What are the true portable or best to define real 4 real 8 integer 4 integer 8, etc?

[vmagnin]

There was that discussion in 2020:

Best way to declare a double precision in Fortran?

Hi, there is several ways to declare a double precision real in Fortran: real*8 :: w real(8) :: x real(kind(0d0)) :: y double precision :: z My habits are to use double precision. Is it a good practice ?

@sblionel is pleading for selected_real_kind()

Personally, I now use iso_fortran_env, which is sufficient for my simple needs (computing on a PC).