Linking multiple main programs in one solution

[Sudarshan_R_]

Hello,

I am working on recompiling a really old fortran code into a 32 bit application. It has multiple 

subroutines, multiple main programs and uses many text files as reference for data. I am not familiar on 

how to recompile them.

For the sake of simplicity, lets assume the following:

Main Programs: MAIN.FOR, A.FOR and B.FOR

Subroutines: SUB1.FOR, SUB2.FOR

Data Files: FILE1, FILE2

The MAKEFILE in the old folder reads like this:

***************************************************************

.SUFFIXES: .for

.for.obj:

 fl /Gt9 $(fflags) >>prog.log

all: log prog.exe

log: del prog.log

prog.exe: MAIN.obj \

 SUB1.obj \

 SUB2.obj \

 A.obj \

 B.obj

 \makefile

 link @prog.rsp >>prog.log

A: A.obj

 link @A.rsp >>prog.log

B: B.obj

 link @B.rsp >>prog.log

*********************************************************************

To do this in visual fortran, I made three different projects under one solution and made project 

PROG (startup project containing MAIN.FOR and subroutines) dependent on project A(with A.FOR and 

subroutines) and project B(with B.FOR and subroutines). I included FILE1 and FILE2 in solution items 

folder and built the solution. However, this doesn't seem to achieve the required objective of linking the 

A and B with MAIN. I am relatively new to VS and would be so grateful if anybody could help me with 

this situation.

[ZlamalJakub]

Create only one project with all *. for files included.

Lines

prog.exe: MAIN.obj \

 SUB1.obj \

 SUB2.obj \

 A.obj \

 B.obj

in makefile mean that prog.exe depends on compiled MAIN, SUB1, SUB2, A and B source files.

[Sudarshan_R_]

That wouldn't work and i would get an error like this

error LNK2005: _MAIN__ already defined in A.obj

error LNK2005: _MAIN__ already defined in B.obj

I have provided the makefile of the old code so that you could get an idea of the order/technique of compilation. I dont want to make another makefile. I would prefer to compile using visual studio.

[jimdempseyatthecove]

On your old platform, how did execution flow pass from MAIN to A and B?

If via CALL, then you wil be required to change "PROGRAM" in A and B to "SUBROUTINE" (without arguments), and if END contained "END PROGRAM" change to "END SUBROUTINE".

If you were on a really old system, then MAIN, A, B would look like a card stack, where A runs after MAIN and B runs after A. For this you would likely write a batch file

: YourBatchFile.bat
MAIN
A
B

Jim Dempsey

[Steven_L_Intel1]

I suspect that these old sources were actually multiple programs that shared some sources. Multiple main programs don't work in any Fortran implementation I have known over 35 years.

[TimP]

As Steve said, multiple main programs never have been allowed to link together.  However, various typos when resurrecting old code may cause the compiler to see multiple main programs.

[Sudarshan_R_]

Thank you for your replies. I had earlier misunderstood the structure of the code. It basically creates multiple applications utilizing the same subroutines.