug in the compilers or in the fortran standard...

  Cheers,

Micael

octopus-notify at tddft.org wrote:
>       Author: acastro
>         Date: Thu Apr 16 17:44:16 2009
> New Revision: 5253
>    Changeset: http://www.tddft.org/trac/octopus/changeset/5253
> 
> Modified:
>    trunk/src/basic/distributed.F90
>    trunk/src/basic/multicomm.F90
>    trunk/src/hamiltonian/epot.F90
>    trunk/src/hamiltonian/hamiltonian.F90
>    trunk/src/ions/geometry.F90
>    trunk/src/ions/simul_box.F90
>    trunk/src/main/run.F90
>    trunk/src/states/states.F90
>    trunk/src/states/states_dim.F90
> 
> Log:
> Added default initialization to NULL() of the pointers included in
> several data types. This allows calling "associated(x)" without crashing.
> 
> In the way, I found one copy of a geometry type variable through the
> previously overloaded "=", which should be done now with
> "geometry_copy". This caused open_systems/03-extended_eigenstate.test
> to fail in my machine; now it still fails, but due to something else.
> 
> 
> _______________________________________________
> Octopus-notify mailing list
> Octopus-notify at tddft.org
> http://www.tddft.org/mailman/listinfo/octopus-notify
>