Re: What does F2023:C8107 mean?

On Thu, 29 Aug 2024 17:59:04 +0000, Steven G. Kargl wrote:

F2023:C8107 states
 
   The namelist-group-name shall not be a name accessed by use association.
 
Are following pieces of code standard conforming?
 
! Code 1
module mod_nml1
   implicit none
   logical :: ldiag
   namelist /nam_nml1/ldiag
end module mod_nml1
 
program ice_nml
   use mod_nml1
   implicit none
   integer :: ilu
   ldiag = .false.
   write(*,nml=nam_nml1)  ! <-- Use assoc. of namelist-group-name
end program ice_nml
 
! Code 2
module mod_nml1
   implicit none
   logical :: ldiag
   namelist /nam_nml1/ldiag
end module mod_nml1
 
program ice_nml
   use mod_nml1
   implicit none
   integer :: ilu, j
   namelist /nam_nml1/j   ! <-- Use assoc of namelist-group-name
   ldiag = .false.
   j = 42
   write(*,nml=nam_nml1)  ! <-- Use assoc of namelist-group-name
end program ice_nml
 
Clarification of the interpretation of C8107 would be appreciated?

After a few minutes re-reading parts of Fortran 2023, I have concluded
that Code 1 is conforming and Code 2 is non-conforming.  My issue was
caused by gfortran accepting Code 2 without and error or warning under
-std=gnu option (default behavior).  If -std=f2023 is used, then gfortran
issues an expected error messages.

--
steve