From rwgk at yahoo.com  Sat Apr  1 01:16:18 2006
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Fri, 31 Mar 2006 15:16:18 -0800 (PST)
Subject: [C++-sig] Problem with Templated class and Boost.python
In-Reply-To: <c2c16f990603311152y593fb82fw56ed53dc04a53dad@mail.gmail.com>
Message-ID: <20060331231619.73542.qmail@web31507.mail.mud.yahoo.com>

--- Arrakis Muadib <marrakis at gmail.com> wrote:
> I'm trying to create a binding for a small c++ class. I decided to
> created directly with Boost.Python and did'nt use Pyste cause it's not
> supported on my distro and I cannot make gccxml working.
> 
> Now I'm confronted with a class having a template parameter.
> 
> The signature is something like this:
> 
> template<typename T>
> class A : virtual public B<T>
> {
>     T methodX() {...}
>     ...
> };
> 
> How can I map the template parameter to the boost.python class definition?

If you have just one A<T> to wrap, simply do it like this:

  class_<A<int> >("A_int")
    .def("method_a", &A<int>::method_a)
  ;

If you have many, write a helper struct like this:

  template <typename T>
  struct A_wrapper
  {
    static void
    wrap(const char* python_name)
    {
      class_<A<T> >(python_name)
        .def("method_a", &A<T>::method_a)
        // etc.
      ;
    }
  };

Later:

  A_wrapper<int>::wrap("A_int");
  A_wrapper<long>::wrap("A_long");
  // etc.

Here is a complete example:

 
http://cvs.sourceforge.net/viewcvs.py/cctbx/scitbx/include/scitbx/stl/vector_wrapper.h?view=markup

http://cvs.sourceforge.net/viewcvs.py/cctbx/scitbx/include/scitbx/stl/vector_ext.cpp?view=markup

Cheers,
        Ralf


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 


From DRandallTaylor at yahoo.com  Sat Apr  1 19:31:49 2006
From: DRandallTaylor at yahoo.com (DRandallTaylor)
Date: Sat, 1 Apr 2006 17:31:49 +0000 (UTC)
Subject: [C++-sig] how to disambiguate 2 overloaded mem funs each with just
	1 arg
Message-ID: <loom.20060401T191818-143@post.gmane.org>

Hello,
N00bie here.
How do I disambiguate 2 member functions each having just 1 arg:

struct C
{
   int len(const char* const val) { return ::strlen(val); }
   int len(const std::string &val) { return val.size(); }
};
BOOST_PYTHON_MODULE(pycore)
{
  boost::python::class_<C>("C")
   .def("len"
        , &C::len
        , boost::python::return_value_policy< python::return_by_value >()
   );
}

There must be some way to tell boost::python that when I pass a python str 
object to C::len then convert the arg to a "const char* const" and call the 
first overload of C::len.  (I told you this was going to be easy! But not for 
me because I'm n00b.).

Cheers,
DRandallTaylor










From dave at boost-consulting.com  Sat Apr  1 20:50:43 2006
From: dave at boost-consulting.com (David Abrahams)
Date: Sat, 01 Apr 2006 10:50:43 -0800
Subject: [C++-sig] how to disambiguate 2 overloaded mem funs each with
	just 1 arg
References: <loom.20060401T191818-143@post.gmane.org>
Message-ID: <ubqvlxh8c.fsf@boost-consulting.com>

DRandallTaylor <DRandallTaylor at yahoo.com> writes:

> Hello,
> N00bie here.
> How do I disambiguate 2 member functions each having just 1 arg:
>
> struct C
> {
>    int len(const char* const val) { return ::strlen(val); }
>    int len(const std::string &val) { return val.size(); }
> };
> BOOST_PYTHON_MODULE(pycore)
> {
>   boost::python::class_<C>("C")
>    .def("len"
>         , &C::len
>         , boost::python::return_value_policy< python::return_by_value >()
>    );
> }
>
> There must be some way to tell boost::python that when I pass a python str 
> object to C::len then convert the arg to a "const char* const" and call the 
> first overload of C::len.  (I told you this was going to be easy! But not for 
> me because I'm n00b.).


  boost::python::class_<C>("C")
   .def("len"
        , (int (C::*)(char const*))&C::len
        , boost::python::return_value_policy< python::return_by_value >()
   );

HTH,
-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com



From DRandallTaylor at yahoo.com  Sat Apr  1 23:12:16 2006
From: DRandallTaylor at yahoo.com (DRandallTaylor)
Date: Sat, 1 Apr 2006 21:12:16 +0000 (UTC)
Subject: [C++-sig] python override of C++ virtual mem fun not exactly
	working yet
Message-ID: <loom.20060401T230825-688@post.gmane.org>

Hello again.  n00b here.
Simple problem, but hard for me, cause I'm n00b.

My Python class derives from a C++ class and overrides a virtual mem fun that 
has an implementation.  In the derived Python class, I need "self" to be 
treated polymorphically with respect to the C++ base class.  The boost::python 
Tutorial gives a sample but the derivation shown of BaseWrap is private, not 
public derivation:  "struct BaseWrap : Base, wrapper<Base>".  I know this is a 
simple problem... but having trouble with it...

My error msg tells me that my 2nd arg is a InboxListener not a Callback.


  File "test.py", line 9, in __init__
    self.inbox = sess.CreateInbox(self)
ArgumentError: Python argument types in
    Session.CreateInbox(Session, InboxListener)
did not match C++ signature:
    CreateInbox(class Session {lvalue}, class Callback {lvalue})


My C++ stuff:


class CallbackWrap : public Callback, python::wrapper<Callback>
{  public:
    void OnData() // iff public derivation used, then this is virtual; else 
non-virtual
    {
        if (python::override the_override = this->get_override("OnData"))
            the_override();
        else  Callback::OnData();
    }
    void default_OnData() { this->Callback::OnData(); }    
};


python::class_ < CallbackWrap, boost::noncopyable > ("Callback")
.def("OnData", &Callback::OnData, &CallbackWrap::default_OnData);


python::class_<Session>("Session")
.def("CreateInbox", &Session::CreateInbox, python::return_value_policy< 
python::return_by_value >());


And my Python stuff:


class InboxListener(Callback):
	def __init__(self, session):
		self.inbox = session.CreateInbox(self)
	def OnData():
		print "OnData called"
		return


Seems like "self" passed to  session.CreateInbox should be treated as a 
Callback type, since I've used public derivatation.  Can you help me spot the 
error of my ways?  And I thank you!

DRandallTaylor











From seefeld at sympatico.ca  Sat Apr  1 23:29:26 2006
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Sat, 01 Apr 2006 16:29:26 -0500
Subject: [C++-sig] python override of C++ virtual mem fun not exactly
 working yet
In-Reply-To: <loom.20060401T230825-688@post.gmane.org>
References: <loom.20060401T230825-688@post.gmane.org>
Message-ID: <442EF0B6.9000506@sympatico.ca>

DRandallTaylor wrote:
> Hello again.  n00b here.
> Simple problem, but hard for me, cause I'm n00b.
> 
> My Python class derives from a C++ class and overrides a virtual mem fun that 
> has an implementation.  In the derived Python class, I need "self" to be 
> treated polymorphically with respect to the C++ base class.  The boost::python 
> Tutorial gives a sample but the derivation shown of BaseWrap is private, not 
> public derivation:  "struct BaseWrap : Base, wrapper<Base>".  I know this is a 
> simple problem... but having trouble with it...

No it isn't. The default access-specifier for struct is public, not private.


> 
> My error msg tells me that my 2nd arg is a InboxListener not a Callback.
> 
> 
>   File "test.py", line 9, in __init__
>     self.inbox = sess.CreateInbox(self)
> ArgumentError: Python argument types in
>     Session.CreateInbox(Session, InboxListener)
> did not match C++ signature:
>     CreateInbox(class Session {lvalue}, class Callback {lvalue})

[...]


> And my Python stuff:
> 
> 
> class InboxListener(Callback):
> 	def __init__(self, session):
> 		self.inbox = session.CreateInbox(self)

In python you have to invoke the base class constructor explicitely in order
to initialize it. I believe this will fix your problem.

HTH,
		Stefan


From DRandallTaylor at yahoo.com  Sat Apr  1 23:45:25 2006
From: DRandallTaylor at yahoo.com (DRandallTaylor)
Date: Sat, 1 Apr 2006 21:45:25 +0000 (UTC)
Subject: [C++-sig] python override of C++ virtual mem fun not exactly
	working yet
References: <loom.20060401T230825-688@post.gmane.org>
	<442EF0B6.9000506@sympatico.ca>
Message-ID: <loom.20060401T233833-671@post.gmane.org>

I've added the invocation of the base class constructor (thanks Stefan):

class InboxListener(Callback):
	def __init__(self, sess):
		Callback.__init__(self)
		self.inbox = sess.CreateInbox(self)  # ArgumentError here
	def OnData():
		print "OnData called"

But I still get this:


ArgumentError: Python argument types in
    VIAmSession.CreateInbox(Session, InboxListener)
did not match C++ signature:
    CreateInbox(class Session {lvalue}, class Callback {lvalue})


Python does not yet know that my InboxListener is derived from Callback.  How 
shall I tell her?

Cheers.








From DRandallTaylor at yahoo.com  Sun Apr  2 02:44:27 2006
From: DRandallTaylor at yahoo.com (DRandallTaylor)
Date: Sun, 2 Apr 2006 00:44:27 +0000 (UTC)
Subject: [C++-sig]
	=?utf-8?q?python_override_of_C++_virtual_mem_fun_not_ex?=
	=?utf-8?q?actly=09working_yet?=
References: <loom.20060401T230825-688@post.gmane.org>
	<442EF0B6.9000506@sympatico.ca>
	<loom.20060401T233833-671@post.gmane.org>
Message-ID: <loom.20060402T024332-32@post.gmane.org>

DRandallTaylor <DRandallTaylor <at> yahoo.com> writes:

> [...]

May I rephrase?

Is conversion from a Python class to the boost::python type that it is derived 
from automatic?  I think it probably is, therefore I must be doing something 
wrong:


class InboxListener(VIAmCallback):
	def __init__(self, sess):
		VIAmCallback.__init__(self)
		self.inbox = sess.CreateInbox(self)


gives me:


  File "test.py", line 32, in run
    listener = InboxListener(sess)
  File "test.py", line 10, in __init__
    self.inbox = sess.CreateInbox(self)
ArgumentError: Python argument types in
    VIAmSession.CreateInbox(VIAmSession, InboxListener)
did not match C++ signature:
    CreateInbox(class tt_stream_ns::VIAmSession {lvalue}, class 
tt_stream_ns::VIAmCallback {lvalue})



And my C++ looked like this:


struct VIAmCallbackPublic : public tt_stream_ns::VIAmCallback
{
   [...]
};
struct VIAmCallbackWrap : public VIAmCallbackPublic, 
python::wrapper<VIAmCallbackPublic>
{
   [...]
]

python::class_ <VIAmCallbackWrap, boost::noncopyable>("VIAmCallback")
        .def("OnData", &VIAmCallbackPublic::OnData,
              &VIAmCallbackWrap::default_OnData);



And I thought that the Python error message above tells me that python does 
NOT believe that InboxListener is actually a tt_stream_ns::VIAmCallback.  So I 
experimented by added a python::bases<> template template arg like this:


python::class_ <
  VIAmCallbackWrap
  , python::bases<VIAmCallbackPublic, tt_stream_ns::VIAmCallback>
  , boost::noncopyable
> ("VIAmCallback")
def("OnData", &VIAmCallbackPublic::OnData,
    &VIAmCallbackWrap::default_OnData);


And then I get a worse error from Python:


Traceback (most recent call last):
  File "test.py", line 1, in ?
    from pycore import *
RuntimeError: extension class wrapper for base class struct VIAmCallbackPublic 
has not been created yet



Am I completely clueless or simply overlooking something.  Please tell me it 
is the latter.


Sincerely,
Your Faithful n00b.

DRandallTaylor




From roman.yakovenko at gmail.com  Sun Apr  2 06:25:08 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Sun, 2 Apr 2006 06:25:08 +0200
Subject: [C++-sig] C++ compile error on .def from n00b.
In-Reply-To: <loom.20060331T082351-967@post.gmane.org>
References: <loom.20060331T082351-967@post.gmane.org>
Message-ID: <7465b6170604012025j17de3ca5qbcbb8a422c732004@mail.gmail.com>

On 3/31/06, Randy Taylor <Randy.Taylor at tradingtechnologies.com> wrote:
> I'm a n00b.
> I'm trying to expose a C++ class that has a virtual mem fun with an
> implementation that I wish to override in Python.  Seems easy enough.
> Compiler disagrees, saying she does not like my def. I must be doing something
> wrong. What is it?  Thanks kindly in advance.
>

Hi. It would be nice if you could provide less source that shows your problem.

Also I think in this case ( noncopyable ) you did not setup class init
method(s).

In order to get more readable errors you can write something like this

typedef boost::python::class_< ... >  my_class_exposer_t;
my_class_exposer_t exposer( ... );
exposer.def( ... );

At least you will be pointed by compiler to the right line.

And an other thing: if you are just start with Boost.Python then try
one of available code
generators: pyplusplus ( has gui ) and/or Pyste. You will be able to
save a lot of time.


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From achim-bpl at mol-net.com  Mon Apr  3 11:11:27 2006
From: achim-bpl at mol-net.com (Achim H.)
Date: Mon, 3 Apr 2006 11:11:27 +0200
Subject: [C++-sig]
	=?iso-8859-1?q?How_to_deep-copy_and_delete_a_mixed_pyth?=
	=?iso-8859-1?q?on/C++_object=09from_C++?=
In-Reply-To: <200603311757.k2VHvFgj015725@libra3.slac.stanford.edu>
References: <200603311048.02242.achim-bpl@mol-net.com>
	<200603311757.k2VHvFgj015725@libra3.slac.stanford.edu>
Message-ID: <200604031111.27888.achim-bpl@mol-net.com>

Am Freitag, 31. M?rz 2006 19:57 schrieb Paul F. Kunz:
> >>>>> On Fri, 31 Mar 2006 10:48:01 +0200, "Achim H."
> >>>>> <achim-bpl at mol-net.com> said:
> >
> > How can I achieve both correct copying and deletion semantics ?
>
>    I'm not sure about deletion, but here is how I do the copying (with
> a lot of help from Dave Abrahams)...
>
>
> 
http://www.slac.stanford.edu/grp/ek/hippodraw/lib/FunctionWrap_8cxx-source.html#l00179 

I came to a similar solution, in my case not refering directly to the owning 
python object, but looking up the class in the __main__ namespace (class name 
was present as a member for introspection reasons anyway). Your solution is 
more widely applicable (though at the moment only working in the case that 
there is no python override, as David Abrahams pointed out). I will therefore 
probably adopt your strategy.

However, your way of cloning IMHO creates the same chicken-and-egg problem 
regarding ownership as my Wrapper did when I played around: the Wrap is owned 
by the python object and the python object is owned by the wrap. A delete on 
the Wrap will auto-destruct the python object, that will then try to delete 
the Wrap.

If you try to delete it from the C++ side, you will get an error (e.g., 
SIGSEGV). I'm not totally sure whether Python at the end tries to delete all 
objects still being around. If python does not, you created a memory leak 
which probably is a minor concern to you if you don't have many instances of 
FunctionWrap around.

I myself played around with an arbitrary Py_INCREF() which at first only 
postponed the error message to the program end. When I started to also employ 
boost::shared_ptr<> (class_<shared_ptr<Wrap> ...>), it vanished. Don't ask me 
why, it kind of works for me now. Still, I would like to understand the 
situation better and use a correct solution, not only one that works in most 
of the cases or on my current platform.

Any how, thanks for your help, it at least provides me with a better 
implementation of the clone() method, though the ownership issues are still 
unsolved for me.

Achim.


From Barnaby.Dalton at brevanhoward.com  Mon Apr  3 19:15:07 2006
From: Barnaby.Dalton at brevanhoward.com (Dalton, Barnaby)
Date: Mon, 3 Apr 2006 18:15:07 +0100
Subject: [C++-sig] boost python how to handle a type of shared_ptr<const T>
Message-ID: <383BF6A1AD220A4796FE90A9EC3FB406010DF782@bhmail1.rivagecapital.com>

How do I use a shared_ptr<const T> type in boost python? I have the
following test code:

#include <vector>
#include <boost/shared_ptr.hpp>

using namespace std;
using namespace boost;

// vector
vector<int> listTest1()
{
  vector<int> a;
  a.push_back(5); a.push_back(10);
  return a;
}

// shared pointer to vector
shared_ptr<vector<int> > listTest2()
{
  shared_ptr<vector<int> > a(new vector<int>());
  a->push_back(6); a->push_back(11);
  return a;
}

// const vector
const vector<int> listTest3()
{
  vector<int> a;
  a.push_back(7); a.push_back(12);
  return a;
}

// shared pointer to const vector
shared_ptr<const vector<int> > listTest4()
{
  shared_ptr<vector<int> > a(new vector<int>());
  a->push_back(8); a->push_back(13);
  return a;
}

#include <boost/python/module.hpp>
#include <boost/python/def.hpp>
#include <boost/python/iterator.hpp>
#include <boost/python/register_ptr_to_python.hpp>

using namespace boost::python;

BOOST_PYTHON_MODULE(hello)
{
  class_<vector<int> >("vec_int")
    .def("push_back", &vector<int>::push_back)
    .def("__iter__", iterator<vector<int> >())
        ;

    def("listTest1", listTest1);
    def("listTest2", listTest2);
    def("listTest3", listTest3);
    def("listTest4", listTest4);

    register_ptr_to_python< boost::shared_ptr<vector <int> > >();
    //register_ptr_to_python< boost::shared_ptr<const vector <int> >
>();
}


listTest 1->3 work fine. If I try and call listTest4 from python I get:

Traceback (most recent call last):
  File "test.py", line 19, in ?
    a = [x for x in hello.listTest4()]
TypeError: No to_python (by-value) converter found for C++ type: class
boost::shared_ptr<class std::vector<int,class std::allocator<int> >
const >

If I uncomment 'register_ptr_to_python< boost::shared_ptr<const vector
<int> > >();' then I get compile errors:

C:\Source\boost_1_33_1\boost\python\object\pointer_holder.hpp(138) :
error C2664: 'boost::python::objects::find_dynamic_type' : cannot
convert parameter 1 from 'const X *' to 'void *'
        Conversion loses qualifiers
        C:\Program Files\Microsoft Visual Studio .NET
2003\Vc7\include\vector(152) : while compiling class-template member
function 'void
*boost::python::objects::pointer_holder<Pointer,Value>::holds(boost::pyt
hon::type_info,bool)'
        with
        [
            Pointer=boost::shared_ptr<const std::vector<int>>,
            Value=const X
        ]
etc....

