[Bug fortran/59537] No "Automatic array cannot have an initializer", for -finit-real without a SAVE statement present in subroutine

[maik.riechert at arcor dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59537

Maik Riechert <maik.riechert at arcor dot de> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |maik.riechert at arcor dot de

--- Comment #8 from Maik Riechert <maik.riechert at arcor dot de> ---
I'm facing the same issue using gfortran 8.2.0.

Often the intent of using -finit-real is for debugging purposes. I'd like to
use '-finit-real=snan -ffpe-trap=invalid' to help find cases where unitialized
variables/values are accidently used. However, in a bigger existing code base I
cannot use -finit-real because of the gfortran behavior described in this
issue. If for some reason -finit-real cannot be applied then there should be a
warning at most, but not an error.

Ideally, I'd also like to -finit-real initialize allocatable arrays, for the
same reasons. In my opinion, this should then happen directly after an
allocate(..) call. I have a feeling that -finit-* is too closely linked to the
allowed initializations one can do in the declarations block, however it should
be more general I think and don't have such constraints really.