Can't link compiled fortran files together

[msgeo]

Please post your link command line (or makefile)!

As an example, the -lm option solved the problem with undefined reference in another case (in an Ubuntu Forums thread)

cc -o sqrt sqrt.c -lm

sorry but by link command line do you mean this?

gfortran -o cor69 lateralf.o qgsjet-II-03.o gheisha_2002d.o

there are a lot of make file...which one should i post here?

[sudodus]

sorry but by link command line do you mean this?

gfortran -o cor69 lateralf.o qgsjet-II-03.o gheisha_2002d.o

there are a lot of make file...which one should i post here?

If you post your makefiles for this particular program (or links to them), it might be easier to help.

I have two questions before we continue to try other things:

1. Did you try to repeat (in Ubuntu) the method that worked with fedora 2 .4.22 using gcc commands?
2. Did you try using the f2c/gcc combination according to this link?

http://valhalla.fcaglp.unlp.edu.ar/~computacion/Manuales/Online/flinux/index.html

[sudodus]

I had some time to fix your program:

1. Declaration of character variables
2. Opening of file (not necessary, otherwise the program writes to the file fort.7)
3. Added a write statement to the terminal write(*,...)

And simply used

 gfortran -o prtime prtime.f

which compiled and linked a working program for me

C-----------------------------------------------------------------------
C  PR(INT) TIME
C
C  PRINTS PRESENT DATE AND TIME AND GIVES IT IN A FORMAT SUITED FOR THE
C  RUNHEADER AND EVENTHEADER.
C  THIS SUBROUTINE IS CALLED FROM AAMAIN AND START.
C  ARGUMENT:
C   TTIME  = TIME (YYMMDD)
C
C  IF OUR DATE ROUTINE DOES NOT FIT TO YOUR COMPUTER, PLEASE REPLACE
C  IT BY A SUITABLE ROUTINE OF YOUR SYSTEM
C  FOR COMPILERS WITH NEWER DATE FUNCTIONS, INCLUDING DEC UNIX f77
C  AND RECENT GNU g77 >0.5.21 (egcs 1.1.x, gcc 2.95, ...)
C  IF YOR COMPUTER DOES NOT KNOW SUBROUT. DATE_AND_TIME
C  REPLACE THIS CALL BY A CALL TO YOUR SYSTEM ROUTINES TO
C  FILL THE INTEGERS: IYEAR, MONTH, IDAY, IHOUR, IMINU, ISEC

      character*8 YYYYMMDD
      character*6 HHMMSS

      open(7,file="printed-time.txt")

      CALL DATE_AND_TIME( YYYYMMDD, HHMMSS )
      READ(YYYYMMDD,'(I4,2I2)') IYEAR,MONTH,IDAY
      READ(HHMMSS,'(3I2)') IHOUR,IMINU,ISEC
      write(7,100) IDAY,MONTH,IYEAR,IHOUR,IMINU,ISEC
      write(*,100) IDAY,MONTH,IYEAR,IHOUR,IMINU,ISEC
      TTIME = MOD(IYEAR,100)*10000 + MONTH*100 + IDAY
 100  FORMAT(' PRESENT TIME : ',I2.2,'.',I2.2,'.',I4,I4.2,':',I2.2,
     *       ':',I2.2)

      RETURN
      END

[msgeo]

I had some time to fix your program:

1. Declaration of character variables
2. Opening of file (not necessary, otherwise the program writes to the file fort.7)
3. Added a write statement to the terminal write(*,...)

And simply used

 gfortran -o prtime prtime.f

which compiled and linked a working program for me

Thank you so much for the time you spent on this.
but the problem seems to be unsolved.


first to your previous question I should say that I used gcc/f2c combination but that one also gives some errors like this:

 
 
/tmp/fort77-29566-1.c:17864:12: error: conflicting types for ‘fctcos_’
/tmp/fort77-29566-1.c:17302:33: note: previous declaration of ‘fctcos_’ was here
/tmp/fort77-29566-1.c:26703:22: error: conflicting types for ‘rtmi_’
/tmp/fort77-29566-1.c:17296:33: note: previous declaration of ‘rtmi_’ was here
/usr/bin/f77: aborting compilation

and about your correction on the PRTIME subroutine I should say that this a only a small part of the program and this is not my program.
this a scientific program for Simulating cosmic rays.
that one also comes with some errors:

 
      character*8 YYYYMMDD                                              
                          1
Error: Symbol 'yyyymmdd' at (1) already has basic type of CHARACTER
lateralf.f:19183.24:
 
      character*6 HHMMSS                                                
                        1
Error: Symbol 'hhmmss' at (1) already has basic type of CHARACTER
lateralf.f:35547.10:

here is the original file:

http://sdrv.ms/Xwz3il

and here is some snapshots of the program directory:
http://sdrv.ms/VRokm2
http://sdrv.ms/Uilgz3

also i mentioned earlier that I'm able to compile files with gcc just like fedora but when I try to link files it comes with some errors!

