[Bug fortran/56675] New: I/O: Check edit descriptors with READ/WRITE used in FORMAT statements

["burnus at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56675

             Bug #: 56675
           Summary: I/O: Check edit descriptors with READ/WRITE used in
                    FORMAT statements
    Classification: Unclassified
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Keywords: diagnostic
          Severity: normal
          Priority: P3
         Component: fortran
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: burnus@gcc.gnu.org


https://groups.google.com/forum/#!topic/comp.lang.fortran/BxWaYSkmV2w

The following code shows that the special checks for READ/WRITE/PRINT do not
get checked for with FORMAT but only with a string.

Seemingly, the format is checked when FORMAT is encountered - one should
consider to do additionally a check if READ/WRITE/PRINT is used.

[The latter becomes even more useful when looking at the types of the arguments
(e.g. "a" edit descriptor with a "real" expression/variable). However,
currently this is only checked at run time. For compile time, it should be
doable if the I/O list doesn't contain nonconst-shape arrays and no
io-do-lists.]




  implicit none
  integer :: z

! OK: Diagnosed.  (Side note: Also diagnosed by NAG.)
!  character(len=*), parameter :: str = '(i0)'
!  read  (*,str) z ! Positive width required in format string
!  read  (*,"(i0)") z ! "Positive width required in format string"

! Not diagnosed but diagnosable. (Also not diagnosed by NAG.)
  read  (*, 99) z ! 'Wrongly' accepted (at compile time)
  99 FORMAT (i0)  ! (gives error at run time)

! RFC: Validly rejected or not? (Note: Also rejected by NAG)
!   read (*, 100) z   ! Rejected at run time (but not used;
!   100 FORMAT (i2,i0)! may it reject it or not?)
end