I'm new to boost python so please feel free to point to any
documentation/examples I might have missed.

Thanks
Barney







From me at elwedgo.de  Tue Apr  4 12:07:01 2006
From: me at elwedgo.de (Peter Dinges)
Date: Tue, 4 Apr 2006 10:07:01 +0000 (UTC)
Subject: [C++-sig] Wrapper for exception translation
References: <2040C0A1CA23D51181A30050BAAC9902011046D8@berexch.ber.haufemg.com>
	<d1ofeg$p6$1@sea.gmane.org> <u1xa77n4g.fsf@boost-consulting.com>
	<d1r4u1$n3q$1@sea.gmane.org> <d3fhsl$33m$1@sea.gmane.org>
	<ull7lpzhh.fsf@boost-consulting.com>
Message-ID: <loom.20060404T115751-923@post.gmane.org>

David Abrahams <dave <at> boost-consulting.com> writes:   
>    
> Colin Irwin <colin_irwin <at> hotmail.com> writes:   
> 
> > Anyway, I think I'm still keen to do whatever on this front to help   
> > with a wrapper to translate exceptions through into Python.   
>    
> Thanks; I'll keep it in mind.   
>    
  
Hello,   
   
just wanted to ask whether anyone found time yet to implement   
the exception wrapping template?   
   
Best regards,   
Peter Dinges   
   



From nail-mail at mail.ru  Tue Apr  4 16:04:45 2006
From: nail-mail at mail.ru (Victor Nakoryakov)
Date: Tue, 04 Apr 2006 18:04:45 +0400
Subject: [C++-sig] Renaming "<string>" filename for PyRun_String
Message-ID: <e0tuga$vqt$1@sea.gmane.org>

Hi all,

I use PyRun_String to run script stored in some file. I can't use 
PyRun_File because of different layouts of FILE struct in my project and 
in python lib. So once I get error I see something like:

Traceback (most recent call last):
   File "<string>", line 14, in ?
NameError: ...

Is it possible somehow to rename "<string>" in this message to the real 
filename?

-- 
Victor (aka nail) Nakoryakov
nail-mail<at>mail<dot>ru

Krasnoznamensk, Moscow, Russia



From kelly at kkcsm.net  Wed Apr  5 06:29:52 2006
From: kelly at kkcsm.net (Kelly Burkhart)
Date: Tue, 4 Apr 2006 23:29:52 -0500
Subject: [C++-sig] Adding a method on the fly
Message-ID: <fa1e4ce70604042129s287cf431x4655d344f794c8@mail.gmail.com>

Is it possible to add a method to an object on the fly?  Something like:

// Add a method 'foo' to object
void addMethod( object obj )
{
    call_method<void>(obj.ptr(),"__setattr__", "foo", "lambda x:x+x" );
}

The "lambda..." string would have to get evaluated by Python to return an
appropriate object to bind to foo.  Is there any way to do something like
this?

TIA,

-K
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060404/c06b0f4f/attachment.htm>

From seefeld at sympatico.ca  Wed Apr  5 07:55:42 2006
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Wed, 05 Apr 2006 01:55:42 -0400
Subject: [C++-sig] Adding a method on the fly
In-Reply-To: <fa1e4ce70604042129s287cf431x4655d344f794c8@mail.gmail.com>
References: <fa1e4ce70604042129s287cf431x4655d344f794c8@mail.gmail.com>
Message-ID: <44335BDE.20101@sympatico.ca>

Kelly Burkhart wrote:
> Is it possible to add a method to an object on the fly?  Something like:
> 
> // Add a method 'foo' to object
> void addMethod( object obj )
> {
>     call_method<void>(obj.ptr(),"__setattr__", "foo", "lambda x:x+x" );
> }
> 
> The "lambda..." string would have to get evaluated by Python to return 
> an appropriate object to bind to foo.  Is there any way to do something 
> like this?

Yes, the following code ought to do what you want:

using namespace boost::python;

int main(int, char **)
{

   Py_Initialize();
   try
   {
     // Retrieve the main module
     object main = import("__main__");

     // Retrieve the main module's namespace
     object global(main.attr("__dict__"));

     // Define the derived class in Python.
     object result = exec(
       "class A:                            \n"
       "    def hello(self):                \n"
       "        return 'Hello from Python!' \n"
       "def func(self, x):                  \n"
       "    return x+x                      \n",
       global, global);
     object A = global["A"];
     A.attr("method") = global["func"];
     result = exec("a = A()                 \n"
		  "print a.hello()         \n"
		  "print a.method(21)      \n",
		  global, global);
   }
   catch (error_already_set)
   {
     PyErr_Print();
   }
}

Note that the first exec defines class 'A' as well as a function 'func'
in python and stores them in 'global'. Then, from within C++, 'func'
is attached to 'A' under the name 'method'. Finally, another chunk
of python is executed, using the same namespace (making sure the
modified 'A' class is exposed), where the newly created method is run
through a new instance of 'A'.

HTH,
		Stefan



From star_1234 at sancharnet.in  Wed Apr  5 10:13:07 2006
From: star_1234 at sancharnet.in (satish)
Date: Wed, 05 Apr 2006 13:43:07 +0530
Subject: [C++-sig] Your confirmation is required to join the C++-sig
	mailing list
References: <mailman.0.1143789233.1406.c++-sig@python.org>
Message-ID: <000e01c65888$c654a370$52b5583b@starsoft3ojkis>


----- Original Message ----- 
From: <c++-sig-confirm+39904eb3065e1b9624fdfd44ebcac837d019c81a at python.org>
To: <star1234 at sancharnet.in>
Sent: Friday, March 31, 2006 12:43 PM
Subject: Your confirmation is required to join the C++-sig mailing list


> Mailing list subscription confirmation notice for mailing list C++-sig
>
> We have received a request from 203.145.188.130 for subscription of
> your email address, "star1234 at sancharnet.in", to the
> c++-sig at python.org mailing list.  To confirm that you want to be added
> to this mailing list, simply reply to this message, keeping the
> Subject: header intact.  Or visit this web page:
>
>
http://mail.python.org/mailman/confirm/c++-sig/39904eb3065e1b9624fdfd44ebcac837d019c81a
>
>
> Or include the following line -- and only the following line -- in a
> message to c++-sig-request at python.org:
>
>     confirm 39904eb3065e1b9624fdfd44ebcac837d019c81a
>
> Note that simply sending a `reply' to this message should work from
> most mail readers, since that usually leaves the Subject: line in the
> right form (additional "Re:" text in the Subject: is okay).
>
> If you do not wish to be subscribed to this list, please simply
> disregard this message.  If you think you are being maliciously
> subscribed to the list, or have any other questions, send them to
> c++-sig-owner at python.org.



From Naceur.Meskini at sophia.inria.fr  Wed Apr  5 10:45:48 2006
From: Naceur.Meskini at sophia.inria.fr (Naceur Meskini)
Date: Wed, 05 Apr 2006 10:45:48 +0200
Subject: [C++-sig] Your confirmation is required to join the
 C++-sig	mailing list
In-Reply-To: <000e01c65888$c654a370$52b5583b@starsoft3ojkis>
References: <mailman.0.1143789233.1406.c++-sig@python.org>
	<000e01c65888$c654a370$52b5583b@starsoft3ojkis>
Message-ID: <443383BC.2000104@sophia.inria.fr>

satish wrote:

>----- Original Message ----- 
>From: <c++-sig-confirm+39904eb3065e1b9624fdfd44ebcac837d019c81a at python.org>
>To: <star1234 at sancharnet.in>
>Sent: Friday, March 31, 2006 12:43 PM
>Subject: Your confirmation is required to join the C++-sig mailing list
>
>
>  
>
>>Mailing list subscription confirmation notice for mailing list C++-sig
>>
>>We have received a request from 203.145.188.130 for subscription of
>>your email address, "star1234 at sancharnet.in", to the
>>c++-sig at python.org mailing list.  To confirm that you want to be added
>>to this mailing list, simply reply to this message, keeping the
>>Subject: header intact.  Or visit this web page:
>>
>>
>>    
>>
>http://mail.python.org/mailman/confirm/c++-sig/39904eb3065e1b9624fdfd44ebcac837d019c81a
>  
>
>>Or include the following line -- and only the following line -- in a
>>message to c++-sig-request at python.org:
>>
>>    confirm 39904eb3065e1b9624fdfd44ebcac837d019c81a
>>
>>Note that simply sending a `reply' to this message should work from
>>most mail readers, since that usually leaves the Subject: line in the
>>right form (additional "Re:" text in the Subject: is okay).
>>
>>If you do not wish to be subscribed to this list, please simply
>>disregard this message.  If you think you are being maliciously
>>subscribed to the list, or have any other questions, send them to
>>c++-sig-owner at python.org.
>>    
>>
>
>_______________________________________________
>C++-sig mailing list
>C++-sig at python.org
>http://mail.python.org/mailman/listinfo/c++-sig
>  
>



From kelly at kkcsm.net  Wed Apr  5 15:17:28 2006
From: kelly at kkcsm.net (Kelly Burkhart)
Date: Wed, 5 Apr 2006 08:17:28 -0500
Subject: [C++-sig] Adding a method on the fly
In-Reply-To: <44335BDE.20101@sympatico.ca>
References: <fa1e4ce70604042129s287cf431x4655d344f794c8@mail.gmail.com>
	<44335BDE.20101@sympatico.ca>
Message-ID: <fa1e4ce70604050617v55e2221s1333144e8f8a94de@mail.gmail.com>

On 4/5/06, Stefan Seefeld <seefeld at sympatico.ca> wrote:
>
>
> Yes, the following code ought to do what you want:
>
> using namespace boost::python;
>
> int main(int, char **)
> {
>
>    Py_Initialize();
>    try
>    {
>      // Retrieve the main module
>      object main = import("__main__");
>
>      // Retrieve the main module's namespace
>      object global(main.attr("__dict__"));
>
>      // Define the derived class in Python.
>      object result = exec(
>        "class A:                            \n"
>        "    def hello(self):                \n"
>        "        return 'Hello from Python!' \n"
>        "def func(self, x):                  \n"
>        "    return x+x                      \n",
>        global, global);
>      object A = global["A"];
>      A.attr("method") = global["func"];
>      result = exec("a = A()                 \n"
>                   "print a.hello()         \n"
>                   "print a.method(21)      \n",
>                   global, global);
>    }
>    catch (error_already_set)
>    {
>      PyErr_Print();
>    }
> }
>
> Note that the first exec defines class 'A' as well as a function 'func'
> in python and stores them in 'global'. Then, from within C++, 'func'
> is attached to 'A' under the name 'method'. Finally, another chunk
> of python is executed, using the same namespace (making sure the
> modified 'A' class is exposed), where the newly created method is run
> through a new instance of 'A'.
>


 Thank you Stefan,

I can't find 'exec'.  Is it in the boost::python namespace?  What header do
I need to include?

-K
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060405/27404d75/attachment.htm>

From seefeld at sympatico.ca  Wed Apr  5 15:19:56 2006
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Wed, 05 Apr 2006 09:19:56 -0400
Subject: [C++-sig] Adding a method on the fly
In-Reply-To: <fa1e4ce70604050617v55e2221s1333144e8f8a94de@mail.gmail.com>
References: <fa1e4ce70604042129s287cf431x4655d344f794c8@mail.gmail.com>	<44335BDE.20101@sympatico.ca>
	<fa1e4ce70604050617v55e2221s1333144e8f8a94de@mail.gmail.com>
Message-ID: <4433C3FC.70807@sympatico.ca>

Kelly Burkhart wrote:

> I can't find 'exec'.  Is it in the boost::python namespace?  What header 
> do I need to include?

It's in the development version, and will be part of the upcoming release.
Sorry for not mentioning that.

Regards,
		Stefan


From kelly at kkcsm.net  Wed Apr  5 18:17:33 2006
From: kelly at kkcsm.net (Kelly Burkhart)
Date: Wed, 5 Apr 2006 11:17:33 -0500
Subject: [C++-sig] Adding a method on the fly
In-Reply-To: <4433C3FC.70807@sympatico.ca>
References: <fa1e4ce70604042129s287cf431x4655d344f794c8@mail.gmail.com>
	<44335BDE.20101@sympatico.ca>
	<fa1e4ce70604050617v55e2221s1333144e8f8a94de@mail.gmail.com>
	<4433C3FC.70807@sympatico.ca>
Message-ID: <fa1e4ce70604050917g16562c51wc3fa1dc9e6253b0b@mail.gmail.com>

On 4/5/06, Stefan Seefeld <seefeld at sympatico.ca> wrote:
>
> Kelly Burkhart wrote:
>
> > I can't find 'exec'.  Is it in the boost::python namespace?  What header
> > do I need to include?
>
> It's in the development version, and will be part of the upcoming release.
> Sorry for not mentioning that.
>


No problem.  Will this development version be production soon?

Is there a way to do what I'm looking for with v1.33.1?

-K
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060405/e9f6422f/attachment.htm>

From seefeld at sympatico.ca  Wed Apr  5 18:21:31 2006
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Wed, 05 Apr 2006 12:21:31 -0400
Subject: [C++-sig] Adding a method on the fly
In-Reply-To: <fa1e4ce70604050917g16562c51wc3fa1dc9e6253b0b@mail.gmail.com>
References: <fa1e4ce70604042129s287cf431x4655d344f794c8@mail.gmail.com>	<44335BDE.20101@sympatico.ca>	<fa1e4ce70604050617v55e2221s1333144e8f8a94de@mail.gmail.com>	<4433C3FC.70807@sympatico.ca>
	<fa1e4ce70604050917g16562c51wc3fa1dc9e6253b0b@mail.gmail.com>
Message-ID: <4433EE8B.9020501@sympatico.ca>

Kelly Burkhart wrote:
> On 4/5/06, *Stefan Seefeld* <seefeld at sympatico.ca 
> <mailto:seefeld at sympatico.ca>> wrote:
> 
>     Kelly Burkhart wrote:
> 
>      > I can't find 'exec'.  Is it in the boost::python namespace?  What
>     header
>      > do I need to include?
> 
>     It's in the development version, and will be part of the upcoming
>     release.
>     Sorry for not mentioning that.
> 
> 
> 
> No problem.  Will this development version be production soon?

As I said, it will be in the upcoming release, i.e. 1.34.

> Is there a way to do what I'm looking for with v1.33.1?

I'm not sure what bits you were particularly interested in.
I showed the 'exec' function only as an example (because constructing
and running a python object in pure C++ would be pretty pointless. :-) ).

You were asking about binding a function to an object, which doesn't
require 'exec'.

Regards,
		Stefan


From dominic.sacre at gmx.de  Wed Apr  5 21:05:32 2006
From: dominic.sacre at gmx.de (Dominic =?iso-8859-1?q?Sacr=E9?=)
Date: Wed, 5 Apr 2006 21:05:32 +0200
Subject: [C++-sig] Wrapping pointers
Message-ID: <200604052105.33008.dominic.sacre@gmx.de>

Hi,

I'm having trouble exporting C++ pointers to Python. My code looks like 
this:

class A
{
    A *p;
    ...
};

BOOST_PYTHON_MODULE(test)
{
    class_<A>("A", init<>())
        .def_readonly("p", &A::p)
        ...
    ;
}

Now every time I try to access A::p from Python I get this error:

TypeError: No to_python (by-value) converter found for C++ type: A*

I searched the list archives and found a message about the same issue, 
where the only suggestion was to use

.def_readonly("p", make_getter(&A::p, return_internal_reference<1>()))

instead, but this doesn't help either. On the contrary, this causes a 
segfault every time I use A::p (although p points to a valid object, at 
least in C++).
So what should I do to access this pointer from Python?

