Re: [Swig-user] Python proxy classes with %import

[Abhinandan J. <ja...@he...>]

William

Thanks for the prompt response. Here is a small example that
illustrates the problem.

A.h file
========

class AA {

};

B.h file
========

#include "A.h"

class BB : public AA {

};

B.i file
========

%module B_py

// using 1 causes %import to used, while 0 enables %include
#if 1
%import "A.h"
#else
%include "A.h"
#endif

%include "B.h"

----------------  end of file contents  ----------------

Generate swig wrappers using 

	swig -c++ -python B.i

twice: once with #if 1 in B.i as is (to use %import) and next with it
changed to #if 0 (to use %include). Do a diff between the B_py.py
files generated. The first one contains the line

   class BB(_object):  
     etc.

while the second one contains (correctly) the line

   class BB(AA):
     etc.

I believe the first case (with #if 1) should also be generating the
latter output for the BB proxy inheritance. Without this, there is no
way for the BB proxy class to get to superclass AA's methods.

I am using swig 1.3.31.

Thanks

Abhi

On Sun, Sep 30, 2007 at 07:53:18PM +0100, William S Fulton wrote:
> Abhinandan Jain wrote:
> > I an having trouble getting Python proxy classes to work when using
> > the %import directive. I have a C++ class X and another Y derived from
> > X. I am using SWIG to build an extension for class X using an X.i
> > file. In Y.i, if I use the
> > 
> > 	%include X.i
> > 
> > directive, then the resulting Python proxy class Y shows up as derived
> > from class X's Python proxy class. However this also leads to
> > duplicate wrapper code for some methods being generated as well. My
> > understanding is that the correct way to avoid this is to use instead
> > the
> > 
> > 	%import X.i
> > 
> > directive in Y.i. This does indeed get rid of the duplicative
> > code. However, the Y's Python proxy class no longer inherits from X's
> > Python proxy class, but instead simply from the 'object' class. This
> > means that I am no longer able to access any of X's methods from an
> > instance of Y in Python.
> > 
> > I suspect that I am either not using the %import directive correctly
> > or am misunderstanding its purpose. Any help/advise on this problem
> > would be much appreciated.
> > 
> > Thanks in advance. I am using pretty recent version of SWIG from the
> > repository.
> > 
> 
> You aren't doing anything wrong from what you have described. Have you 
> looked carefully at the warnings? SWIG usually warns if it hasn't parsed 
> the base class.
> 
> What is happening sounds a bit odd and possibly you have another 
> %include "X.i" somewhere... you should never mix %import and %include of 
> the same file. You'll need to cut the code down to a small example in 
> order for us to help any further. If you are using the SVN version of 
> SWIG, do you get the same problem using the latest release (1.3.31)?
> 
> William
> 
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2005.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> Swig-user mailing list
> Swi...@li...
> https://lists.sourceforge.net/lists/listinfo/swig-user

-- 
Abhinandan Jain, Ph.D.
Dynamics and Real-Time Simulation (DARTS) Lab
Mobilty and Robotic Systems Section
Jet Propulsion Laboratory
4800 Oak Grove Drive, Pasadena, CA 91109
(818)-354-3412	  (818)-393-5007 (fax)