[Bug fortran/35348] New: g77 fails to warn critical real/complex assignment errors

[Bug fortran/35348] New: g77 fails to warn critical real/complex assignment errors

[ohyeahq at yahoo dot co dot jp]

Dear Fortrunners:

The following code has two very common errors, but complies without errors or
warnings: (1) a complex-to-real assignment without explicit REALPART(); (2) a
subroutine call with an erroneous real-array arg in place of a complex-arrray
arg.  

      PROGRAM TESTCMPL
      IMPLICIT NONE
      REAL R,RA(2)
      COMPLEX C
      C=COMPLEX(1.0,1.0)
      R=C
      WRITE(*,*) 'C=',C,' -> R=',R
      RA(1)=1.0
      RA(2)=2.0
      WRITE(*,*) 'CALLS  AS:',RA(1),RA(2)
      CALL ARRAY(RA)
      STOP
      END

      SUBROUTINE ARRAY(CA)
      IMPLICIT NONE
      COMPLEX CA(2)
      WRITE(*,*) 'CALLED AS:',CA(1),CA(2)
      RETURN
      END

Here is the console output:

>g77 -mno-cygwin -O -fbounds-check -ftrapv -W -Wall -Wsurprising -pedantic -o testcmpl.exe testcmpl.f

>testcmpl

 C= (1.,1.) -> R=  1.
 CALLS  AS:  1.  2.
 CALLED AS: (1.,2.) (7.72198524E+033,1.)

g77 should give errors or warnings, because these errors are very common in
real life.  I suffered for one week locating these :-)

Cheeeeeers


-- 
           Summary: g77 fails to warn critical real/complex assignment
                    errors
           Product: gcc
           Version: 3.4.4
            Status: UNCONFIRMED
          Severity: major
          Priority: P3
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: ohyeahq at yahoo dot co dot jp


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35348

[Bug fortran/35348] g77 fails to warn critical real/complex assignment errors

[fxcoudert at gcc dot gnu dot org]

------- Comment #1 from fxcoudert at gcc dot gnu dot org  2008-02-24 12:22 -------
g77 is not supported any more; starting with GCC 4.0.0, it was replaced by
gfortran. Cygwin doesn't yet offer gfortran, so you'll have to ask them to do
so and/or download unofficial binaries from
http://gcc.gnu.org/wiki/GFortranBinaries.

For your problem: I think assignment of a complex value to a real variable is
valid Fortran, and behaves as you'd expect (assigning real part of the complex
value). The second issue is real, but hard to diagnose: it should be
incorporated in gfortran starting with 4.4.0, which is under development.


-- 

fxcoudert at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |fxcoudert at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |INVALID


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35348

[Bug fortran/35348] g77 fails to warn critical real/complex assignment errors

[kargl at gcc dot gnu dot org]

------- Comment #2 from kargl at gcc dot gnu dot org  2008-02-24 17:42 -------
FX is correct with respect to assignment of a complex expression
to a real variable.  See page 141 of the F2003 standard especially
table 7.9.

real x
complex :: z = (1,-1)
x = z
end 

The 'x = z' is treated as if you wrote 'x = real(z,kind(x))'

As to the subroutine problem, your program is technically
illegal for at least one reasone.  You've never assigned
a value to CA(2), so you're not allowed to reference it.

To avoid these types of issues if make the subroutine an
internal subprogram or use an explicit interface for the
subroutine.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35348

[Bug fortran/35348] g77 fails to warn critical real/complex assignment errors

[jvdelisle at gcc dot gnu dot org]

------- Comment #3 from jvdelisle at gcc dot gnu dot org  2008-02-24 18:05 -------
If you do decide to give gfortran a try from the binary you download at the
link given in comment #1, please let us know how it works for you.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35348