Thanks,
    Dominic


From kelly at kkcsm.net  Wed Apr  5 21:58:48 2006
From: kelly at kkcsm.net (Kelly Burkhart)
Date: Wed, 5 Apr 2006 14:58:48 -0500
Subject: [C++-sig] Adding a method on the fly
In-Reply-To: <4433EE8B.9020501@sympatico.ca>
References: <fa1e4ce70604042129s287cf431x4655d344f794c8@mail.gmail.com>
	<44335BDE.20101@sympatico.ca>
	<fa1e4ce70604050617v55e2221s1333144e8f8a94de@mail.gmail.com>
	<4433C3FC.70807@sympatico.ca>
	<fa1e4ce70604050917g16562c51wc3fa1dc9e6253b0b@mail.gmail.com>
	<4433EE8B.9020501@sympatico.ca>
Message-ID: <fa1e4ce70604051258x2013e3b2nda636086823e580f@mail.gmail.com>

On 4/5/06, Stefan Seefeld <seefeld at sympatico.ca> wrote:
>
> Kelly Burkhart wrote:
> > Is there a way to do what I'm looking for with v1.33.1?
>
> I'm not sure what bits you were particularly interested in.
> I showed the 'exec' function only as an example (because constructing
> and running a python object in pure C++ would be pretty pointless. :-) ).
>
> You were asking about binding a function to an object, which doesn't
> require 'exec'.
>

I'm trying to programatically create a python function and bind it to a C++
object.  Using your code above as an example, I belive it may be something
like this:


>From C++:

using namespace boost;
using namespace boost::python;

class Foo {...};

void mkMethod( object self, const std::string &name ) {
    // Get Foo instance namespace
    object selfdict(self.attr("__dict__"));

    // Define class
    exec(str(format("def %s( arg ): print arg") % name.c_str),
         selfdict, selfdict);
}
...
// Wrap class Foo
class_<Foo> cls("Foo");

// Wrap mkMethod function
def("mkMethod", mkMethod);

// Attach mkMethod to Foo
cls.setattr("mkMethod",&mkMethod)


>From Python:

from foomod import Foo
f = Foo()
f.mkMethod('newmethod')

# now a brand new method of f exists
f.newmethod(123)


-K
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060405/fc461616/attachment.htm>

From mharidev at qualcomm.com  Wed Apr  5 22:44:34 2006
From: mharidev at qualcomm.com (Haridev, Meghana)
Date: Wed, 5 Apr 2006 13:44:34 -0700
Subject: [C++-sig] Problem running pygccxml example
Message-ID: <5383A81D5EB8F14FBDB487D9B5FBEA88A23716@NAEX11.na.qualcomm.com>

Hi Folks,

 

I'm trying to run the pygccxml example
(http://www.language-binding.net/pygccxml/pygccxml.html#i-d-like-to-see-
an-example). (Also found under pygccxml-0.6.9/docs from the installation
tarball)

 

I modified example.py
(http://www.language-binding.net/pygccxml/example.py ) to set the
correct path for gccxml:

config = pygccxml.parser.config_t(
gccxml_path=r'/opt/buildtools/bin/gccxml' )

 

The source file 'core_class_hierarchy.hpp' has no changes:
http://www.language-binding.net/pygccxml/core_class_hierarchy.hpp

 

But I get the following error when I run example.py: 

 

Traceback (most recent call last):

  File "example.py", line 13, in ?

    namespaces = pygccxml.parser.parse( ['core_class_hierarchy.hpp'],
config )

  File
"/opt/buildtools/lib/python2.4/site-packages/pygccxml/parser/__init__.py
", line 25, in parse

    answer = parser.read_files(files, compilation_mode)

  File
"/opt/buildtools/lib/python2.4/site-packages/pygccxml/parser/project_rea
der.py", line 61, in read_files    return
self.__parse_file_by_file(files)

  File
"/opt/buildtools/lib/python2.4/site-packages/pygccxml/parser/project_rea
der.py", line 82, in __parse_file_by_file

    decls = reader.read_file( header )

  File
"/opt/buildtools/lib/python2.4/site-packages/pygccxml/parser/source_read
er.py", line 140, in read_file

    raise error

KeyError: u'complex long double'

 

I am using:

GCC: 4.0.2

Python: 2.4.2

Gccxml: 0.7.0

Pygccxml: 0.6.9

 

Any pointers to what I'm doing wrong? Am I missing anything in the
configuration?

 

 

Thanks,

-Meghana.

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060405/2b274c41/attachment.htm>

From kelly at kkcsm.net  Thu Apr  6 03:29:19 2006
From: kelly at kkcsm.net (Kelly Burkhart)
Date: Wed, 5 Apr 2006 20:29:19 -0500
Subject: [C++-sig] Adding a method on the fly
In-Reply-To: <fa1e4ce70604051258x2013e3b2nda636086823e580f@mail.gmail.com>
References: <fa1e4ce70604042129s287cf431x4655d344f794c8@mail.gmail.com>
	<44335BDE.20101@sympatico.ca>
	<fa1e4ce70604050617v55e2221s1333144e8f8a94de@mail.gmail.com>
	<4433C3FC.70807@sympatico.ca>
	<fa1e4ce70604050917g16562c51wc3fa1dc9e6253b0b@mail.gmail.com>
	<4433EE8B.9020501@sympatico.ca>
	<fa1e4ce70604051258x2013e3b2nda636086823e580f@mail.gmail.com>
Message-ID: <fa1e4ce70604051829x7a6bc757h3e688b3e46b942a6@mail.gmail.com>

Success!

#include <boost/python.hpp>
#include <boost/format.hpp>
#include <string>

using namespace boost::python;

struct Foo {};

void mkMethod( object self, const std::string &name, int val )
{
    // Get self instance namespace
    object selfdict( self.attr("__dict__") );

    // Create function and bind to self.name
    std::string cmd( boost::str(boost::format("lambda x:x+%d") % val ));
    object result((handle<>(
                            PyRun_String(cmd.c_str()
                                         , Py_eval_input
                                         , selfdict.ptr()
                                         , selfdict.ptr()))
                   ));
    call_method<void>(self.ptr(),"__setattr__", name.c_str(), result);
}

BOOST_PYTHON_MODULE(bptst)
{
    class_< Foo > cls( "Foo", init<>() );
    def("mkMethod", &mkMethod);
    cls.setattr("mkMethod", &mkMethod);
}

>From Python:

>>> import bptst
>>> f = bptst.Foo()
>>> print f.__dict__
{}
>>> f.mkMethod('bar', 123)
>>> print f.__dict__
{'bar': <function <lambda> at 0x402ed3ac>}
>>> print f.bar(1)
124
>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060405/9cbecb81/attachment.htm>

From roman.yakovenko at gmail.com  Thu Apr  6 07:33:53 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Thu, 6 Apr 2006 07:33:53 +0200
Subject: [C++-sig] Problem running pygccxml example
In-Reply-To: <5383A81D5EB8F14FBDB487D9B5FBEA88A23716@NAEX11.na.qualcomm.com>
References: <5383A81D5EB8F14FBDB487D9B5FBEA88A23716@NAEX11.na.qualcomm.com>
Message-ID: <7465b6170604052233s758fe662pf2879a9fedf4a5a0@mail.gmail.com>

On 4/5/06, Haridev, Meghana <mharidev at qualcomm.com> wrote:
> Hi Folks,
>
>
>
>
> I am using:
>
> GCC: 4.0.2
>
> Python: 2.4.2
>
> Gccxml: 0.7.0
>
> Pygccxml: 0.6.9
>
>
>
> Any pointers to what I'm doing wrong? Am I missing anything in the
> configuration?

You just hited the bug found and fixed in cvs.  Can you use CVS version?

Today I am going to commit final version
of pygccxml and pyplusplus for the next release. I also update the example.

>
> Thanks,
>
> -Meghana.
>
>
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>
>
>


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From lloyd at fusion.net.nz  Thu Apr  6 11:04:01 2006
From: lloyd at fusion.net.nz (Lloyd Weehuizen)
Date: Thu, 06 Apr 2006 21:04:01 +1200
Subject: [C++-sig] Problems with Python "Factories"
In-Reply-To: <7465b6170604052233s758fe662pf2879a9fedf4a5a0@mail.gmail.com>
References: <5383A81D5EB8F14FBDB487D9B5FBEA88A23716@NAEX11.na.qualcomm.com>
	<7465b6170604052233s758fe662pf2879a9fedf4a5a0@mail.gmail.com>
Message-ID: <4434D981.6070603@fusion.net.nz>

Hi

I've run into a problem wrapping a C++ base class thats extended in 
Python and contains a factory method.

Here's a simplified example:

class FactoryBaseType
{
};

class FactoryBase
{
    virtual FactoryBaseType*  InstanceType();
};

In python I override this with a concrete type:

import mylib

class ConcreatePythonType( mylib.FactoryBaseType ):
     pass

class Factory( mylib.FactoryBase )
  def InstanceType():
       return ConcreatePythonType()

I've managed to set up the wrappers for the types and everything works 
great up until I call the InstanceType method on a python based factory 
from C++. The object is created and returned correctly to C++ from 
python. The C++ wrapper then uses extract<> to pull out the 
FactoryBaseType and return it to the callee and thats where it all goes 
wrong. When the wrappers InstanceType completes the local 
boost::python::object that has its pointer extracted from goes out of 
scope and the extract'd pointer becomes a dangling pointer as nobody is 
referencing the constructed python object anymore.

I'm not exactly sure how to solve this, do I have to hack around this by 
implementing some kind of temporary reference on the object? Or is there 
some way around this problem that I've missed?

Thanks for your help
Lloyd


From roman.yakovenko at gmail.com  Thu Apr  6 11:31:31 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Thu, 6 Apr 2006 11:31:31 +0200
Subject: [C++-sig] Problems with Python "Factories"
In-Reply-To: <4434D981.6070603@fusion.net.nz>
References: <5383A81D5EB8F14FBDB487D9B5FBEA88A23716@NAEX11.na.qualcomm.com>
	<7465b6170604052233s758fe662pf2879a9fedf4a5a0@mail.gmail.com>
	<4434D981.6070603@fusion.net.nz>
Message-ID: <7465b6170604060231w5a2c2900sc6c581622057281d@mail.gmail.com>

On 4/6/06, Lloyd Weehuizen <lloyd at fusion.net.nz> wrote:
> Hi
>
> I've run into a problem wrapping a C++ base class thats extended in
> Python and contains a factory method.
>
> Here's a simplified example:
>
> class FactoryBaseType
> {
> };
>
> class FactoryBase
> {
>     virtual FactoryBaseType*  InstanceType();
> };
>
> In python I override this with a concrete type:
>
> import mylib
>
> class ConcreatePythonType( mylib.FactoryBaseType ):
>      pass


you should override __init__ method. You have to call
mylib.FactoryBaseType.__init__

> class Factory( mylib.FactoryBase )

same as above

>   def InstanceType():
>        return ConcreatePythonType()
>
> I've managed to set up the wrappers for the types and everything works
> great up until I call the InstanceType method on a python based factory
> from C++. The object is created and returned correctly to C++ from
> python. The C++ wrapper then uses extract<> to pull out the
> FactoryBaseType and return it to the callee and thats where it all goes
> wrong. When the wrappers InstanceType completes the local
> boost::python::object that has its pointer extracted from goes out of
> scope and the extract'd pointer becomes a dangling pointer as nobody is
> referencing the constructed python object anymore.
>
> I'm not exactly sure how to solve this, do I have to hack around this by
> implementing some kind of temporary reference on the object? Or is there
> some way around this problem that I've missed?

I am almost sure the code that you will add will solve your problem

> Thanks for your help
> Lloyd
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From achim-bpl at mol-net.com  Thu Apr  6 11:37:08 2006
From: achim-bpl at mol-net.com (Achim H.)
Date: Thu, 6 Apr 2006 11:37:08 +0200
Subject: [C++-sig] Problems with Python "Factories"
In-Reply-To: <4434D981.6070603@fusion.net.nz>
References: <5383A81D5EB8F14FBDB487D9B5FBEA88A23716@NAEX11.na.qualcomm.com>
	<7465b6170604052233s758fe662pf2879a9fedf4a5a0@mail.gmail.com>
	<4434D981.6070603@fusion.net.nz>
Message-ID: <200604061137.08954.achim-bpl@mol-net.com>

Am Donnerstag, 6. April 2006 11:04 schrieb Lloyd Weehuizen:
> I've managed to set up the wrappers for the types and everything works
> great up until I call the InstanceType method on a python based factory
> from C++. The object is created and returned correctly to C++ from
> python. The C++ wrapper then uses extract<> to pull out the
> FactoryBaseType and return it to the callee and thats where it all goes
> wrong. When the wrappers InstanceType completes the local
> boost::python::object that has its pointer extracted from goes out of
> scope and the extract'd pointer becomes a dangling pointer as nobody is
> referencing the constructed python object anymore.
>
> I'm not exactly sure how to solve this, do I have to hack around this by
> implementing some kind of temporary reference on the object? Or is there
> some way around this problem that I've missed?

I asked a similar question last week and there is no real solution.
See the thread with these messages:

http://article.gmane.org/gmane.comp.python.c++/9528
http://article.gmane.org/gmane.comp.python.c++/9546

Achim.

P.S.: Somehow, I can only find my own mails from this thread on gmane, not the 
answers from other people. Does anybody know the reason why?



From abhi at qualcomm.com  Thu Apr  6 17:25:12 2006
From: abhi at qualcomm.com (Abhi)
Date: Thu, 06 Apr 2006 08:25:12 -0700
Subject: [C++-sig] Problem running pygccxml example
In-Reply-To: <7465b6170604052233s758fe662pf2879a9fedf4a5a0@mail.gmail.com>
References: <5383A81D5EB8F14FBDB487D9B5FBEA88A23716@NAEX11.na.qualcomm.com>
	<7465b6170604052233s758fe662pf2879a9fedf4a5a0@mail.gmail.com>
Message-ID: <E98A57531FBC36B923CEBA2C@[10.50.16.122]>

This is due to the mismatch between the versions of gcc and pygccxml? or is 
it python and pygccxml?
Just curious
thanks
- Abhi


--On Thursday, April 06, 2006 7:33 AM +0200 Roman Yakovenko 
<roman.yakovenko at gmail.com> wrote:

> On 4/5/06, Haridev, Meghana <mharidev at qualcomm.com> wrote:
>> Hi Folks,
>>
>>
>>
>>
>> I am using:
>>
>> GCC: 4.0.2
>>
>> Python: 2.4.2
>>
>> Gccxml: 0.7.0
>>
>> Pygccxml: 0.6.9
>>
>>
>>
>> Any pointers to what I'm doing wrong? Am I missing anything in the
>> configuration?
>
> You just hited the bug found and fixed in cvs.  Can you use CVS version?
>
> Today I am going to commit final version
> of pygccxml and pyplusplus for the next release. I also update the
> example.
>
>>
>> Thanks,
>>
>> -Meghana.
>>
>>
>> _______________________________________________
>> C++-sig mailing list
>> C++-sig at python.org
>> http://mail.python.org/mailman/listinfo/c++-sig
>>
>>
>>
>
>
> --
> Roman Yakovenko
> C++ Python language binding
> http://www.language-binding.net/
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig






From roman.yakovenko at gmail.com  Thu Apr  6 17:35:16 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Thu, 6 Apr 2006 17:35:16 +0200
Subject: [C++-sig] Problem running pygccxml example
In-Reply-To: <E98A57531FBC36B923CEBA2C@10.50.16.122>
References: <5383A81D5EB8F14FBDB487D9B5FBEA88A23716@NAEX11.na.qualcomm.com>
	<7465b6170604052233s758fe662pf2879a9fedf4a5a0@mail.gmail.com>
	<E98A57531FBC36B923CEBA2C@10.50.16.122>
Message-ID: <7465b6170604060835n59644cecu1dc06ff2dbbdab03@mail.gmail.com>

On 4/6/06, Abhi <abhi at qualcomm.com> wrote:
> This is due to the mismatch between the versions of gcc and pygccxml? or is
> it python and pygccxml?

There is an other option: gccxml and pygccxml. Brad King, the author
of gccxml, did not
released gccxml for a long time. He added a lot of useful features to
it. Thus he forces
people to use cvs version. pygccxml ( 0.6.9 ) has been checked against
gccxml latest
release. cvs version of pygccxml uses almost latest version of gccxml.
So this bug has been
fixed.

> Just curious
> thanks
> - Abhi


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From greg.landrum at gmail.com  Thu Apr  6 20:21:09 2006
From: greg.landrum at gmail.com (Greg Landrum)
Date: Thu, 6 Apr 2006 11:21:09 -0700
Subject: [C++-sig] operator definitions
Message-ID: <60825b0f0604061121v2f4578fs2aa8ce73d759b943@mail.gmail.com>

I just ran across this syntax for defining operators on exposed classes:
class_<FilePos>("FilePos")
    .def(self + int())          // __add__
    .def(int() + self)          // __radd__
    .def(self - self)           // __sub__
    .def(self - int())          // __sub__
    .def(self += int())         // __iadd__

(example from: http://www.boost.org/libs/python/doc/tutorial/doc/html/python/exposing.html)

I don't know when this was added (or if it's been here all along and I
have just never noticed), but this is amazingly clean and useful.
Thanks!

-greg


From mharidev at qualcomm.com  Fri Apr  7 00:53:28 2006
From: mharidev at qualcomm.com (Haridev, Meghana)
Date: Thu, 6 Apr 2006 15:53:28 -0700
Subject: [C++-sig] Problem running pygccxml example
Message-ID: <5383A81D5EB8F14FBDB487D9B5FBEA88C0DAD0@NAEX11.na.qualcomm.com>

Thanks for the tip Roman - I am now able to run the pygccxml example
after installing pygccxml from CVS.

Now, moving to the pyplusplus example (pyplusplus also installed from
CVS today), I'm trying to run the hello_world example at
/pyplusplus/examples/tutorials. 
I made the appropriate changes to environment.py.

I got this error when I ran simple.py:

------------------------------------------------------------------------
--
Traceback (most recent call last):   File "simple.py", line 32, in ?
    extmodule = module_creator.create( decls=decls,
module_name="hello_world",recursive=False)
TypeError: create() got an unexpected keyword argument 'recursive'

------------------------------------------------------------------------
--

After looking at the __init__.py for module_creator which defines
create() as:

def create( decls, module_name ) 

I removed the extra "recursive" keyword from the argument list. 

But then I run into this error:
------------------------------------------------------------------------
--
Traceback (most recent call last):
  File "simple.py", line 32, in ?
    extmodule = module_creator.create( decls=decls,
module_name="hello_world")
  File
"/opt/data/install-pyplusplus/lib/python2.4/site-packages/pyplusplus/mod
ule_creator/__init__.py", line 11, in create
    maker = creator_t(decls, module_name)
  File
"/opt/data/install-pyplusplus/lib/python2.4/site-packages/pyplusplus/mod
ule_creator/creator.py", line 84, in __init__
    self.__decls = self._filter_decls( self._reorder_decls(
self._prepare_decls( decls ) ) )
  File
"/opt/data/install-pyplusplus/lib/python2.4/site-packages/pyplusplus/mod
ule_creator/creator.py", line 98, in _prepare_decls
    decls = filter( lambda x: not x.ignore, decls )
  File
"/opt/data/install-pyplusplus/lib/python2.4/site-packages/pyplusplus/mod
ule_creator/creator.py", line 98, in <lambda>
    decls = filter( lambda x: not x.ignore, decls )
AttributeError: 'enumeration_t' object has no attribute 'ignore'
------------------------------------------------------------------------
--

Any suggestions?

Thanks,
-Meghana.


