Date: Thu, 18 Aug 1994 20:46:12 -0400 (EDT)
From: Kimberley Burchett <OKRA AT max DOT tiac DOT net>
Subject: libraries
To: DJGPP Mailing List <djgpp AT sun DOT soe DOT clarkson DOT edu>


  Okay, I admit it - having come from Turbo Pascal, I'm new to "real" 
compilers.  So I can't figure this out.
  What I want to do is compile my graphics library so that it isn't 
re-compiled for every program.  I was thinking that the object file 
format was good for this - the one that ends with .o
  So I split my thing up into a header file and a source file and tried 
to compile with the "-c" option.

  Up to this point, everything is working (at least I think so).

  If I compile my .c source without putting the variables in, it doesn't 
work (which makes sense).  So what I did was include the .h file in the 
.c file to get the variables.
  That lets the .c file compile to a .o file with no problems.

  Now, if I want to use this graphics library in another program, I 
#include the .h file.  In order for that to work, however, it needs 
access to the variables again >>before<< linking so that it can assemble 
and know what sizes, etc...
  Now here's my problem.  The variables are in the .o file AND in the 
file I'm currently compiling so I get multiple definition errors at the 
linking stage.

  Does that make any sense?  I know it seems like a jumble to me...

  So let me restate the problem.  In order to compile graph.c to a .o 
file, it needs access to the variables.  In order to compile a program 
that uses graph.c, it _also_ needs access to the variables.  So I end up 
with two definitions of each variable used and, thus, no executable program.

  Can someone help straighten me out?  I'm trying to figure this out on 
my own and I just have the feeling that I have some basic 
misunderstanding somewhere. :(
							Kim