Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!wuarchive!gem.mps.ohio-state.edu!tut.cis.ohio-state.edu!purdue!mentor.cc.purdue.edu!pur-ee!pc.ecn.purdue.edu!filoj
From: filoj@pc.ecn.purdue.edu (Jeffrey J Filo)
Newsgroups: comp.lang.c++
Subject: Re: Using C++ AND C together
Message-ID: <1272@pc.ecn.purdue.edu>
Date: 14 Sep 89 15:32:03 GMT
References: <953@ncratl.Atlanta.NCR.COM> <6590251@hplsla.HP.COM> <6497@columbia.edu>
Reply-To: filoj@pc.ecn.purdue.edu.UUCP (Jeffrey J Filo)
Organization: Purdue University Engineering Computer Network
Lines: 33

In article <6497@columbia.edu> kearns@cs.columbia.edu writes:
>
>Is it possible to link a C program with a C++ library?  
>The only a priori problem I could see is that the c++ global
>constructors would not be properly inited.  


In my research in solid modeling, I have had success linking C++ 
routines with C and even FORTRAN code. 

The way I approach it is to write a number of C functions that "turn
around" and call C++ methods.  For example, if an class Curve has a 
method SelfIntersect the C calling function can be written as:
	
	int	
	SelfIntersect( Curve A)
		{ 
		A->SelfIntersect();
		}

A collection of these functions can be written for the set of methods.
This set of functions can be compiled into an interface library.  The
C and FORTRAN code calls these interface functions and accesses the 
C++ routines in this way.

One problem is that C doesn't really know about C++ objects, but 
a work-around is to treat the object pointers as integers.  This
approach is commonly used when manipulating C structures within
FORTRAN code.

------
Jeff Filo
filoj@pc.ecn.purdue.edu