-----Original Message-----
From: c++-sig-bounces at python.org [mailto:c++-sig-bounces at python.org] On
Behalf Of Roman Yakovenko
Sent: Wednesday, April 05, 2006 10:34 PM
To: Development of Python/C++ integration
Subject: Re: [C++-sig] Problem running pygccxml example

On 4/5/06, Haridev, Meghana <mharidev at qualcomm.com> wrote:
> Hi Folks,
>
>
>
>
> I am using:
>
> GCC: 4.0.2
>
> Python: 2.4.2
>
> Gccxml: 0.7.0
>
> Pygccxml: 0.6.9
>
>
>
> Any pointers to what I'm doing wrong? Am I missing anything in the
> configuration?

You just hited the bug found and fixed in cvs.  Can you use CVS version?

Today I am going to commit final version
of pygccxml and pyplusplus for the next release. I also update the
example.

>
> Thanks,
>
> -Meghana.
>
>
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>
>
>


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/
_______________________________________________
C++-sig mailing list
C++-sig at python.org
http://mail.python.org/mailman/listinfo/c++-sig


From roman.yakovenko at gmail.com  Fri Apr  7 19:15:46 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Fri, 7 Apr 2006 20:15:46 +0300
Subject: [C++-sig] Problem running pygccxml example
In-Reply-To: <5383A81D5EB8F14FBDB487D9B5FBEA88C0DAD0@NAEX11.na.qualcomm.com>
References: <5383A81D5EB8F14FBDB487D9B5FBEA88C0DAD0@NAEX11.na.qualcomm.com>
Message-ID: <7465b6170604071015q60021a4gc855a9e3546926c3@mail.gmail.com>

On 4/7/06, Haridev, Meghana <mharidev at qualcomm.com> wrote:
> Thanks for the tip Roman - I am now able to run the pygccxml example
> after installing pygccxml from CVS.

> Any suggestions?

I don't know why you have an old example.  I attach new version of example
and relevant file. This is actually a CVS version. So may be you can
check out a cvs version one more time?

The code in cvs version is frozen and will be release in a week or two.
We are working on documentation. The example I provide should be enough to
start with pyplusplus.

> Thanks,
> -Meghana.
>


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: generate_code.py
Type: text/x-python
Size: 2155 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060407/5326ed11/attachment.py>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: environment.py
Type: text/x-python
Size: 1115 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060407/5326ed11/attachment-0001.py>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: hello_world.hpp
Type: text/x-c++hdr
Size: 1359 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060407/5326ed11/attachment.hpp>

From ndbecker2 at gmail.com  Sat Apr  8 15:36:11 2006
From: ndbecker2 at gmail.com (Neal Becker)
Date: Sat, 08 Apr 2006 09:36:11 -0400
Subject: [C++-sig] operator definitions
References: <60825b0f0604061121v2f4578fs2aa8ce73d759b943@mail.gmail.com>
Message-ID: <e18e7t$cg7$1@sea.gmane.org>

Greg Landrum wrote:

> I just ran across this syntax for defining operators on exposed classes:
> class_<FilePos>("FilePos")
>     .def(self + int())          // __add__
>     .def(int() + self)          // __radd__
>     .def(self - self)           // __sub__
>     .def(self - int())          // __sub__
>     .def(self += int())         // __iadd__
> 
> (example from:
>
http://www.boost.org/libs/python/doc/tutorial/doc/html/python/exposing.html)
> 
> I don't know when this was added (or if it's been here all along and I
> have just never noticed), but this is amazingly clean and useful.
> Thanks!
> 

It looks cool, but IMHO a bit too cute.  AFAIK, it doesn't do anything
different than the IMHO much more clear:
    .def("__iadd__", plusAssignOp, return_self<>())

Of course, beauty is in the eye and all that.



From dave at boost-consulting.com  Sat Apr  8 16:46:06 2006
From: dave at boost-consulting.com (David Abrahams)
Date: Sat, 08 Apr 2006 10:46:06 -0400
Subject: [C++-sig] operator definitions
References: <60825b0f0604061121v2f4578fs2aa8ce73d759b943@mail.gmail.com>
	<e18e7t$cg7$1@sea.gmane.org>
Message-ID: <m2r748dt1t.fsf@boonen.local>

Neal Becker <ndbecker2 at gmail.com> writes:

>> 
>> I don't know when this was added (or if it's been here all along and I
>> have just never noticed), but this is amazingly clean and useful.
>> Thanks!
>> 
>
> It looks cool, but IMHO a bit too cute.  AFAIK, it doesn't do anything
> different than the IMHO much more clear:
>     .def("__iadd__", plusAssignOp, return_self<>())

How is that more clear?  *I* certainly can't tell what it means.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com



From greg.landrum at gmail.com  Sat Apr  8 17:02:41 2006
From: greg.landrum at gmail.com (Greg Landrum)
Date: Sat, 8 Apr 2006 08:02:41 -0700
Subject: [C++-sig] operator definitions
In-Reply-To: <m2r748dt1t.fsf@boonen.local>
References: <60825b0f0604061121v2f4578fs2aa8ce73d759b943@mail.gmail.com>
	<e18e7t$cg7$1@sea.gmane.org> <m2r748dt1t.fsf@boonen.local>
Message-ID: <60825b0f0604080802i44bd476fp5ee66c1c9db9d921@mail.gmail.com>

On 4/8/06, David Abrahams <dave at boost-consulting.com> wrote:
> Neal Becker <ndbecker2 at gmail.com> writes:
>
> > It looks cool, but IMHO a bit too cute.  AFAIK, it doesn't do anything
> > different than the IMHO much more clear:
> >     .def("__iadd__", plusAssignOp, return_self<>())
>
> How is that more clear?  *I* certainly can't tell what it means.

I'm with David on this one, and the contrast is even more extreme when
you define __add__ (the "clear" syntax for that stymied me, and it was
the resulting web search that turned up the new-to-me syntax).

-greg


From ndbecker2 at gmail.com  Sat Apr  8 17:18:40 2006
From: ndbecker2 at gmail.com (Neal Becker)
Date: Sat, 08 Apr 2006 11:18:40 -0400
Subject: [C++-sig] operator definitions
References: <60825b0f0604061121v2f4578fs2aa8ce73d759b943@mail.gmail.com>
	<e18e7t$cg7$1@sea.gmane.org> <m2r748dt1t.fsf@boonen.local>
Message-ID: <e18k81$sc4$1@sea.gmane.org>

David Abrahams wrote:

> Neal Becker <ndbecker2 at gmail.com> writes:
> 
>>> 
>>> I don't know when this was added (or if it's been here all along and I
>>> have just never noticed), but this is amazingly clean and useful.
>>> Thanks!
>>> 
>>
>> It looks cool, but IMHO a bit too cute.  AFAIK, it doesn't do anything
>> different than the IMHO much more clear:
>>     .def("__iadd__", plusAssignOp, return_self<>())
> 
> How is that more clear?  *I* certainly can't tell what it means.
> 

It's clear (to me) because it's consistent with the same method used to
define any other function.  In this case "clear" assumes the observer is
already familiar with the usual useage of .def.



From dave at boost-consulting.com  Sun Apr  9 04:37:09 2006
From: dave at boost-consulting.com (David Abrahams)
Date: Sat, 08 Apr 2006 22:37:09 -0400
Subject: [C++-sig] operator definitions
References: <60825b0f0604061121v2f4578fs2aa8ce73d759b943@mail.gmail.com>
	<e18e7t$cg7$1@sea.gmane.org> <m2r748dt1t.fsf@boonen.local>
	<e18k81$sc4$1@sea.gmane.org>
Message-ID: <m2ek07zd7u.fsf@boonen.local>

Neal Becker <ndbecker2 at gmail.com> writes:

> David Abrahams wrote:
>
>> Neal Becker <ndbecker2 at gmail.com> writes:
>> 
>>> It looks cool, but IMHO a bit too cute.  AFAIK, it doesn't do anything
>>> different than the IMHO much more clear:
>>>     .def("__iadd__", plusAssignOp, return_self<>())
>> 
>> How is that more clear?  *I* certainly can't tell what it means.
>> 
>
> It's clear (to me) because it's consistent with the same method used to
> define any other function.  In this case "clear" assumes the observer is
> already familiar with the usual useage of .def.

And with the meaning of plusAssignOp and return_self<>()

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com



From app26 at ast.cam.ac.uk  Sun Apr  9 23:24:13 2006
From: app26 at ast.cam.ac.uk (Andrew Pontzen)
Date: Sun, 9 Apr 2006 22:24:13 +0100 (BST)
Subject: [C++-sig] Newcomer - constructors not working as I'd expect
Message-ID: <Pine.LNX.4.64.0604092209530.13994@hermes-1.csi.cam.ac.uk>

Hi - apologies in advance if this is a question covered elsewhere. I did 
as thorough a search as I could for information before posting.

I am using boost.python (from boost package 1_33_1) to expose some C++ 
classes. I have a problem with constructors - here is a simple test case:


   class test1 {
   public:
     virtual string getText() const {
       return "Hello test1";
     }
     virtual ~test1() { } ;
   };

   class test2 : public test1 {
   public:
     virtual string getText() const {
       return "Hello test2";
     }
     virtual ~test2() { } ;
   };


   class test3 : public test2 {
   public:

     test3(const test1 &basedon) : ourobj(basedon) { };

     virtual string getText() const {
       return ourobj.getText();
     }

     const test1 & ourobj;

     virtual ~test3() { } ;
   };


   void printMessage(test1 & obj) {
     cout << obj.getText() << endl;
   }



   BOOST_PYTHON_MODULE(testme)
   {


     def("pm", &printMessage);

     class_<test1>("test1",init<>());
     class_<test2, bases<test1> >("test2",init<>());
     class_<test3, bases<test2> >("test3",init<test1>());
   }


In Python, things happen as expected until constructing a test3 object:

>>> a=testme.test1()
>>> b=testme.test2()
>>> testme.pm(a)
Hello test1
>>> testme.pm(b)
Hello test2
>>> c=testme.test3(a)
>>> testme.pm(c)
Segmentation fault


Using gdb, it seems the C++ constructor for test3 is never called, so the 
getText() call accesses an invalid reference.

I don't understand why this should be - clearly I've missed something!

TIA for any help.

Andrew

--
Andrew Pontzen
app26 at ast.cam.ac.uk

Institute of Astronomy
Madingley Road
Cambridge
CB3 0HA
UK


From dave at boost-consulting.com  Mon Apr 10 02:17:22 2006
From: dave at boost-consulting.com (David Abrahams)
Date: Sun, 09 Apr 2006 20:17:22 -0400
Subject: [C++-sig] Newcomer - constructors not working as I'd expect
References: <Pine.LNX.4.64.0604092209530.13994@hermes-1.csi.cam.ac.uk>
Message-ID: <m24q12wagd.fsf@boonen.local>

Andrew Pontzen <app26 at ast.cam.ac.uk> writes:

> I am using boost.python (from boost package 1_33_1) to expose some C++ 
> classes. I have a problem with constructors - here is a simple test case:

I can't see anything wrong with your code.  It's admittedly pretty
simple, but could you try to reduce it further?  It seems as though if
it's broken as is, you could remove one level of inheritance at least,
without changing anything.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com



From rwgk at yahoo.com  Mon Apr 10 06:26:43 2006
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Sun, 9 Apr 2006 21:26:43 -0700 (PDT)
Subject: [C++-sig] Newcomer - constructors not working as I'd expect
In-Reply-To: <Pine.LNX.4.64.0604092209530.13994@hermes-1.csi.cam.ac.uk>
Message-ID: <20060410042643.94482.qmail@web31501.mail.mud.yahoo.com>

General remark: use valgrind you have memory problems: http://valgrind.org/

It is really easy, except that you need a suppression file for Python 2.4 or
higher: Python-2.4*/Misc/valgrind-python.supp

Then: valgrind --tool=memcheck --suppressions=valgrind-python.supp python
script.py

--- Andrew Pontzen <app26 at ast.cam.ac.uk> wrote:

>      test3(const test1 &basedon) : ourobj(basedon) { };
> ...
>      class_<test3, bases<test2> >("test3",init<test1>());
>    }

This doesn't match. I'd try: init<const test1&>()

Even if this helps I'd be really worried. If you call methods of test3 after
the test1 instance was destroyed you'll get a crash. I'd consider using
boost::shared_ptr<test1> as a member of test3. If that's not an option a call
policy may be the way to achieve safety, but I am not sure since I avoid this
kind ...

>     const test1 & ourobj;

of unsafe C++, except in very low-level code with no direct connection to the
Python layer.

Cheers,
        Ralf


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 


From app26 at ast.cam.ac.uk  Mon Apr 10 10:56:40 2006
From: app26 at ast.cam.ac.uk (Andrew Pontzen)
Date: Mon, 10 Apr 2006 09:56:40 +0100 (BST)
Subject: [C++-sig] Newcomer - constructors not working as I'd expect
In-Reply-To: <20060410042643.94482.qmail@web31501.mail.mud.yahoo.com>
References: <20060410042643.94482.qmail@web31501.mail.mud.yahoo.com>
Message-ID: <Pine.LNX.4.64.0604100917180.23464@hermes-1.csi.cam.ac.uk>


On Sun, 9 Apr 2006, Ralf W. Grosse-Kunstleve wrote:

> Then: valgrind --tool=memcheck --suppressions=valgrind-python.supp python
> script.py

Thanks for the tip


> This doesn't match. I'd try: init<const test1&>()

You're quite right - this fixes it. In retrospect, it is quite obvious 
that what I was attempting to do would cause undefined behaviour!
(Although I'm suprised it didn't give me a compile-time error?)


> Even if this helps I'd be really worried. If you call methods of test3 after
> the test1 instance was destroyed you'll get a crash. I'd consider using
> boost::shared_ptr<test1> as a member of test3. If that's not an option a call
> policy may be the way to achieve safety, but I am not sure since I avoid this
> kind ...
>
>>     const test1 & ourobj;
>
> of unsafe C++, except in very low-level code with no direct connection to the
> Python layer.


I agree it's not ideal. However, in the case of my current project, I 
don't really see a way of avoiding keeping hold of the reference.
Of course in the real code, I'll be using a policy! (Is there any 
particular reason why using boost::shared_ptr would be preferable to using 
a policy? I'd rather avoid adding support for shared_ptr through my 
existing code if possible, but am unsure if it might have any advantages.)

Very grateful for the help,

Andrew


-- 
Andrew Pontzen
app26 at ast.cam.ac.uk

Institute of Astronomy
Madingley Road
Cambridge
CB3 0HA
UK


From app26 at ast.cam.ac.uk  Mon Apr 10 11:01:08 2006
From: app26 at ast.cam.ac.uk (Andrew Pontzen)
Date: Mon, 10 Apr 2006 10:01:08 +0100 (BST)
Subject: [C++-sig] Newcomer - constructors not working as I'd expect
In-Reply-To: <m24q12wagd.fsf@boonen.local>
References: <Pine.LNX.4.64.0604092209530.13994@hermes-1.csi.cam.ac.uk>
	<m24q12wagd.fsf@boonen.local>
Message-ID: <Pine.LNX.4.64.0604100957240.23464@hermes-1.csi.cam.ac.uk>

On Sun, 9 Apr 2006, David Abrahams wrote:

> I can't see anything wrong with your code.  It's admittedly pretty
> simple, but could you try to reduce it further?  It seems as though if
> it's broken as is, you could remove one level of inheritance at least,
> without changing anything.

You're right, removing test2 and even disinheriting test3 didn't 
change anything. It was just the constructor with the wrong call 
pattern which mucked things up (see Ralf's post.)

I originally constructed the example to explore another problem
which I cleared up for myself, but should have simplified it!

Thanks,

Andrew

-- 
Andrew Pontzen
app26 at ast.cam.ac.uk

Institute of Astronomy
Madingley Road
Cambridge
CB3 0HA
UK


From vishant at ama-inc.com  Mon Apr 10 16:11:52 2006
From: vishant at ama-inc.com (Vishant Shahnawaz)
Date: Mon, 10 Apr 2006 10:11:52 -0400
Subject: [C++-sig] Unloading module/Freeing memory
Message-ID: <200604101411.KAA18078@elephant.cnchost.com>

Hi all,

I have a C++ application in which I am importing a custom Python module and
calling a function from that module that parses a file and writes out
another file. After this the module does nothing. Hence I want to free up
the memory which is very large. I have tried several things but none have
worked. There are other modules that are imported that do other things and
hence I cannot unload all the modules that are loaded, ONLY the custom
parser module should be unloaded and the memory used while invoking the
function should also be freed. 

 

Here is the C++ method body.  I would sincerely appreciate any help I can
get here as soon as possible.

 

Vishant

 

{

  PyObject *p_moduleName = (PyString_FromString("cppwriter"));

      if(!p_moduleName)

      {

            throw NPVException("Cannot make string\n" + getPythonError());

      }

 

      // MEMORY USED HERE - 6-7 MB

      PyObject *p_module = (PyImport_Import(p_moduleName));

 

      if(!p_module)

      {

            throw NPVException("Cannot find module cppwriter\n" +
getPythonError());

      }

 

  PyObject *p_script = (PyString_FromString(script.c_str()));

  if(!p_script)

  {

    throw NPVException("Cannot make string\n" + getPythonError());

  }

 

  std::string implFile = outputDirectory + string(dllName + ".cpp");

  std::string headerFile = outputDirectory + string("cppwriterheader.h");

 

  PyObject *p_implFile = PyString_FromString(implFile.c_str());

  PyObject *p_headerFile = PyString_FromString(headerFile.c_str());

  PyObject *p_modelName = (PyString_FromString("CPPWriter"));

  PyObject *p_function = PyObject_GetAttrString(p_module,
"runCPPWriterFromString");

  if(!p_function)

  {

    throw NPVException("Cannot find function\n" + getPythonError());

  }

  PyObject *p_pyTuple = PyTuple_New(6);

  PyTuple_SetItem(p_pyTuple, 0, p_script);

  PyTuple_SetItem(p_pyTuple, 1, p_implFile);

  PyTuple_SetItem(p_pyTuple, 2, p_headerFile);

  PyTuple_SetItem(p_pyTuple, 3, p_modelName);

  PyTuple_SetItem(p_pyTuple, 4, Py_None);

  PyTuple_SetItem(p_pyTuple, 5, Py_None);

  Py_INCREF(Py_None);

  Py_INCREF(Py_None);

 

  // MEMORY USED HERE - 20 MB

  PyObject *p_pyRetVal = (PyObject_CallObject(p_function, p_pyTuple));

 

  std::string retString;

  

  try

  {

        retString = getPythonStdOut();

  }

  catch (...)

