unclear compile error

[Boyd_G_1]

When I compile (build) my fortran code, which compiled and ran some time ago on an IBM mainframe, I get the error:

#5507 there are multiple unnamed main program declarations in this file

and I don't know where to start to repair.  Help?

[mecej4]

The source code probably got corrupted during the transfer from the mainframe. Typically, in fixed form Fortran, if the source contains tabs or if some characters got lost, you may have a source file in which, say, the second subroutine statement in the file starts before column 7. The rest of the subroutine is seen as a main program by the compiler and, if there was a main program already scanned, the compiler would give the error message that you saw.

We could be more specific if you can show the source code.

[Boyd_G_1]

Thank you for your help.  I reviewed the file looking for any mis placed letter in col 6 - none found.  However, it turns out that I have multiple BLOCK DATA blocks - and naming them solved error.  I'm thinking the multiple BLOCK SATA's must have been interpreted as a MAIN.

So I'm moving forward!  Thank you!

[mecej4]

Normally, it is at link time that multiple definitions of code symbols (subroutines, functions) and data symbols (named blocks) are detected and reported. The Intel compiler can see that there is something wrong when two blocks with the same name (or with no names) occur in the same source file. However, the messages that it gives can be confusing! The code

      program too_many_blocks
      implicit none
      integer a,b,c, p,q,r
      common /abc/a,b,c
      common /pqr/p,q,r

      write(*,*)a,b,c
      write(*,*)p,q,r
      end program

      block data bb!B_abc
      implicit none
      integer a,b,c
      common /abc/a,b,c
      data a,b,c/10,11,12/
      end

      block data bb!B_pqr
      implicit none
      integer p,q,r
      common /pqr/p,q,r
      data p,q,r/22,21,20/
      end

results in the message

blks.f(23): error #5507: There are multiple unnamed main program declarations in this file
      end
------^
compilation aborted for blks.f (code 1)

If I name both blocks as 'bb', the message becomes

blks.f(18): error #5508: Declaration of routine 'BB' conflicts with a previous declaration
      block data bb!B_pqr

If the two block data-s are placed in separate files, at link time we see

blk2.obj : error LNK2005: __UNNAMED_DATA$ already defined in blks.obj
blks.exe : fatal error LNK1169: one or more multiply defined symbols found

It should not be too hard for the compiler to differentiate between code symbols and data symbols.