https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122957

--- Comment #3 from Martin Köhler <martin.pleissa at web dot de> ---
Even if the code compiled with the -fdefault-integer-8 option has some smell,
due to fact that it is most likely legacy code, it would be nice to allow some
workaround, which is portable over the Fortran compilers. I checked the
flang-based compilers and it seems that there the integer size in the user
defined IO functions chnage with their "integer-8" flag. Thus, to ensure the
compatibility, having the user defined io working with integer-8 would be
great. By now at least having a proper warning/error message that indicate this
point would be helpful as well. 

Beside that, it is clear to me, that the legacy codes need to be restyled.

Reply via email to