  {

            retString = "Debug unavailable";

  }

 

  std::string error = getPythonError();

 

  // TRYING TO GET RID OF MEMORY HERE

      Py_DECREF(p_moduleName);

      Py_DECREF(p_module);

  //Py_DECREF(p_script);

  //Py_DECREF(p_implFile);

  //Py_DECREF(p_headerFile);

  //Py_DECREF(p_modelName);

 

  Py_DECREF(p_function);

  Py_DECREF(p_pyTuple);

  Py_DECREF(p_pyRetVal);

 

    PyObject *modules = PyImport_GetModuleDict();

      _PyModule_Clear(p_function);

      _PyModule_Clear(p_module);

       PyDict_SetItemString(modules, "transformer2cpp", Py_None);

       int delsuccess = PyDict_DelItemString(modules, "transformer2cpp");

 

      PyErr_Clear();

  // END TRYING TO GET RID OF MEMORY

 

  if(error != "")

    throw NPVException(error);

 

  return retString;

}

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060410/f8919098/attachment.htm>

From dave at boost-consulting.com  Mon Apr 10 17:51:24 2006
From: dave at boost-consulting.com (David Abrahams)
Date: Mon, 10 Apr 2006 11:51:24 -0400
Subject: [C++-sig] Newcomer - constructors not working as I'd expect
References: <20060410042643.94482.qmail@web31501.mail.mud.yahoo.com>
	<Pine.LNX.4.64.0604100917180.23464@hermes-1.csi.cam.ac.uk>
Message-ID: <m264lhe8eb.fsf@boonen.local>

Andrew Pontzen <app26 at ast.cam.ac.uk> writes:

>> This doesn't match. I'd try: init<const test1&>()
>
> You're quite right - this fixes it. In retrospect, it is quite obvious 
> that what I was attempting to do would cause undefined behaviour!
> (Although I'm suprised it didn't give me a compile-time error?)

Why should it, and how could it?

A ctor taking a const reference to a temporary is perfectly
legitimate, as long as the ctor doesn't attempt to store it.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com



From dave at boost-consulting.com  Mon Apr 10 17:49:18 2006
From: dave at boost-consulting.com (David Abrahams)
Date: Mon, 10 Apr 2006 11:49:18 -0400
Subject: [C++-sig] Newcomer - constructors not working as I'd expect
References: <Pine.LNX.4.64.0604092209530.13994@hermes-1.csi.cam.ac.uk>
	<20060410042643.94482.qmail@web31501.mail.mud.yahoo.com>
Message-ID: <m2bqv9e8ht.fsf@boonen.local>

"Ralf W. Grosse-Kunstleve" <rwgk at yahoo.com> writes:

> General remark: use valgrind you have memory problems: http://valgrind.org/
>
> It is really easy, except that you need a suppression file for Python 2.4 or
> higher: Python-2.4*/Misc/valgrind-python.supp
>
> Then: valgrind --tool=memcheck --suppressions=valgrind-python.supp python
> script.py
>
> --- Andrew Pontzen <app26 at ast.cam.ac.uk> wrote:
>
>>      test3(const test1 &basedon) : ourobj(basedon) { };
>> ...
>>      class_<test3, bases<test2> >("test3",init<test1>());
>>    }
>
> This doesn't match. I'd try: init<const test1&>()

Ah, that explains everything.


-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com



From abhi at qualcomm.com  Mon Apr 10 21:34:44 2006
From: abhi at qualcomm.com (Abhi)
Date: Mon, 10 Apr 2006 12:34:44 -0700
Subject: [C++-sig] using the same module in different files
Message-ID: <6CD9E7723ED55130B7FAEE2F@[10.30.5.17]>

I have 2 files
- A.h which I want to wrap in a file boost_A.cpp
- and B.h which I want to wrap in a file boost_B.cpp

>>> File A.h

class A
{
  public:
    void fooA();
};

>>> File B.h

class B
{
  public:
    void fooB();
};


File boost_A.cpp wraps class A, while file boost_B.cpp wraps class B.

I want both these files to expose methods in the same module, ie, I want to 
"semantically" use BOOST_PYTHON_MODULE(common) for both these files, so 
that all the methods exposed from boost_A.cpp and boost_B.cpp get imported 
by doing

>> import common
>> a = common.A()
>> b = common.B()


Is this possible?

thanks
- Abhi






From seefeld at sympatico.ca  Mon Apr 10 22:07:50 2006
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Mon, 10 Apr 2006 16:07:50 -0400
Subject: [C++-sig] using the same module in different files
In-Reply-To: <6CD9E7723ED55130B7FAEE2F@[10.30.5.17]>
References: <6CD9E7723ED55130B7FAEE2F@[10.30.5.17]>
Message-ID: <443ABB16.90600@sympatico.ca>

Abhi wrote:

> I want both these files to expose methods in the same module, ie, I want to 
> "semantically" use BOOST_PYTHON_MODULE(common) for both these files, so 
> that all the methods exposed from boost_A.cpp and boost_B.cpp get imported 
> by doing
> 
>>> import common
>>> a = common.A()
>>> b = common.B()
> 
> 
> Is this possible?

Wouldn't it be simpler to integrate both from within python, using

# inside common.py
import A, B


?

Regards,
		Stefan


From abhi at qualcomm.com  Mon Apr 10 22:57:27 2006
From: abhi at qualcomm.com (Abhi)
Date: Mon, 10 Apr 2006 13:57:27 -0700
Subject: [C++-sig] using the same module in different files
In-Reply-To: <443ABB16.90600@sympatico.ca>
References: <6CD9E7723ED55130B7FAEE2F@[10.30.5.17]>
	<443ABB16.90600@sympatico.ca>
Message-ID: <D16F326AA411DF788DA91669@[10.30.5.17]>

Well the problem is that we have a large number of files that need to be 
wrapped by different people. The methods in these files are tightly 
coupled, ie, you will need to use methods from multiple files to execute 
code in python.

Hence
1) It is not possible to put them in one boost wrapper file.
2) Also it is not good to have a long import line in python every time 
methods from these files need to be included.

thanks
- Abhi



--On Monday, April 10, 2006 4:07 PM -0400 Stefan Seefeld 
<seefeld at sympatico.ca> wrote:

> Abhi wrote:
>
>> I want both these files to expose methods in the same module, ie, I want
>> to  "semantically" use BOOST_PYTHON_MODULE(common) for both these files,
>> so  that all the methods exposed from boost_A.cpp and boost_B.cpp get
>> imported  by doing
>>
>>>> import common
>>>> a = common.A()
>>>> b = common.B()
>>
>>
>> Is this possible?
>
> Wouldn't it be simpler to integrate both from within python, using
>
># inside common.py
> import A, B
>
>
> ?
>
> Regards,
> 		Stefan
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig






From seefeld at sympatico.ca  Mon Apr 10 23:16:47 2006
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Mon, 10 Apr 2006 17:16:47 -0400
Subject: [C++-sig] using the same module in different files
In-Reply-To: <D16F326AA411DF788DA91669@[10.30.5.17]>
References: <6CD9E7723ED55130B7FAEE2F@[10.30.5.17]>	<443ABB16.90600@sympatico.ca>
	<D16F326AA411DF788DA91669@[10.30.5.17]>
Message-ID: <443ACB3F.3030403@sympatico.ca>

Abhi wrote:
> Well the problem is that we have a large number of files that need to be 
> wrapped by different people. The methods in these files are tightly 
> coupled, ie, you will need to use methods from multiple files to execute 
> code in python.

The types and methods you wrap may be coupled, but do the wrappers have
to be coupled (at compile-time) ? In fact, I'v already split large
wrapper files into smaller ones to cut down on compile- and link-time.
You could have a set of python extension modules all using / wrapping
from the same shared library underneath.

> Hence
> 1) It is not possible to put them in one boost wrapper file.
> 2) Also it is not good to have a long import line in python every time 
> methods from these files need to be included.

I'm not sure I understand what you mean by 'every time'. The common.py
module I suggested would only be written once. Users would then only
call

import common

and would see exactly the same thing as with your attempted solution.
See http://boost.org/libs/python/doc/tutorial/doc/html/python/techniques.html#python.creating_packages
for more on packaging extension modules.

HTH,
		Stefan


From rwgk at yahoo.com  Tue Apr 11 01:54:19 2006
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Mon, 10 Apr 2006 16:54:19 -0700 (PDT)
Subject: [C++-sig] Newcomer - constructors not working as I'd expect
In-Reply-To: <Pine.LNX.4.64.0604100917180.23464@hermes-1.csi.cam.ac.uk>
Message-ID: <20060410235419.94328.qmail@web31515.mail.mud.yahoo.com>

--- Andrew Pontzen <app26 at ast.cam.ac.uk> wrote:
> I agree it's not ideal. However, in the case of my current project, I 
> don't really see a way of avoiding keeping hold of the reference.
> Of course in the real code, I'll be using a policy! (Is there any 
> particular reason why using boost::shared_ptr would be preferable to using 
> a policy?

I find using shared_ptr easier than figuring out which one is the right call
policy.

> I'd rather avoid adding support for shared_ptr through my 
> existing code if possible, but am unsure if it might have any advantages.)

With shared_ptr your C++ code is also safe, not just the Python interface.

Cheers,
        Ralf


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 


From rwgk at yahoo.com  Tue Apr 11 02:13:53 2006
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Mon, 10 Apr 2006 17:13:53 -0700 (PDT)
Subject: [C++-sig] using the same module in different files
In-Reply-To: <6CD9E7723ED55130B7FAEE2F@[10.30.5.17]>
Message-ID: <20060411001353.14642.qmail@web31510.mail.mud.yahoo.com>

--- Abhi <abhi at qualcomm.com> wrote:

> I have 2 files
> - A.h which I want to wrap in a file boost_A.cpp
> - and B.h which I want to wrap in a file boost_B.cpp
> 
> >>> File A.h
> 
> class A
> {
>   public:
>     void fooA();
> };
> 
> >>> File B.h
> 
> class B
> {
>   public:
>     void fooB();
> };
> 
> 
> File boost_A.cpp wraps class A, while file boost_B.cpp wraps class B.
> 
> I want both these files to expose methods in the same module, ie, I want to 
> "semantically" use BOOST_PYTHON_MODULE(common) for both these files, so 
> that all the methods exposed from boost_A.cpp and boost_B.cpp get imported 
> by doing
> 
> >> import common
> >> a = common.A()
> >> b = common.B()
> 
> 
> Is this possible?

IIUC, yes. Create a file boost_common.cpp, e.g.:

void wrap_A();
void wrap_B();

BOOST_PYTHON_MODULE(common)
{
  wrap_A();
  wrap_B();
}


boost_A.cpp would look like this:

void
wrap_A()
{
  class_<A>("A")
  // ...
  ;
}

And boost_B.cpp the same with A replaced by B. Compile the three files
(boost_common.cpp, boost_A.cpp, boost_B.cpp) separately, but link like this:

g++ -shared -o common.so boost_common.o boost_A.o boost_B.o -lboost_python

HTH,
        Ralf


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 


From ljreeder77 at hotmail.com  Tue Apr 11 07:20:19 2006
From: ljreeder77 at hotmail.com (Jason Reeder)
Date: Tue, 11 Apr 2006 00:20:19 -0500
Subject: [C++-sig] (no subject)
Message-ID: <BAY102-DAV9DD7C0CF68D1C1DA4CE56C8CD0@phx.gbl>

Dunno if this is still a valid email for you. Its Jason!! Reply if this is
Ben Collar.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060411/039949ae/attachment.htm>

From roman.yakovenko at gmail.com  Tue Apr 11 07:48:06 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 11 Apr 2006 07:48:06 +0200
Subject: [C++-sig] using the same module in different files
In-Reply-To: <20060411001353.14642.qmail@web31510.mail.mud.yahoo.com>
References: <6CD9E7723ED55130B7FAEE2F@10.30.5.17>
	<20060411001353.14642.qmail@web31510.mail.mud.yahoo.com>
Message-ID: <7465b6170604102248v2ad59867lfe05bc1bb577e08f@mail.gmail.com>

On 4/11/06, Ralf W. Grosse-Kunstleve <rwgk at yahoo.com> wrote:
> --- Abhi <abhi at qualcomm.com> wrote:
>
> > I have 2 files
> > - A.h which I want to wrap in a file boost_A.cpp
> > - and B.h which I want to wrap in a file boost_B.cpp
> >
> > >>> File A.h
> >
> > class A
> > {
> >   public:
> >     void fooA();
> > };
> >
> > >>> File B.h
> >
> > class B
> > {
> >   public:
> >     void fooB();
> > };
> >
> >
> > File boost_A.cpp wraps class A, while file boost_B.cpp wraps class B.
> >
> > I want both these files to expose methods in the same module, ie, I want to
> > "semantically" use BOOST_PYTHON_MODULE(common) for both these files, so
> > that all the methods exposed from boost_A.cpp and boost_B.cpp get imported
> > by doing
> >
> > >> import common
> > >> a = common.A()
> > >> b = common.B()
> >
> >
> > Is this possible?
>
> IIUC, yes. Create a file boost_common.cpp, e.g.:
>
> void wrap_A();
> void wrap_B();
>
> BOOST_PYTHON_MODULE(common)
> {
>   wrap_A();
>   wrap_B();
> }
>
>
> boost_A.cpp would look like this:
>
> void
> wrap_A()
> {
>   class_<A>("A")
>   // ...
>   ;
> }
>
> And boost_B.cpp the same with A replaced by B. Compile the three files
> (boost_common.cpp, boost_A.cpp, boost_B.cpp) separately, but link like this:

An other approach it to create common package in pure Python that will hide
implementation/building blocks.

> HTH,
>         Ralf


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From guillaume.desticourt at gmail.com  Tue Apr 11 18:07:34 2006
From: guillaume.desticourt at gmail.com (Guillaume Desticourt)
Date: Tue, 11 Apr 2006 17:07:34 +0100
Subject: [C++-sig] wrapping an abstract class with out parameters using
	boost.python
Message-ID: <609a651c0604110907y13d1ecd5qe2135642956a8866@mail.gmail.com>

hello,

i try to wrap an abstract class so i can create class that inherit from it
in python
but at run time i got the following error message:
RuntimeError: extension class wrapper for base class
pimp::gfx2d::DisplayData has not been created yet

my class is something like that:

class DisplayData : public IdData, public LoadData, public
Subject<DisplayData>
{
public:
 virtual void accept(DisplayDataVisitor &visitor) = 0;
 virtual void getZone(Zone &zone, long time) const = 0;
// ...
}

so i wrote a class:

class DisplayDataWrap :
  public DisplayData,
  public wrapper<DisplayData>;

to encapsulate the DisplayData::accept, i ve defined:

  void DisplayDataWrap::accept(DisplayDataVisitor &visitor)
  {
    this->get_override("accept")(visitor);
  }

and in the module definition i do:
 class_<DisplayDataWrap, boost::noncopyable>("PyDisplayData", no_init)
    .def("accept", pure_virtual(&DisplayData::accept))
    ;

the problem is that i have no idea how to encapsulate:
the DisplaData::getZone method...
i tried
  Zone getZone(void) const
  {
    Zone zone;
    this->get_override("getZone")(zone);
    return zone;
  }

 Zone (DisplayDataWrap::*getZone1)(void) const = &DisplayDataWrap::getZone;

  class_<DisplayDataWrap, boost::noncopyable>("PyDisplayData", no_init)
    .def("getZone", getZone1) ;

and then i get the error messge
RuntimeError: extension class wrapper for base class
pimp::gfx2d::DisplayData has not been created yet
and i am not even sure i ll be able to override the getZone in python...

any help would be appreciated, thanks.

--
Guillaume Desticourt
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060411/65d14232/attachment.htm>

From ndbecker2 at gmail.com  Wed Apr 12 15:59:13 2006
From: ndbecker2 at gmail.com (Neal Becker)
Date: Wed, 12 Apr 2006 09:59:13 -0400
Subject: [C++-sig] IMPORTANT 2.5 API changes for C Extension Modules and
	Embedders
Message-ID: <e1j130$n3j$3@sea.gmane.org>

Hopefully you've seen this notice.  I'm wondering if anyone has evaluated
it's possible effect on boost::python?



