[Bug fortran/117798] Audit intrinsic subprograms with scalar INTENT(OUT) character strings

[kargls at comcast dot net via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117798

--- Comment #6 from kargls at comcast dot net ---
On 2/3/25 02:14, tkoenig at gcc dot gnu.org wrote:
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117798
> 
> Thomas Koenig <tkoenig at gcc dot gnu.org> changed:
> 
>             What    |Removed                     |Added
> ----------------------------------------------------------------------------
>                   CC|                            |tkoenig at gcc dot gnu.org
> 
> --- Comment #5 from Thomas Koenig <tkoenig at gcc dot gnu.org> ---
> (In reply to kargls from comment #2)
>> I think that Fortran 2023 may break the ABI for gfortran.
> 
> We have two choices. Either bump the ABI versoin (ugh) or have
> two versions, only call the new one for a new compiler version, and
> remove the old one on an ABI break.  Maybe it could be useful
> to use the old version to implement part of the new one, I haven't
> looked.

I believe the effected intrinsic subprograms won't occur
in a hot-loop, so in pseudo-code, we can do:

if (string is allocatable) then
    create tmp
    use current implementation with tmp
    copy tmp to string with current re-allocation algorithm
else
    use current implementation with string
end if

The only question would be how to determine size of tmp.

For IO, read(string,...) we should be able to determine
the needed length from a format statement or the io-list.