g77 bug

[Tony Willis]

Hello:

I am using the following version of g77 from egcs 1.0.3

g77 -v
Reading specs from /usr/local/lib/gcc-lib/i686-pc-linux-gnu/egcs-2.90.29/specs
gcc version egcs-2.90.29 980515 (egcs-1.0.3 release)    

to compile some really ugly code adapted from the fftpak library for
doing ffts.

If I compile the following piece of code (cut out from the original
and pared down to a bare minumum) with
g77 -c -fugly

      SUBROUTINE DEFFT1 (N,WA,IFAC)
      DOUBLE PRECISION WA(1), ARG1, ARGH, CH1, CH1H, DCH1, DSH1, SH1,
     1     TPI
      INTEGER IFAC(1), NTRYH(4)
      DATA NTRYH(1), NTRYH(2), NTRYH(3), NTRYH(4) /4, 2, 3, 5/
      DATA TPI   /  6.2831853071 7958647692 5286766559 00577D0/
C
      NL = N
      RETURN
      END

      SUBROUTINE DEFFTI (N,WSAVE)
      DOUBLE PRECISION WSAVE(*)
C
      IF (N .EQ. 1) RETURN
C
      CALL DEFFT1 (N,WSAVE(2*N+1),WSAVE(3*N+1))
C
      RETURN
      END

I get the following messages, which treat the difference in
type of variables  WSAVE and IFAC as an error:

f771: warning: -fugly is overloaded with meanings and likely to be removed;
f771: warning: use only the specific -fugly-* options you need
temp.f: In subroutine `deffti':
temp.f:2: 
         SUBROUTINE DEFFT1 (N,WA,IFAC)
                    1
temp.f:18: (continued):
         CALL DEFFT1 (N,WSAVE(2*N+1),WSAVE(3*N+1))
              2
Argument #3 (named `ifac') of `defft1' is one type at (2) but is some other type at (1) [info -f g77 M GLOBALS]

Hoever if I reverse the order and put SUBROUTINE DEFFTI first in the code:

      SUBROUTINE DEFFTI (N,WSAVE)
      DOUBLE PRECISION WSAVE(*)
C
      IF (N .EQ. 1) RETURN
C
      CALL DEFFT1 (N,WSAVE(2*N+1),WSAVE(3*N+1))
C
      RETURN
      END

      SUBROUTINE DEFFT1 (N,WA,IFAC)
      DOUBLE PRECISION WA(1), ARG1, ARGH, CH1, CH1H, DCH1, DSH1, SH1,
     1     TPI
      INTEGER IFAC(1), NTRYH(4)
      DATA NTRYH(1), NTRYH(2), NTRYH(3), NTRYH(4) /4, 2, 3, 5/
      DATA TPI   /  6.2831853071 7958647692 5286766559 00577D0/
C
      NL = N
      RETURN
      END

and then compile I get only a warning, and the compiler produces a .o
file, as it should, since this really ugly construct was allowed in f66.

f771: warning: -fugly is overloaded with meanings and likely to be removed;
f771: warning: use only the specific -fugly-* options you need
temp.f: In subroutine `defft1':
temp.f:6: warning:
         CALL DEFFT1 (N,WSAVE(2*N+1),WSAVE(3*N+1))
              1
temp.f:11: (continued):
         SUBROUTINE DEFFT1 (N,WA,IFAC)
                    2
Argument #3 (named `ifac') of `defft1' is one type at (2) but is some other type at (1) [info -f g77 M GLOBALS]

According to me g77 should have produced just a warning in both these
cases.

Cheers,

Tony
-- 
Tony Willis

Internet  :   Tony.Willis@hia.nrc.ca    obsolete: twillis@drao.nrc.ca
Snailnet  :   Dominion Radio Astrophysical Observatory
              P.O. Box 248, Penticton, BC, Canada V2A 6K3
BC Tel net:   (250) 493-2277    Faxnet    :   (250) 493-7767
voicemailnet: (250) 490-4343    Localnet  :   ext 343