From sbarranx at modulonet.fr  Wed Apr 12 15:59:47 2006
From: sbarranx at modulonet.fr (=?iso-8859-1?b?U3TpcGhhbmU=?= Barranx)
Date: Wed, 12 Apr 2006 15:59:47 +0200
Subject: [C++-sig] Big error in GCCXML compilation :: segmentation fault
	termination
Message-ID: <1144850387.443d07d3c3aab@webmail.modulonet.fr>

Hi, I'm trying to wrap a C++ project for motion correction. Like
in tutorials, I made the environment and py++ run scripts. But
when I try to run the script, I got tons of error but obviously
only two major problems. Here is the beginning of the error
log:

Traceback (most recent call last):
  File "D:\MotionCorrection\Foundation\wrap\pypp_Foundation.py",
line 112, in ?
    export()
  File "D:\MotionCorrection\Foundation\wrap\pypp_Foundation.py",
line 107, in export
    exporter.create()
  File "D:\MotionCorrection\Foundation\wrap\pypp_Foundation.py",
line 98, in create
    decls = self.read_declarations()
  File "D:\MotionCorrection\Foundation\wrap\pypp_Foundation.py",
line 65, in read_declarations
    return reader.read_files( self.__header_files )
  File "C:\Program
Files\Python24\Lib\site-packages\pygccxml\parser\project_reader.py",
line 61, in read_files
    return self.__parse_file_by_file(files)
  File "C:\Program
Files\Python24\Lib\site-packages\pygccxml\parser\project_reader.py",
line 82, in __parse_file_by_file
    decls = reader.read_file( header )
  File "C:\Program
Files\Python24\Lib\site-packages\pygccxml\parser\source_reader.py",
line 140, in read_file
    raise error
gccxml_runtime_error_t: Error occured while running GCC-XML: In
file included from
D:/MotionCorrection/Foundation/VectorField.h:24,
                 from D:/MotionCorrection/Foundation/Image.h:23:
D:/MotionCorrection/Foundation/Matrix.h: In constructor `
   Xform2DParams::Xform2DParams(int)':
D:/MotionCorrection/Foundation/Matrix.h:40: error: `fill_n' 
   undeclared in namespace `std'
D:/MotionCorrection/Foundation/Matrix.h: In copy constructor `
   Xform2DParams::Xform2DParams(const Xform2DParams&)':
D:/MotionCorrection/Foundation/Matrix.h:46: error: `copy' 
   undeclared in namespace `std'

//** ?? 'fill_n' and 'copy' are actually members of the standard
library **//

In file included from
D:/MotionCorrection/Foundation/VectorField.h:28,
                 from D:/MotionCorrection/Foundation/Image.h:23:
C:/Program Files/Microsoft Visual Studio .NET
2003/Vc7/Include/valarray: At global scope:
C:/Program Files/Microsoft Visual Studio .NET
2003/Vc7/Include/valarray:30: error: `
   size_t' was not declared in this scope
C:/Program Files/Microsoft Visual Studio .NET
2003/Vc7/Include/valarray:30: error: template
   argument 1 is invalid

//** I stop here but there is hundreds of errors next but only in
'valarray' **//

So, can anybody help me? 

Here is my run script:


import os, time
from environment import settings
from pygccxml import parser
from pygccxml import declarations
from pyplusplus import code_creators
from pyplusplus import module_creator
from pyplusplus import file_writers

def full_path(path, file):
    return os.path.join( path, file )  

def identify_call_policies( decl ):
    build_in_resolver = module_creator.build_in_resolver_t()
    policy = build_in_resolver( decl )
    if policy:
        return policy

    if isinstance( decl, declarations.member_operator_t ) \
       and decl.symbol == '()' \
       and decl.parent.name == 'Foundation':
        return code_creators.return_internal_reference()
    return None

class Foundation_exporter_t:
    def __init__(self):
        path = settings.Foundation_path
        self.__header_files = []
        self.__header_files.append(full_path(path,
'Foundation.h'))
        self.__header_files.append(full_path(path, 'Image.h'))
        self.__header_files.append(full_path(path,
'LSPoly2Approx.h'))
        self.__header_files.append(full_path(path, 'Mask.h'))
        self.__header_files.append(full_path(path, 'Matrix.h'))
        self.__header_files.append(full_path(path,
'Polynomial1.h'))
        self.__header_files.append(full_path(path,
'Polynomial2.h'))
        self.__header_files.append(full_path(path, 'RecPar.h'))
        self.__header_files.append(full_path(path,
'SOUnwrapper.h'))
        self.__header_files.append(full_path(path,
'TUnwrapper.h'))
        self.__header_files.append(full_path(path, 'Vector2.h'))
        self.__header_files.append(full_path(path,
'VectorField.h'))
        self.__header_files.append(full_path(path,
'WLSPoly2Approx.h'))
    
    def read_declarations(self):
        config = parser.config_t(
gccxml_path=settings.gccxml_path
                                  ,
working_directory=settings.MotionCorrection_path
                                  ,
include_paths=settings.vtk_paths)
        reader = parser.project_reader_t( config )
        return reader.read_files( self.__header_files )
    
    def filter_declarations(self, decls ):
        return declarations.filtering.by_location( decls,
[settings.MotionCorrection_path] )
    
    def create_extmodule(self, decls):
        creator = module_creator.creator_t( decls=decls
                                            ,
module_name=settings.Foundation.module_name
                                            ,
create_castinig_constructor=True
                                            ,
call_policies_resolver_=identify_call_policies) 
        return creator.create()
    
    def customize_extmodule( self, extmodule ):
        global license
        extmodule.license = license
        extmodule.std_directories.append( settings.boost_path )
        extmodule.std_directories.extend( settings.vtk_paths )
        #butifying include code generation
        extmodule.user_defined_directories.append(
settings.MotionCorrection_path )
        extmodule.user_defined_directories.append(
settings.Foundation_path )
        extmodule.user_defined_directories.append(
settings.working_dir )
        extmodule.user_defined_directories.extend(
config.include_paths )
        extmodule.precompiled_header = 'boost/python.hpp'
        #Adding headers as first includes. It comes because of
internal structure of Foundation
        extmodule.adopt_creator( code_creators.include_t(
header=self.__header_files ), 2 )
        #body_index = extmodule.creators.index( extmodule.body )
        
    def write_files( self, extmodule ):
        file_writers.write_file(extmodule
                                ,
os.path.join(settings.Foundation.generated_files_dir
                                               ,
settings.Foundation.module_name + '.cpp') )

    def create(self):
        decls = self.read_declarations()
        self.filter_declarations(decls)
        extmodule = self.create_extmodule( decls )
        self.customize_extmodule( extmodule )
        self.write_files( extmodule )

def export():
    start = time.clock()
    exporter = Foundation_exporter_t()
    exporter.create()
    finish = time.clock() - start
    print '%0.2f seconds' % finish

if __name__ == '__main__':
    export()
    print 'done'

//** end of the script **//

Where is my mistake?? Did I forgot an inclusion? Why the GCCXML
compiler doesn't recognize 'fill_n' and 'copy'? What's wrong
with 'valarray' file?

Thanks for helping




From seefeld at sympatico.ca  Wed Apr 12 16:08:43 2006
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Wed, 12 Apr 2006 10:08:43 -0400
Subject: [C++-sig] IMPORTANT 2.5 API changes for C Extension Modules and
 Embedders
In-Reply-To: <e1j130$n3j$3@sea.gmane.org>
References: <e1j130$n3j$3@sea.gmane.org>
Message-ID: <443D09EB.7060703@sympatico.ca>

Neal Becker wrote:
> Hopefully you've seen this notice.

Which notice ? Do you have a reference ?

Thanks,
		Stefan



From rwgk at yahoo.com  Wed Apr 12 17:59:45 2006
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Wed, 12 Apr 2006 08:59:45 -0700 (PDT)
Subject: [C++-sig] IMPORTANT 2.5 API changes for C Extension Modules and
	Embedders
In-Reply-To: <e1j130$n3j$3@sea.gmane.org>
Message-ID: <20060412155945.68663.qmail@web31506.mail.mud.yahoo.com>

--- Neal Becker <ndbecker2 at gmail.com> wrote:

> Hopefully you've seen this notice.  I'm wondering if anyone has evaluated
> it's possible effect on boost::python?

I've successfully tested Python 2.5a1 with Boost.Python after locally making a
few small changes related to http://www.python.org/dev/peps/pep-0353/ . I also
used Fredrik Lundh's scanner (see PEP) to pin-point places in Boost.Python that
need attention, but I don't have the time right now to go through
systematically (I think it just needs a few int -> Py_ssize_t changes). Any
help would be highly appreciated. Attached are my minimal changes.

Cheers,
        Ralf

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: py25a1_bpl_minimal_patch
Type: application/octet-stream
Size: 1658 bytes
Desc: 2207447082-py25a1_bpl_minimal_patch
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060412/8ab1447f/attachment.obj>

From roman.yakovenko at gmail.com  Wed Apr 12 20:48:13 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Wed, 12 Apr 2006 21:48:13 +0300
Subject: [C++-sig] Big error in GCCXML compilation :: segmentation fault
	termination
In-Reply-To: <1144850387.443d07d3c3aab@webmail.modulonet.fr>
References: <1144850387.443d07d3c3aab@webmail.modulonet.fr>
Message-ID: <7465b6170604121148t5fc1ae03he188eedd7c1709a3@mail.gmail.com>

On 4/12/06, St?phane Barranx <sbarranx at modulonet.fr> wrote:
> Hi, I'm trying to wrap a C++ project for motion correction. Like
> in tutorials, I made the environment and py++ run scripts. But
> when I try to run the script, I got tons of error but obviously
> only two major problems.

Hi.

First of all you should treat GCC-XML as just an other compiler. So,
basically before
you can use pyplusplus/pygccxml GCC-XML should be able to compile your code.

There are 2 ways you can fix the situation. The simple one is next. When GCC-XML
compiles a code it defines new preprocessed definition  __GCCXML__. So you can
use it in order to fix your code.

Now about your script. Nice work and you did it almost without documentation!
But there is small problem with it. You use old version of pyplusplus. Few days
ago I put at SourceForge next version of pyplusplus and pygccxml. It
simplifies a
lot the usage of pyplusplus.

If you need help with porting your script to new version, I will be
glad to help you.
Anyway, there is a mailing list for pyplusplus and pygccxml:
http://lists.sourceforge.net/lists/listinfo/pygccxml-development.

Anonymous CVS access is not synchronized( SourceForge problems ), so I
attach tutorials here.

If you still need help, consider to use pyplusplus/pygccxml mailing list

Thanks

--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: hello_world.hpp
Type: text/x-c++hdr
Size: 1357 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060412/48503ebc/attachment.hpp>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: generate_code.py
Type: text/x-python
Size: 2325 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060412/48503ebc/attachment.py>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: module_builder.rest
Type: application/octet-stream
Size: 5579 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060412/48503ebc/attachment.obj>

From tkirke at gmail.com  Thu Apr 13 22:16:18 2006
From: tkirke at gmail.com (Tony Kirke)
Date: Thu, 13 Apr 2006 13:16:18 -0700
Subject: [C++-sig] Wrapping pointers
In-Reply-To: <200604052105.33008.dominic.sacre@gmx.de>
References: <200604052105.33008.dominic.sacre@gmx.de>
Message-ID: <4dcfdcab0604131316q89ceaccgc2c3a17f6da2af08@mail.gmail.com>

Did you get an answer to this?
I am trying to figure out the same thing except my pointer is to an object
of a different class instead of  the same class

On 4/5/06, Dominic Sacr? <dominic.sacre at gmx.de> wrote:
>
> Hi,
>
> I'm having trouble exporting C++ pointers to Python. My code looks like
> this:
>
> class A
> {
>     A *p;
>     ...
> };
>
> BOOST_PYTHON_MODULE(test)
> {
>     class_<A>("A", init<>())
>         .def_readonly("p", &A::p)
>         ...
>     ;
> }
>
> Now every time I try to access A::p from Python I get this error:
>
> TypeError: No to_python (by-value) converter found for C++ type: A*
>
> I searched the list archives and found a message about the same issue,
> where the only suggestion was to use
>
> .def_readonly("p", make_getter(&A::p, return_internal_reference<1>()))
>
> instead, but this doesn't help either. On the contrary, this causes a
> segfault every time I use A::p (although p points to a valid object, at
> least in C++).
> So what should I do to access this pointer from Python?
>
> Thanks,
>     Dominic
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060413/4be851e0/attachment.htm>

From dominic.sacre at gmx.de  Fri Apr 14 02:18:53 2006
From: dominic.sacre at gmx.de (Dominic =?iso-8859-1?q?Sacr=E9?=)
Date: Fri, 14 Apr 2006 02:18:53 +0200
Subject: [C++-sig] Wrapping pointers
In-Reply-To: <4dcfdcab0604131316q89ceaccgc2c3a17f6da2af08@mail.gmail.com>
References: <200604052105.33008.dominic.sacre@gmx.de>
	<4dcfdcab0604131316q89ceaccgc2c3a17f6da2af08@mail.gmail.com>
Message-ID: <200604140218.53896.dominic.sacre@gmx.de>

On Thursday, 13. April 2006 22:16, Tony Kirke wrote:
> Did you get an answer to this?
> I am trying to figure out the same thing except my pointer is to an
> object of a different class instead of  the same class

Nope, no reply. So I'm still looking for hints how to do this.


> On 4/5/06, Dominic Sacr? <dominic.sacre at gmx.de> wrote:
> > Hi,
> >
> > I'm having trouble exporting C++ pointers to Python. My code looks
> > like this:
> >
> > class A
> > {
> >     A *p;
> >     ...
> > };
> >
> > BOOST_PYTHON_MODULE(test)
> > {
> >     class_<A>("A", init<>())
> >         .def_readonly("p", &A::p)
> >         ...
> >     ;
> > }
> >
> > Now every time I try to access A::p from Python I get this error:
> >
> > TypeError: No to_python (by-value) converter found for C++ type: A*


From rwgk at yahoo.com  Fri Apr 14 03:01:38 2006
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Thu, 13 Apr 2006 18:01:38 -0700 (PDT)
Subject: [C++-sig] Wrapping pointers
In-Reply-To: <200604140218.53896.dominic.sacre@gmx.de>
Message-ID: <20060414010138.83299.qmail@web31506.mail.mud.yahoo.com>

--- Dominic Sacr? <dominic.sacre at gmx.de> wrote:

> On Thursday, 13. April 2006 22:16, Tony Kirke wrote:
> > Did you get an answer to this?
> > I am trying to figure out the same thing except my pointer is to an
> > object of a different class instead of  the same class
> 
> Nope, no reply. So I'm still looking for hints how to do this.

CAVEAT: Raw pointers are intrinsically difficult to work with, terribly unsafe,
and therefore not easily compatible with the safe Python layer. I avoid them
completely (except in core code not exposed via public interfaces), i.e. I have
no working experience related to your question.

FWIW: You may want to try

  .add_property("p", make_getter(&A::p, xxx));

where xxx is a return value policy you have to find out about.

> > On 4/5/06, Dominic Sacr? <dominic.sacre at gmx.de> wrote:
> > > Hi,
> > >
> > > I'm having trouble exporting C++ pointers to Python. My code looks
> > > like this:
> > >
> > > class A
> > > {
> > >     A *p;
> > >     ...
> > > };
> > >
> > > BOOST_PYTHON_MODULE(test)
> > > {
> > >     class_<A>("A", init<>())
> > >         .def_readonly("p", &A::p)
> > >         ...
> > >     ;
> > > }
> > >
> > > Now every time I try to access A::p from Python I get this error:
> > >
> > > TypeError: No to_python (by-value) converter found for C++ type: A*


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 


From faheem at email.unc.edu  Sat Apr 15 04:26:53 2006
From: faheem at email.unc.edu (Faheem Mitha)
Date: Fri, 14 Apr 2006 22:26:53 -0400 (EDT)
Subject: [C++-sig] wrapping function objects
Message-ID: <Pine.LNX.4.62.0604142219270.12285@Chrestomanci>


Hi,

Is there any way I can expose function objects to python, in such a way 
that I can call the associated operator() as a method in python?

The only thing I could find that looked relavant was

http://www.boost.org/libs/python/todo.html#wrapping-function-objects

but I didn't understand how to use the method described in the linked 
message.

Thanks in advance. Please cc me on any reply. Thanks.

                                                            Faheem.


From ludwig at fh-worms.de  Sat Apr 15 13:03:59 2006
From: ludwig at fh-worms.de (Christoph Ludwig)
Date: Sat, 15 Apr 2006 13:03:59 +0200
Subject: [C++-sig] GCC version compatibility
In-Reply-To: <uacgo49s7.fsf@boost-consulting.com>
References: <42CEF948.3010908@v.loewis.de>
	<20050709102010.GA3836@lap200.cdc.informatik.tu-darmstadt.de>
	<42D0D215.9000708@v.loewis.de>
	<20050710125458.GA3587@lap200.cdc.informatik.tu-darmstadt.de>
	<42D15DB2.3020300@v.loewis.de>
	<20050716101357.GC3607@lap200.cdc.informatik.tu-darmstadt.de>
	<20051012120917.GA11058@lap200.cdc.informatik.tu-darmstadt.de>
	<ufyqg5r4t.fsf@boost-consulting.com>
	<20051101161534.GC8657@lap200.cdc.informatik.tu-darmstadt.de>
	<uacgo49s7.fsf@boost-consulting.com>
Message-ID: <20060415110358.GA25764@castellio.textgrid.it.fh-worms.de>

Hi,

On Tue, Nov 01, 2005 at 11:23:20AM -0500, David Abrahams wrote:
> Christoph Ludwig <cludwig at cdc.informatik.tu-darmstadt.de> writes:
> 
> > On Tue, Nov 01, 2005 at 10:23:14AM -0500, David Abrahams wrote:
> >> Christoph Ludwig <cludwig at cdc.informatik.tu-darmstadt.de> writes:
> >> > I finally had the spare time to look into this problem again and submitted
> >> > patch #1324762. The proposed patch implements the following:
> >> 
> >> <snip awesome patch>
> >> 
> >> Christoph, thanks so much for doing this! 
> >
> > You are welcome. It was pretty straight forward once I realized how the C++
> > compiler is communicated from configure to distutils. But the patch has not
> > been applied to Python's CVS yet; I don't know how the Python developers think
> > about it. 
> 
> Hmm, maybe I should post encouraging someone to look at this.

FYI, the patch got in.

