Optional (empty) string argument

[Philippe Schaffnit]

Hi!

I should definitly look it up myself in the standard, but I somehow
don't seem to be able to get the meaning hidden behind the technical
wording...

So, I am definitly not saying GFortan gets it wrong, but it begs to
differs with the Intel and Sun compilers (and I would personnaly
probably try to go for the safety in numbers... plus I have to admit
that I find GFortran's behaviour slightly counter-intuitive).

If you consider the following example: an empty string is handed over
twice, the last time as an optional argument: GFortran, seems to
consider that the optional argument is not present!

Can anyone comment on this?

Thanks!

Philippe

PS: this example:
!
      PROGRAM Test_Optional_Empty_String
!
      CHARACTER ( LEN = * ), PARAMETER  ::   String = " "
!
      CALL First_Layer ( TRIM(String) )
!
      CONTAINS
          SUBROUTINE First_Layer ( Dummy )
              CHARACTER ( LEN = * ), INTENT(IN)            ::   Dummy
              CALL Second_Layer ( TRIM(Dummy) )
          END SUBROUTINE First_Layer
          SUBROUTINE Second_Layer ( Dummy )
              CHARACTER ( LEN = * ), OPTIONAL, INTENT(IN)  ::   Dummy
              IF ( PRESENT ( Dummy ) ) THEN
                  WRITE ( 6, "(//A//)" ) " Argument: |" // TRIM(Dummy) &
     &                                                            // "|"
              ELSE
                  WRITE ( 6, "(//A//)" ) " No argument:"
              END IF
          END SUBROUTINE Second_Layer
!
      END PROGRAM Test_Optional_Empty_String
!

PPS: I am using freshly updated sources (from this morning: revision
127505)