WHIZARD crashes immediately due to gfortran regression (?)

[kilian]

With today's snapshot (3 Nov 2009) of gfortran, WHIZARD crashes right at the beginning, even if the input file whizard.in.empty is provided.

The reason is a mismatch in optional arguments. An argument that is not PRESENT is suddenly PRESENT in a subroutine.

For checking, add a line in models.f90, procedure model_set_particle_data:

  ...
  logical, parameter :: is_stable = .true.
  print *, "model_set_particle_data:  present(name) = ", present (name) !NEW
  call particle_data_set (model%prt(i), &
  ...

and similarly in procedure particle_data_set:

  ...
  type(parameter_t), intent(in), target, optional :: mass_src, width_src
  print *, "particle_data_set: present (name) = ", present (name) !NEW
  if (present (is_visible))  prt%is_visible = is_visible
  ...

This results in the following output:

|=============================================================================|
|                               WHIZARD 2.0.0                                 |
|=============================================================================|
| Initializing process library 'processes'
| Loading process library 'processes'
| [No compiled process library 'processes']
| Reading model file 'SM.mdl'
 model_set_particle_data:  present(name) =  F
 particle_data_set: present (name) =  T
Operating system error: Cannot allocate memory
Out of memory

If this can be reproduced, a bug report to gfortran is in order (or maybe this is already known?)

[Juergen Reuter]

I reported this as bug report 41948 at gcc bugzilla but it seems to be already known as bug

41907. This has been fixed in the gfortran trunk in the meanwhile, at least from revision r153950

on. Therefore, closing this ticket.

[Juergen Reuter]

Milestone v2.0-beta deleted