[sudodus]

first to your previous question I should say that I used gcc/f2c combination but that one also gives some errors like this:

 
/tmp/fort77-29566-1.c:17864:12: error: conflicting types for ‘fctcos_’
/tmp/fort77-29566-1.c:17302:33: note: previous declaration of ‘fctcos_’ was here
/tmp/fort77-29566-1.c:26703:22: error: conflicting types for ‘rtmi_’
/tmp/fort77-29566-1.c:17296:33: note: previous declaration of ‘rtmi_’ was here
/usr/bin/f77: aborting compilation

I see. That did not help

Probably you had some other compiler or linker version or a library available in Fedora, that you don't have (yet) in Ubuntu.

and about your correction on the PRTIME subroutine I should say that this a only a small part of the program and this is not my program.
this a scientific program for Simulating cosmic rays.
that one also comes with some errors:

 
      character*8 YYYYMMDD                                              
                          1
Error: Symbol 'yyyymmdd' at (1) already has basic type of CHARACTER
lateralf.f:19183.24:
 
      character*6 HHMMSS                                                
                        1
Error: Symbol 'hhmmss' at (1) already has basic type of CHARACTER
lateralf.f:35547.10:

Well, I had only the snippet you posted, and I needed declarations of the character variables to make the compiler happy. Obviously there are some kind of include statements or global statements, that declare the character variables.

But I would like to know if you can create a working program of the fortran code in post #13 using the command line in the same post. That will show if your compiler and linker have enough 'horsepower' to do that (for example if there is such a time function or subroutine).

I am running Ubuntu 12.04 LTS 32-bit with pae

uname -a
Linux ssd-grund 3.2.0-36-generic-pae #57-Ubuntu SMP Tue Jan 8 22:01:06 UTC 2013 i686 i686 i386 GNU/Linux

here is the original file:

http://sdrv.ms/Xwz3il

and here is some snapshots of the program directory:
http://sdrv.ms/VRokm2
http://sdrv.ms/Uilgz3

also i mentioned earlier that I'm able to compile files with gcc just like fedora but when I try to link files it comes with some errors!

I'll look at that later.

[sudodus]

sorry but by link command line do you mean this?

gfortran -o cor69 lateralf.o qgsjet-II-03.o gheisha_2002d.o

there are a lot of make file...which one should i post here?

The .o files are compiled object code files. It might be enough, but you might need some kind of reference to the linker or some libraries (if more than the default ones are needed).

This could be possible to see in one of the makefiles.
--
With the snippet in post #13 it was possible for me to compile to an object file with

gfortran -c prtime.f

and to link with

gfortran -o prtime-from-o prtime.o

It created a file identical to the one created directly from the fortran source code file. So yes, it could be possible to link with the command you suggested, at least concerning the prtime subroutine.

[sudodus]

...

here is the original file:

http://sdrv.ms/Xwz3il

and here is some snapshots of the program directory:
http://sdrv.ms/VRokm2
http://sdrv.ms/Uilgz3

also i mentioned earlier that I'm able to compile files with gcc just like fedora but when I try to link files it comes with some errors!

I looked at it. Yes, I could see the whole time subroutine

      SUBROUTINE PRTIME( TTIME )

with declaration of variables, that were not there in the snippet.

I also had a second look at the command line in post #1

gfortran -o file1.o file2.o file3.o -o outputfilename

I think you'd make the compiler/linker confused with the two -o options. First you tell it to write the final program to file1.o and then to outputfilename. The correct syntax should be

gfortran -o outputfilename file1.o file2.o file3.o

The command line in post #11 looks correct.

gfortran -o cor69 lateralf.o qgsjet-II-03.o gheisha_2002d.o

and I guess this is the real thing

If you still have problems, maybe you should check, if you have double definitions or calls to the time subroutine, that there is another one, that cannot be resolved.

What about this one?

      INTEGER          ILEFTA,ILEFTB,TIME
      EXTERNAL         TIME
...
C  TIME AT BEGINNING

      ILEFTA = TIME()
...
...
C  TIME SINCE BEGINNING

      ILEFTB = TIME()

      TDIFF  = ILEFTB - ILEFTA

These are not calls to or via PRTIME

[msgeo]

But I would like to know if you can create a working program of the fortran code in post #13 using the command line in the same post.

yes.that file compiled and linked successfully with gfortran.

[msgeo]

I'm really grateful for your help...
this ""time"" function isn't really important but i don't know how to get rid of it!

just downloaded the fedora 18 to see if i can compile and link files on that

[sudodus]

yes.that file compiled and linked successfully with gfortran.

Then I think the problem is the call of

TIME()

So you should make a function or subroutine instead of that to use the same time call as in PRTIME,

CALL DATE_AND_TIME( YYYYMMDD, HHMMSS )

or some other call to the system time that is supported, and use the output of that instead of the external TIME() call. You should also check what unit it is supposed to be (seconds?)