Suggest that iargc and getarg be replaced by command_argument_count and get_command_argument

[Beliavsky]

For a code such as

PROGRAM test_getarg
  implicit none
  INTEGER :: i
  CHARACTER(len=32) :: arg
  DO i = 1, iargc()
    CALL getarg(i, arg)
    WRITE (*,*) arg
  END DO
END PROGRAM test_getarg

from the gfortran documentation, Fortitude could suggest that iargc and getarg be replaced by command_argument_count and get_command_argument.

[ZedThree]

There's also similar functions in other compilers that could be replaced.

This feels like a new category, something like Portability or Standard?

[LiamPattinson]

This should be quite nice to implement, and I imagine auto-fixes would be fairly easy (at least for the GNU functions, I can't speak for the other compiler-specific functions).

@ZedThree I'm torn, as I think Portability is the better category name but Standard would allow rule codes to have the form STD001. Maybe the latter? I was just thinking Style for this one, but I think you're right that a new category would make sense here.

[ZedThree]

There's lots of other compiler extensions that could go in this category, e.g. abort.

Yes, maybe Standard and STD

[imciner2]

I had actually been thinking about a Portability category that would have rules for the various non-portable things, it is just going to be a pain to gather a list of all the non-portable things. I found where the gfortran source code defines their intrinsics, but I haven't found good lists for other compilers. I would actually fold the integer(4) type of rule into this category as well.

I also was thinking about rules for standards-compliance, where you could say you wanted code that was compliant with Fortran 2008, and then rules could identify the use of things from future standards (for instance, using intrinsics that were added in 2018/2023).