Regards

Christoph


-- 
FH Worms - University of Applied Sciences
Fachbereich Informatik / Telekommunikation
Erenburgerstr. 19, 67549 Worms, Germany
-------------- next part --------------
An embedded message was scrubbed...
From: "SourceForge.net" <noreply at sourceforge.net>
Subject: [ python-Patches-1324762 ] Compiling and linking main() with C++
	compiler
Date: Fri, 14 Apr 2006 07:37:09 -0700
Size: 9470
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060415/853dd9dc/attachment.eml>

From dave at boost-consulting.com  Sat Apr 15 15:12:24 2006
From: dave at boost-consulting.com (David Abrahams)
Date: Sat, 15 Apr 2006 09:12:24 -0400
Subject: [C++-sig] GCC version compatibility
In-Reply-To: <20060415110358.GA25764@castellio.textgrid.it.fh-worms.de>
	(Christoph Ludwig's message of "Sat, 15 Apr 2006 13:03:59 +0200")
References: <42CEF948.3010908@v.loewis.de>
	<20050709102010.GA3836@lap200.cdc.informatik.tu-darmstadt.de>
	<42D0D215.9000708@v.loewis.de>
	<20050710125458.GA3587@lap200.cdc.informatik.tu-darmstadt.de>
	<42D15DB2.3020300@v.loewis.de>
	<20050716101357.GC3607@lap200.cdc.informatik.tu-darmstadt.de>
	<20051012120917.GA11058@lap200.cdc.informatik.tu-darmstadt.de>
	<ufyqg5r4t.fsf@boost-consulting.com>
	<20051101161534.GC8657@lap200.cdc.informatik.tu-darmstadt.de>
	<uacgo49s7.fsf@boost-consulting.com>
	<20060415110358.GA25764@castellio.textgrid.it.fh-worms.de>
Message-ID: <m27j5rx9s7.fsf@boonen.local>

Christoph Ludwig <ludwig at fh-worms.de> writes:

> Hi,
>
> On Tue, Nov 01, 2005 at 11:23:20AM -0500, David Abrahams wrote:
>> Christoph Ludwig <cludwig at cdc.informatik.tu-darmstadt.de> writes:
>> 
>> > On Tue, Nov 01, 2005 at 10:23:14AM -0500, David Abrahams wrote:
>> >> Christoph Ludwig <cludwig at cdc.informatik.tu-darmstadt.de> writes:
>> >> > I finally had the spare time to look into this problem again and submitted
>> >> > patch #1324762. The proposed patch implements the following:
>> >> 
>> >> <snip awesome patch>
>> >> 
>> >> Christoph, thanks so much for doing this! 
>> >
>> > You are welcome. It was pretty straight forward once I realized how the C++
>> > compiler is communicated from configure to distutils. But the patch has not
>> > been applied to Python's CVS yet; I don't know how the Python developers think
>> > about it. 
>> 
>> Hmm, maybe I should post encouraging someone to look at this.
>
> FYI, the patch got in.

Great!

IIRC, this had some practical consequences W.R.T. Boost.Python and/or
debug builds of Python, but I can't quite remember what the upshot
was.  Can you?

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com


From ludwig at fh-worms.de  Sat Apr 15 18:48:09 2006
From: ludwig at fh-worms.de (Christoph Ludwig)
Date: Sat, 15 Apr 2006 18:48:09 +0200
Subject: [C++-sig] GCC version compatibility
In-Reply-To: <m27j5rx9s7.fsf@boonen.local>
References: <42D0D215.9000708@v.loewis.de>
	<20050710125458.GA3587@lap200.cdc.informatik.tu-darmstadt.de>
	<42D15DB2.3020300@v.loewis.de>
	<20050716101357.GC3607@lap200.cdc.informatik.tu-darmstadt.de>
	<20051012120917.GA11058@lap200.cdc.informatik.tu-darmstadt.de>
	<ufyqg5r4t.fsf@boost-consulting.com>
	<20051101161534.GC8657@lap200.cdc.informatik.tu-darmstadt.de>
	<uacgo49s7.fsf@boost-consulting.com>
	<20060415110358.GA25764@castellio.textgrid.it.fh-worms.de>
	<m27j5rx9s7.fsf@boonen.local>
Message-ID: <20060415164808.GA26536@castellio.textgrid.it.fh-worms.de>

On Sat, Apr 15, 2006 at 09:12:24AM -0400, David Abrahams wrote:
> Christoph Ludwig <ludwig at fh-worms.de> writes:
> > On Tue, Nov 01, 2005 at 11:23:20AM -0500, David Abrahams wrote:
> >> Christoph Ludwig <cludwig at cdc.informatik.tu-darmstadt.de> writes:
> >> 
> >> > On Tue, Nov 01, 2005 at 10:23:14AM -0500, David Abrahams wrote:
> >> >> Christoph Ludwig <cludwig at cdc.informatik.tu-darmstadt.de> writes:
> >> >> > I finally had the spare time to look into this problem again and submitted
> >> >> > patch #1324762. The proposed patch implements the following:
> >> >> 
> >> >> <snip awesome patch>
> >> >> 
> >> >> Christoph, thanks so much for doing this! 
> >> >
> >> > You are welcome. It was pretty straight forward once I realized how the C++
> >> > compiler is communicated from configure to distutils. But the patch has not
> >> > been applied to Python's CVS yet; I don't know how the Python developers think
> >> > about it. 
> >> 
> >> Hmm, maybe I should post encouraging someone to look at this.
> >
> > FYI, the patch got in.
> 
> Great!
> 
> IIRC, this had some practical consequences W.R.T. Boost.Python and/or
> debug builds of Python, but I can't quite remember what the upshot
> was.  Can you?

I do not see how the patch could affect Boost.Python (in the sense that you
have to change anything in Boost.Python's code or its Jamfile). I only
modified the configuration and build scripts of Python in such a way that
Python's main() is compiled and linked with a C++ compiler only if the user
explicitly asks for it. Therefore, the dependency of the Python executable on
the C++ runtime library is removed. On many platforms (notably, Linux/ELF) you
can load C++ extension modules anyway, but you are no longer forced to build
your extension module with a particular C++ compiler version. (On platforms
that support C++ extension modules only if main() was compiled and linked with
a C++ compiler you need to build Python with the --with-cxx-main configure
option. Of course, then you are no longer free to choose which compiler to
build your C++ extensions with. I.e., then you have exactly the situation that
used to be forced on all users.) 

The same holds for a Python debug build - I did not mess with the compiler
option settings at all. If you want your Python executable built with debug
info, you do it as before.

Regards

Christoph

-- 
FH Worms - University of Applied Sciences
Fachbereich Informatik / Telekommunikation
Erenburgerstr. 19, 67549 Worms, Germany


From stummer at gmail.com  Sat Apr 15 20:13:36 2006
From: stummer at gmail.com (david stummer)
Date: Sat, 15 Apr 2006 19:13:36 +0100
Subject: [C++-sig] boost python references
Message-ID: <4d0bf7320604151113w3e967cd5p9e3873f94ad4e18@mail.gmail.com>

hi guys,

i am trying to use references with boost python.

in python you can do:

class person:
    name='david'

a=person()
b=a
print a.name
a.name='john'
print b.name

because b is a reference, it will always change as a changes. hence
b.nameis also john.

i want this functionality between boost and python.

I have the C++ class:

class_<Window>("Window")
    .def_readwrite("seat", &Window::seat)
 ;

where seat is an object of class Seat. Seat contains an integer attribute
named position. It's default value is 45.

My python code is

s=Seat()
w=Window()
print s.position # prints 45
w.seat.position=s.position
s.position=66
print w.seat.position # prints 45
print s.position # prints 66

this shows that it isn't passed by reference as print w.seat.position needs
to be 66

How can i used references?

thanks
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060415/ca16d606/attachment.htm>

From stummer at gmail.com  Sun Apr 16 12:44:46 2006
From: stummer at gmail.com (david stummer)
Date: Sun, 16 Apr 2006 11:44:46 +0100
Subject: [C++-sig] boost python references
In-Reply-To: <4d0bf7320604151113w3e967cd5p9e3873f94ad4e18@mail.gmail.com>
References: <4d0bf7320604151113w3e967cd5p9e3873f94ad4e18@mail.gmail.com>
Message-ID: <4d0bf7320604160344y573655f1m496ebcdf925827d9@mail.gmail.com>

s
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060416/df367668/attachment.htm>

From stummer at gmail.com  Sun Apr 16 12:51:29 2006
From: stummer at gmail.com (david stummer)
Date: Sun, 16 Apr 2006 11:51:29 +0100
Subject: [C++-sig] boost python references
In-Reply-To: <4d0bf7320604151113w3e967cd5p9e3873f94ad4e18@mail.gmail.com>
References: <4d0bf7320604151113w3e967cd5p9e3873f94ad4e18@mail.gmail.com>
Message-ID: <4d0bf7320604160351t5a12745evb33cae48c376397f@mail.gmail.com>

Let me simplify my question. In python you can do something like

class Seat:
    position="a"

class Window:
    seat=None

s=Seat()
s2=Seat()
w=Window()
s2 = s
w.seat=s

print s.position
print s2.position
print w.seat.position

s.position="b"

print s.position
print s2.position
print w.seat.position

and the output would be :
a
a
a
b
b
b

indicating that s2 and w.seat reference s. This is what i want. However,
when i try this with c++ classes it doesn't work, i.e.

class Seat
{
public:
    std::string position;
};

Seat::Seat()
{
    position="david";
}

class Window
{
public:
    Seat seat;
};

class_<Seat>("Seat")
    .def_readwrite("position", &Seat::position)
;

class_<Window>("Window")
    .def_readwrite("seat", &Window::seat)
;

when using similar python code above with these classes instead of the
python-defined ones, i get the following output:
a
a
a
b
b
a

The Seat object in the Window object isn't reference: it is a new object.
Why is this?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060416/008724f7/attachment.htm>

From roman.yakovenko at gmail.com  Sun Apr 16 12:53:36 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Sun, 16 Apr 2006 13:53:36 +0300
Subject: [C++-sig] boost python references
In-Reply-To: <4d0bf7320604151113w3e967cd5p9e3873f94ad4e18@mail.gmail.com>
References: <4d0bf7320604151113w3e967cd5p9e3873f94ad4e18@mail.gmail.com>
Message-ID: <7465b6170604160353mf87237bm9389fae547ad19b9@mail.gmail.com>

On 4/15/06, david stummer <stummer at gmail.com> wrote:
> hi guys,
>
> i am trying to use references with boost python.
>
> in python you can do:
>
> class person:
>     name='david'
>
> a=person()
> b=a
> print a.name
> a.name='john'
>  print b.name
>
> because b is a reference, it will always change as a changes. hence b.name
> is also john.
>
> i want this functionality between boost and python.
>
> I have the C++ class:
>
> class_<Window>("Window")
>     .def_readwrite("seat", &Window::seat)
>  ;
>
> where seat is an object of class Seat. Seat contains an integer attribute
> named position. It's default value is 45.
>
> How can i used references?

There is one important difference between Python code and C++.
In Python you create value that belongs to class, while in C++ you
create value that
belongs to class instance. I think you should use add_static_property function.
Also, I think you will need to use boost::python::make_function in order to
construct getter/setter with desired call policies.

http://www.boost.org/libs/python/doc/v2/class.html
http://www.boost.org/libs/python/doc/v2/make_function.html#make_function-spec

> thanks

--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From stummer at gmail.com  Sun Apr 16 13:07:55 2006
From: stummer at gmail.com (david stummer)
Date: Sun, 16 Apr 2006 12:07:55 +0100
Subject: [C++-sig] boost python references
In-Reply-To: <7465b6170604160353mf87237bm9389fae547ad19b9@mail.gmail.com>
References: <4d0bf7320604151113w3e967cd5p9e3873f94ad4e18@mail.gmail.com>
	<7465b6170604160353mf87237bm9389fae547ad19b9@mail.gmail.com>
Message-ID: <4d0bf7320604160407oa2361dfm86438c20cd17c3da@mail.gmail.com>

ok, i added a member function to Window which takes a Seat reference:

    void Window::AddSeat(Seat &s)
    {
        seat=s;
    }

    class_<Window>("Window")
        .def("FindWindow", make_function(&Window::AddSeat))
        .def_readwrite("seat", &Window::seat)
    ;

using make_function like you said, although this doesn't seem to have any
effect.

I thank you for your patience - i am a newbie
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060416/b7cdef39/attachment.htm>

From roman.yakovenko at gmail.com  Sun Apr 16 13:12:52 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Sun, 16 Apr 2006 14:12:52 +0300
Subject: [C++-sig] boost python references
In-Reply-To: <4d0bf7320604160407oa2361dfm86438c20cd17c3da@mail.gmail.com>
References: <4d0bf7320604151113w3e967cd5p9e3873f94ad4e18@mail.gmail.com>
	<7465b6170604160353mf87237bm9389fae547ad19b9@mail.gmail.com>
	<4d0bf7320604160407oa2361dfm86438c20cd17c3da@mail.gmail.com>
Message-ID: <7465b6170604160412q500d183dm82a20ed75f86de82@mail.gmail.com>

On 4/16/06, david stummer <stummer at gmail.com> wrote:
> ok, i added a member function to Window which takes a Seat reference:
>
>     void Window::AddSeat(Seat &s)
>     {
>         seat=s;
>     }
>
>     class_<Window>("Window")
>         .def("FindWindow", make_function(&Window::AddSeat))
>
>         .def_readwrite("seat", &Window::seat)
>     ;
>
> using make_function like you said, although this doesn't seem to have any
> effect.
>
> I thank you for your patience - i am a newbie

:-). You missed important note:

There is one important difference between Python code and C++.
In Python you create value that belongs to class, while in C++ you
create value that belongs to class instance. I think you should use
add_static_property
function.

make_function you should use in order to setup call policies

--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From stummer at gmail.com  Sun Apr 16 13:30:43 2006
From: stummer at gmail.com (david stummer)
Date: Sun, 16 Apr 2006 12:30:43 +0100
Subject: [C++-sig] boost python references
In-Reply-To: <7465b6170604160412q500d183dm82a20ed75f86de82@mail.gmail.com>
References: <4d0bf7320604151113w3e967cd5p9e3873f94ad4e18@mail.gmail.com>
	<7465b6170604160353mf87237bm9389fae547ad19b9@mail.gmail.com>
	<4d0bf7320604160407oa2361dfm86438c20cd17c3da@mail.gmail.com>
	<7465b6170604160412q500d183dm82a20ed75f86de82@mail.gmail.com>
Message-ID: <4d0bf7320604160430k4153ca9buc0b92055c1fb9b30@mail.gmail.com>

the trouble is i don't want it to be a static member..
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060416/9a1f0144/attachment.htm>

From stummer at gmail.com  Sun Apr 16 13:36:26 2006
From: stummer at gmail.com (david stummer)
Date: Sun, 16 Apr 2006 12:36:26 +0100
Subject: [C++-sig] boost python references
In-Reply-To: <4d0bf7320604160430k4153ca9buc0b92055c1fb9b30@mail.gmail.com>
References: <4d0bf7320604151113w3e967cd5p9e3873f94ad4e18@mail.gmail.com>
	<7465b6170604160353mf87237bm9389fae547ad19b9@mail.gmail.com>
	<4d0bf7320604160407oa2361dfm86438c20cd17c3da@mail.gmail.com>
	<7465b6170604160412q500d183dm82a20ed75f86de82@mail.gmail.com>
	<4d0bf7320604160430k4153ca9buc0b92055c1fb9b30@mail.gmail.com>
Message-ID: <4d0bf7320604160436r7c92b6f0m7726e8ea4af4e7a3@mail.gmail.com>

i was wondering if you could provide some simple code for me to follow - i'd
really appreciate this as i am lost :(
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060416/70a93587/attachment.htm>

From roman.yakovenko at gmail.com  Sun Apr 16 13:38:11 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Sun, 16 Apr 2006 14:38:11 +0300
Subject: [C++-sig] boost python references
In-Reply-To: <4d0bf7320604160430k4153ca9buc0b92055c1fb9b30@mail.gmail.com>
References: <4d0bf7320604151113w3e967cd5p9e3873f94ad4e18@mail.gmail.com>
	<7465b6170604160353mf87237bm9389fae547ad19b9@mail.gmail.com>
	<4d0bf7320604160407oa2361dfm86438c20cd17c3da@mail.gmail.com>
	<7465b6170604160412q500d183dm82a20ed75f86de82@mail.gmail.com>
	<4d0bf7320604160430k4153ca9buc0b92055c1fb9b30@mail.gmail.com>
Message-ID: <7465b6170604160438y5ec1e203m7b6b4dcfca3b2fac@mail.gmail.com>

On 4/16/06, david stummer <stummer at gmail.com> wrote:
> the trouble is i don't want it to be a static member..

Somewhere should be defined an unique instance of the class, you want to share
across all instances. How you implement this, it is up to you.

It is also possible that I don't understand you. So may be other
people can help you?

> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>
>
>


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From stummer at gmail.com  Sun Apr 16 13:46:21 2006
From: stummer at gmail.com (david stummer)
Date: Sun, 16 Apr 2006 12:46:21 +0100
Subject: [C++-sig] boost python references
In-Reply-To: <7465b6170604160438y5ec1e203m7b6b4dcfca3b2fac@mail.gmail.com>
References: <4d0bf7320604151113w3e967cd5p9e3873f94ad4e18@mail.gmail.com>
	<7465b6170604160353mf87237bm9389fae547ad19b9@mail.gmail.com>
	<4d0bf7320604160407oa2361dfm86438c20cd17c3da@mail.gmail.com>
	<7465b6170604160412q500d183dm82a20ed75f86de82@mail.gmail.com>
	<4d0bf7320604160430k4153ca9buc0b92055c1fb9b30@mail.gmail.com>
	<7465b6170604160438y5ec1e203m7b6b4dcfca3b2fac@mail.gmail.com>
Message-ID: <4d0bf7320604160446y5d803a2ftfa0c58a58dec049d@mail.gmail.com>

>Somewhere should be defined an unique instance of the class, you want to
share
>across all instances. How you implement this, it is up to you.

i don't know how to implement this - which is why i posted here in the first
place..

as you can see i have tried references, but the object for some reason is
being coied.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060416/91fe3c34/attachment.htm>

From roman.yakovenko at gmail.com  Sun Apr 16 18:31:12 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Sun, 16 Apr 2006 19:31:12 +0300
Subject: [C++-sig] boost python references
In-Reply-To: <4d0bf7320604160446y5d803a2ftfa0c58a58dec049d@mail.gmail.com>
References: <4d0bf7320604151113w3e967cd5p9e3873f94ad4e18@mail.gmail.com>
	<7465b6170604160353mf87237bm9389fae547ad19b9@mail.gmail.com>
	<4d0bf7320604160407oa2361dfm86438c20cd17c3da@mail.gmail.com>
	<7465b6170604160412q500d183dm82a20ed75f86de82@mail.gmail.com>
	<4d0bf7320604160430k4153ca9buc0b92055c1fb9b30@mail.gmail.com>
	<7465b6170604160438y5ec1e203m7b6b4dcfca3b2fac@mail.gmail.com>
	<4d0bf7320604160446y5d803a2ftfa0c58a58dec049d@mail.gmail.com>
Message-ID: <7465b6170604160931s262c5e8clebde71057eb74ac@mail.gmail.com>

On 4/16/06, david stummer <stummer at gmail.com> wrote:
> >Somewhere should be defined an unique instance of the class, you want to
> share
> >across all instances. How you implement this, it is up to you.

