XEXP fails in Gfortran 64 bit

[oph]

I was updating a program that uses EXP LABELS for Y aXIS. I tried XEXP and FEXP, and as XEXP look nice, I adopted it. All with Gfortran 32 bit.

But when I tried Gfortran 64 bit, the screen is dark (no black) and the program aborts.
 
64 bit is required for sobe extra big files to plot.

Regards,
OPH. 2023-06-26 09:20

[Helmut Michels]

Hi,

I have made some tests with a simple axis scaling, but they do not reproduce the  XEXP and FEXP problem for GFortran 64-bit.

Can you post a short code with just an axis system that shows the program abort?

With best regards,

Helmut

[oph]

I tried but could not reproduce the error in a simple example. In the routine the
error still occurs, and as I said, FEXP works fine.

In my program I use TEXMOD to allow the user units such as Ohm (Omega) or
micro (mu). I thought that this might interfere with the multiplication sign, but
the problem is not there

I could not reproduce the error in simple case nor avoid it in the working
program...

For now I'll keep using EXP with the asterisk, meanwhile I'll keep trying to find the bug,
to find the error.

Regards,

OPH. 2023-06-26 17:38

[oph]

Well, I found were the program failed.

C
C TEST OF XEXP
C
      IMPLICIT REAL*8 (A-H,O-Z)
      CHARACTER*5 CAX,CAY,CAZ
      IXPOS=100
      IYPOS=2000
      IXLEN=2000
      CALL SETPAG ('DA4L')
      CALL METAFL ('XWIN')
      CALL DISINI
      CALL LABELS ('FLOAT','X')
C      CALL LABELS ('XEXP','Y')
      CALL LABELS ('FEXP','Y')
      CALL LABELS ('FLOAT','Z')
      CALL LABDIG(-1,'Y')
      CALL GETLAB(CAX,CAY,CAZ)
      WRITE(*,*)CAX,CAY,CAZ
      XINI=0.00D0
      XFIN=1.00D0
      REDX= 0.10D0
      YINI=0.00D0
      YFIN=1.0D8
      REDY=0.1D8
      CALL GRAF(XINI,XFIN,XINI,REDX,YINI,YFIN,YINI,REDY)
      CALL RLINE(XINI,YINI,XFIN,YFIN)
      CALL DISFIN
      STOP
      END

The CALL GETLAB (CAX,CAY,CAZ) fails for 'XEXP'. For the rest of possible formats it works,

but with 'XEXP' the program aborts with a error message (Intel compiler). Gfortran simply aborts without messages.

I suppresed  this CALL GETLAB, and the nice XEXP format for labels is OK.

Regards,

OPH. 2023-06-26 20:15

[Helmut Michels]

Hi,

thank you for the example code. I could reproduce the problem and will let you know when I have

corrected it.

With best regards,

Helmut

[Helmut Michels]

Hi,

yes, the bug happens in the routine GETLAB. The option 'XEXP' is added later to LABELS without updating

GETLAB, so that an index overflow caused the crash. Thank you for the hint and the example code.

I will fix the bug with the next update. A workaround is not to use GETLAB for the option 'XEXP', which you have

already done.  Since GETLAB is just a requesting routine, it's not absolutely necessary.

With best regards,

Helmut