FORTRAN READ problems

[jasonbot]

Hi,

I'm having trouble with some FORTRAN code I'm trying to use.

I receive an error:

At line 769 of file ../src/mpolar.f (unit = 11, file = 'polarx.387')
Fortran runtime error: Sequential READ or WRITE not allowed after EOF marker, possibly use REWIND or BACKSPACE

Upon inspection of the code the following appears:

  55 READ(LU11,END=60) DUMMY
      GO TO 55
C
C
C---- the polar dump file doesn't exist, so write new header
   56 CONTINUE
      WRITE(LU11) NAME, ' MSES   ', VERSION
      WRITE(LU11) MACHIN, REYNIN/1.0E6, ACRIT
      WRITE(LU11) IMATYP, IRETYP
      WRITE(LU11) NBL, II
      WRITE(LU11) (ILEB(N), ITEB(N), IIB(N), N=1, NBL)
      DO 59 N=1, NBL
        WRITE(LU11) (XB(IB,N), YB(IB,N), IB=1, IIB(N))
   59 CONTINUE
C
   60 CONTINUE
C
      CLOSE(LU11)
      RETURN
      END ! PXINIT

as far as I understand the file is closing before it is written to. Upon inspection of the file system the file does get created but it is empty.

Could someone possibly explain to me what the code is trying to do and possibly suggest a solution?

I'm new to fortran code so I dont understand the syntax. Is there a good online manual for fortran?

 

[SteamKing]

Post your complete code. That may get you more suggestions. It's hard to follow what the routine is trying to do with just this snippet of code.

 

[jasonbot]

I have attached the file that is giving me problems. The file is compiled with a makefile

 

[DrClaude]

The file is open with status='unknown', such that if the file does not exist, it is created. There is then a read attempt, which in your case reaches the end of the file because it has just been created by the open statement and is empty. This is followed by a write to the file (since line 769 can only be reached if the file did not previously exist), where the program complains that it can't write since the file pointer is now beyond the EOF marker.

This is a strange behavior, probably a quirk of gcc, as a Fortran program would normally be expected to point just before the EOF marker when the EOF is reached. (I've tried with another compiler and the program executed normally.) To remedy the situation, just add the following line right after 56 CONTINUE:

REWIND(LU11)

 

[jasonbot]

The file is open with status='unknown', such that if the file does not exist, it is created. There is then a read attempt, which in your case reaches the end of the file because it has just been created by the open statement and is empty. This is followed by a write to the file (since line 769 can only be reached if the file did not previously exist), where the program complains that it can't write since the file pointer is now beyond the EOF marker.

This is a strange behavior, probably a quirk of gcc, as a Fortran program would normally be expected to point just before the EOF marker when the EOF is reached. (I've tried with another compiler and the program executed normally.) To remedy the situation, just add the following line right after 56 CONTINUE:

REWIND(LU11)

Thank you so much! It seems to be working! I guess the error shouldve been self explanatory but since I have no working knowledge of fortran I didnt know the syntax that I needed to get the programme working.

Just to add, however, I was using gfortran not gcc. I dont know if that made a difference?

As I asked before, what is a good online resource to learn fortran?

 

[DrClaude]

Just to add, however, I was using gfortran not gcc. I dont know if that made a difference?

gfortran is the Fortran "wrapper" over gcc, which links implicitly to all the necessary libraries.

As I asked before, what is a good online resource to learn fortran?

A couple of good sources I know:

Professional Programmer's Guide to Fortran77
Michael Metcalf's Fortran 90 CNL Articles