Welcome to the wonderful world of HP-C.  This problem has been on-going for
years and many versions.  I first ran across this _feature_ when I sent
programs compiled on 3.1 to 2.0-2.2 systems.  And yes, you guessed it there
was a new module which did not exist on 2.x.  If memory serves it was
X11_init_globals or some such silly thing.  What I did was create a dummy
routine called X11_init_globals and included it in the link command.
 
But if you want to see how crazy it gets, read on.
 
I sell a utility which uses IMAGE intrinsics and SQL commands.  When I
compile the C code, I must pass it through the C SQL preprocessor, then
compile the output.  The pre-processor introduces new externals on 5.0 which
are not found on 4.0.  Unfortunately, I cannot fake this one out.  I must
therefore keep two sets of compiled code for my customers.  One for 4.0 and
the other for 5.0  What is even more fun, is that since IMAGE/SQL and
ALLBASE/SQL are decoupled from the operating system, I can encounter what I
call 5.0 IMAGE/ALLBASE on a 4.0 machine.  Wait, the fun is not over yet.
 
Now with the 5.0 C compiler, there are new routines which are not found on
4.0, so even a 5.0 program with INAGE/ALLBASE for 5.0 but needed for 4.0
cannot run on 4.0.  Wait there is more.  The ATCINFO file format is different
on the ...   but you get the point.
 
It is a character building experience.
 
 
Kind regards,
 
Denys. . .