Somewhere should be some kind of static:

1. It could be module attribute, then all your get/set functions
should work on it
2. Regular C++ static defined somewhere

Try to implement some solution in C++, after that I will try to help
you with boost.python

--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From stummer at gmail.com  Sun Apr 16 19:17:55 2006
From: stummer at gmail.com (david stummer)
Date: Sun, 16 Apr 2006 18:17:55 +0100
Subject: [C++-sig] call python from c++ (not in dll)
Message-ID: <4d0bf7320604161017v5a273912me9a04468e3d21e4@mail.gmail.com>

i have managed to put together dlls for python to call, but is it possible
to call a python module from an exe file?

if so where is this info as i cannot see it on boost.org?

cheers
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060416/46e84fdb/attachment.htm>

From roman.yakovenko at gmail.com  Sun Apr 16 19:28:46 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Sun, 16 Apr 2006 20:28:46 +0300
Subject: [C++-sig] call python from c++ (not in dll)
In-Reply-To: <4d0bf7320604161017v5a273912me9a04468e3d21e4@mail.gmail.com>
References: <4d0bf7320604161017v5a273912me9a04468e3d21e4@mail.gmail.com>
Message-ID: <7465b6170604161028j19a8f648j6e5a98d80b4c4e6@mail.gmail.com>

On 4/16/06, david stummer <stummer at gmail.com> wrote:
> i have managed to put together dlls for python to call, but is it possible
> to call a python module from an exe file?

Why not, may be I am missing something, but it should work out of the box.
Just link your exe with relevant libs and enjoy.

> if so where is this info as i cannot see it on boost.org?
>
> cheers


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From stummer at gmail.com  Sun Apr 16 19:55:24 2006
From: stummer at gmail.com (david stummer)
Date: Sun, 16 Apr 2006 18:55:24 +0100
Subject: [C++-sig] call python from c++ (not in dll)
In-Reply-To: <7465b6170604161028j19a8f648j6e5a98d80b4c4e6@mail.gmail.com>
References: <4d0bf7320604161017v5a273912me9a04468e3d21e4@mail.gmail.com>
	<7465b6170604161028j19a8f648j6e5a98d80b4c4e6@mail.gmail.com>
Message-ID: <4d0bf7320604161055h33db71b5o509667272b3d7bc6@mail.gmail.com>

i don't mean to be rude but your answer doesn't help me in the slightest

i was loking for maybe a tutorial which i haven't found yet

boost has absolutely nothing on calling python from an exe which i can see

what is the point in using boost if relative newcommers like me cannot get
anywhere
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060416/e009085f/attachment.htm>

From roman.yakovenko at gmail.com  Sun Apr 16 20:12:14 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Sun, 16 Apr 2006 21:12:14 +0300
Subject: [C++-sig] call python from c++ (not in dll)
In-Reply-To: <4d0bf7320604161055h33db71b5o509667272b3d7bc6@mail.gmail.com>
References: <4d0bf7320604161017v5a273912me9a04468e3d21e4@mail.gmail.com>
	<7465b6170604161028j19a8f648j6e5a98d80b4c4e6@mail.gmail.com>
	<4d0bf7320604161055h33db71b5o509667272b3d7bc6@mail.gmail.com>
Message-ID: <7465b6170604161112mfa68ea2x4442d737bec94ff3@mail.gmail.com>

On 4/16/06, david stummer <stummer at gmail.com> wrote:
> i don't mean to be rude but your answer doesn't help me in the slightest

Sorry, I was lazy:
http://boost.org/libs/python/doc/tutorial/doc/html/python/embedding.html

> i was loking for maybe a tutorial which i haven't found yet

Try google: boost python embedding exe tutorial. 3'rd link will point
you to the url I provided

> boost has absolutely nothing on calling python from an exe which i can see
>
> what is the point in using boost if relative newcommers like me cannot get
> anywhere

boost is relatively big project with some culture. It takes time to
learn it and
understand it's design. Almost all libraries have documentation,
sometimes it even
good. If you not satisfy with something you can:

1. ask( you get the answers, right? )
2. search for answers( you are not the first newcomer! )
3. try to write some document and post it on boost development mailing list.
    In that document you can specify what is wrong, on your opinion, and
    how it should be.
4. you can make it better( hint: contribute, contribute, contribute )
5. you can complain and do nothing

Enter your choice: ...

:-)

--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From stummer at gmail.com  Sun Apr 16 20:44:14 2006
From: stummer at gmail.com (david stummer)
Date: Sun, 16 Apr 2006 19:44:14 +0100
Subject: [C++-sig] call python from c++ (not in dll)
In-Reply-To: <7465b6170604161112mfa68ea2x4442d737bec94ff3@mail.gmail.com>
References: <4d0bf7320604161017v5a273912me9a04468e3d21e4@mail.gmail.com>
	<7465b6170604161028j19a8f648j6e5a98d80b4c4e6@mail.gmail.com>
	<4d0bf7320604161055h33db71b5o509667272b3d7bc6@mail.gmail.com>
	<7465b6170604161112mfa68ea2x4442d737bec94ff3@mail.gmail.com>
Message-ID: <4d0bf7320604161144l77c7ba93qe2fb0eee1d6c241d@mail.gmail.com>

ok thanks

i will take a look

cheers
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060416/41b81f6a/attachment.htm>

From stummer at gmail.com  Sun Apr 16 20:48:01 2006
From: stummer at gmail.com (david stummer)
Date: Sun, 16 Apr 2006 19:48:01 +0100
Subject: [C++-sig] c++ funciton slowing
Message-ID: <4d0bf7320604161148ve0f87afq9528941431b17d39@mail.gmail.com>

one of my c++ class functions which performs fine 'natively' slows right
down (i.e. takes nearly 1 minute) when i call it from python.

void Window::LoadImages()
{
    int ids[3] = {IDB_BITMAP1,IDB_BITMAP2,IDB_BITMAP3};

    for(int i=0;i<3;i++)
    {
        HANDLE hBm = ::LoadImage(::GetModuleHandle(NULL),
MAKEINTRESOURCE(ids[i]),
                                 IMAGE_BITMAP, 0, 0, LR_DEFAULTCOLOR);
        BITMAP bm;
        ::GetObject(hBm, sizeof(BITMAP), &bm);
        int size = bm.bmHeight * bm.bmWidth * bm.bmBitsPixel / 8;
        pixels[i] = new BYTE[size];
        ::GetBitmapBits((HBITMAP)hBm, size, pixels[i]);
    }
}

What it does is load resources from within visual studio. i have three
individual resources. it stalls on this line:

pixels[i] = new BYTE[size];

where pixels is declared as

BYTE* pixels[4]

any ideas?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060416/636146ba/attachment.htm>

From dave at boost-consulting.com  Mon Apr 17 00:09:23 2006
From: dave at boost-consulting.com (David Abrahams)
Date: Sun, 16 Apr 2006 18:09:23 -0400
Subject: [C++-sig] wrapping an abstract class with out parameters using
	boost.python
References: <609a651c0604110907y13d1ecd5qe2135642956a8866@mail.gmail.com>
Message-ID: <m28xq5jhpo.fsf@boonen.local>

"Guillaume Desticourt" <guillaume.desticourt at gmail.com> writes:

> class DisplayDataWrap :
>   public DisplayData,
>   public wrapper<DisplayData>;
>
> to encapsulate the DisplayData::accept, i ve defined:
>
>   void DisplayDataWrap::accept(DisplayDataVisitor &visitor)
>   {
>     this->get_override("accept")(visitor);
>   }
>
> and in the module definition i do:
>  class_<DisplayDataWrap, boost::noncopyable>("PyDisplayData", no_init)
>     .def("accept", pure_virtual(&DisplayData::accept))
>     ;
>
> the problem is that i have no idea how to encapsulate:
> the DisplaData::getZone method...
> i tried
>   Zone getZone(void) const
>   {
>     Zone zone;
>     this->get_override("getZone")(zone);
>     return zone;
>   }
>
>  Zone (DisplayDataWrap::*getZone1)(void) const = &DisplayDataWrap::getZone;
>
>   class_<DisplayDataWrap, boost::noncopyable>("PyDisplayData", no_init)
>     .def("getZone", getZone1) ;
>
> and then i get the error messge
> RuntimeError: extension class wrapper for base class
> pimp::gfx2d::DisplayData has not been created yet
> and i am not even sure i ll be able to override the getZone in python...

Please post a minimal, reproducible testcase that shows your problem.
This should ideally be one C++ file (the extension module) and one
Python file that I can invoke to exercise it and demonstrate the problem.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com



From nyamatongwe at gmail.com  Mon Apr 17 00:54:03 2006
From: nyamatongwe at gmail.com (Neil Hodgson)
Date: Mon, 17 Apr 2006 08:54:03 +1000
Subject: [C++-sig] c++ funciton slowing
In-Reply-To: <4d0bf7320604161148ve0f87afq9528941431b17d39@mail.gmail.com>
References: <4d0bf7320604161148ve0f87afq9528941431b17d39@mail.gmail.com>
Message-ID: <50862ebd0604161554s71e902dby7efd275ee7bceba6@mail.gmail.com>

david stummer:

> one of my c++ class functions which performs fine 'natively' slows right
> down (i.e. takes nearly 1 minute) when i call it from python.
> ...
> any ideas?

   Check the return codes of all the functions. You are most likely
not loading resources from the file you think you are.

   Neil


From roman.yakovenko at gmail.com  Mon Apr 17 09:40:49 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Mon, 17 Apr 2006 10:40:49 +0300
Subject: [C++-sig] ANN: pyplusplus - v0.7.2
Message-ID: <7465b6170604170040j24b2d1dbue0cc07933f6243d5@mail.gmail.com>

Hi. The next version of pyplusplus is available.
Download page: http://www.language-binding.net/pyplusplus/download.html

What is it?
pyplusplus  is an object-oriented framework for creating a code
generator for  boost.python  library.

Project home page: http://www.language-binding.net/pyplusplus/pyplusplus.html

What is new?
    New user friendly API has been created. You can find tutorials here:
    http://www.language-binding.net/pyplusplus/tutorials/tutorials.html

    Logging functionality has been added to almost all operations

    Documentation has been improved a lot.

    Bug fixes

Full list of changes you can find here:
http://www.language-binding.net/pyplusplus/history/history.html

Many thanks to Matthias Baas  and Allen Bierbaum. They fixed bugs, wrote
documentation, designed new user friendly API.

Enjoy

--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From stummer at gmail.com  Mon Apr 17 11:06:01 2006
From: stummer at gmail.com (david stummer)
Date: Mon, 17 Apr 2006 10:06:01 +0100
Subject: [C++-sig] c++ funciton slowing
In-Reply-To: <50862ebd0604161554s71e902dby7efd275ee7bceba6@mail.gmail.com>
References: <4d0bf7320604161148ve0f87afq9528941431b17d39@mail.gmail.com>
	<50862ebd0604161554s71e902dby7efd275ee7bceba6@mail.gmail.com>
Message-ID: <4d0bf7320604170206n17b4ec8fxe2bf89e027a8d208@mail.gmail.com>

ok my fault :(

i forgot to switch to 'multi threaded dll'

cheers
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060417/13bdd35d/attachment.htm>

From stummer at gmail.com  Mon Apr 17 13:38:24 2006
From: stummer at gmail.com (david stummer)
Date: Mon, 17 Apr 2006 12:38:24 +0100
Subject: [C++-sig] sharing objects
Message-ID: <4d0bf7320604170438j267af19g3b2057850f16e2db@mail.gmail.com>

ok guys i was wondering if anyone had some examples of sharing class objects
between c++ and python.

i have made a few posts recently about sharing and referencing but nobody
seems to have an answer so far.

i understand it's quite difficult because of the differences the two
languages have in referencing objects, storing them etc,

cheers
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060417/5211d227/attachment.htm>

From stummer at gmail.com  Mon Apr 17 16:25:24 2006
From: stummer at gmail.com (david stummer)
Date: Mon, 17 Apr 2006 15:25:24 +0100
Subject: [C++-sig] vector in class
Message-ID: <4d0bf7320604170725w748fce67id6b2ebea7ed08e70@mail.gmail.com>

i managed to get a vector working on it's own using

    typedef std::vector<int> IntVec;

    class_<IntVec>("bankrolls")
        .def(vector_indexing_suite<IntVec>())
    ;

but now i would like to make a vector a member of a class.

what steps would i take to make the vector a member of a c++ class and have
it callable from python?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060417/94d28170/attachment.htm>

From mharidev at qualcomm.com  Mon Apr 17 22:18:03 2006
From: mharidev at qualcomm.com (Haridev, Meghana)
Date: Mon, 17 Apr 2006 13:18:03 -0700
Subject: [C++-sig] Generation of wrapper for arguments of type T*&
Message-ID: <5383A81D5EB8F14FBDB487D9B5FBEA88C0E8C0@NAEX11.na.qualcomm.com>

Hi Folks,

 

I want to wrap the following piece of code using pyplusplus:

 

=================================================

// Simulating an out using foo (T*& ptr)

Class_Z

{

    public:

 

Class_X  foo (Class_Y*& ptr)

{

            Class_Y *newptr = new Class_Y();

            ...

            ...

            ptr = newptr;

 

            return Class_X ();

      }

} 

 

=================================================

 

Manually, I am wrapping it like this in Boost:

 

=================================================

 

boost::python::tuple foo(Class_Z& self )

{

    Class_Y *ptr;

    Class_X objx = self.foo (ptr);

    return boost::python::make_tuple(objx, ptr);

}

 

class_<Class_Z>("Class_Z")

{

.def( "foo", &::foo)

}

  

=================================================

 

I'm a newbie to boost/python/pyplusplus! Any suggestions much
appreciated!

 

Thanks,

-Meghana.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060417/9cf6e617/attachment.htm>

From roman.yakovenko at gmail.com  Mon Apr 17 22:35:07 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Mon, 17 Apr 2006 23:35:07 +0300
Subject: [C++-sig] Generation of wrapper for arguments of type T*&
In-Reply-To: <5383A81D5EB8F14FBDB487D9B5FBEA88C0E8C0@NAEX11.na.qualcomm.com>
References: <5383A81D5EB8F14FBDB487D9B5FBEA88C0E8C0@NAEX11.na.qualcomm.com>
Message-ID: <7465b6170604171335x5f7a581exf58a0d9ff6c5c721@mail.gmail.com>

On 4/17/06, Haridev, Meghana <mharidev at qualcomm.com> wrote:
> Hi Folks,
> I want to wrap the following piece of code using pyplusplus:
> I'm a newbie to boost/python/pyplusplus! Any suggestions much appreciated!

I plan to add this feature to the next release of pyplusplus.

Right now you have few ways to go:

1. If you have only few such functions, then you can wrap them manually.
    after this you can add

    .def( "for", &foo )

   code to the class, using add_code method

   mb = module_builder_t( ... )

   Class_Z = mb.class_( name="Class_Z" )
   foo = Class_Z.member_function( name="foo", arg_types=[None] )


> Thanks,
>
> -Meghana.


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From roman.yakovenko at gmail.com  Mon Apr 17 22:45:53 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Mon, 17 Apr 2006 23:45:53 +0300
Subject: [C++-sig] Generation of wrapper for arguments of type T*&
In-Reply-To: <5383A81D5EB8F14FBDB487D9B5FBEA88C0E8C0@NAEX11.na.qualcomm.com>
References: <5383A81D5EB8F14FBDB487D9B5FBEA88C0E8C0@NAEX11.na.qualcomm.com>
Message-ID: <7465b6170604171345p392d9ffav18c443705c8004e0@mail.gmail.com>

On 4/17/06, Haridev, Meghana <mharidev at qualcomm.com> wrote:
> Hi Folks,
> I want to wrap the following piece of code using pyplusplus:
> // Simulating an out using foo (T*& ptr)
> Class_Z
> {
>     public:
> Class_X  foo (Class_Y*& ptr)
> {
>
>             Class_Y *newptr = new Class_Y();
>             ?
>             ptr = newptr;
>             return Class_X ();
>       }
> }
> Manually, I am wrapping it like this in Boost:
> boost::python::tuple foo(Class_Z& self )
>
> {
>
>     Class_Y *ptr;
>
>     Class_X objx = self.foo (ptr);
>
>     return boost::python::make_tuple(objx, ptr);
>
> }
>
>
>
> class_<Class_Z>("Class_Z")
>
> {
>
> .def( "foo", &::foo)
>
> }
>
> I'm a newbie to boost/python/pyplusplus! Any suggestions much appreciated!


I plan to add this feature to the next release of pyplusplus.

Right now you have few ways to go:

1. If you have only few such functions, then you can wrap them manually.
    after this you can add

    .def( "for", &foo )

   code to the class, using add_code method
   #I assume that you already has module_builder_t class instance
   mb = module_builder_t( ... )
   #Get declaration of Class_Z class
   Class_Z = mb.class_( name="Class_Z" )
   #get declaration of function "foo", If "foo" function is overloaded, then
   #you can specify additional criteria: argument types
   foo = Class_Z.member_function( name="foo" )
   foo.exclude()
   Class_Z.add_code( 'def( "foo", &foo )' )

This is a simple way to do this.

Lets start with this one, if it will not help you, we will proceed to
an other one.


> Thanks,
>
> -Meghana.
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>
>
>


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From mharidev at qualcomm.com  Tue Apr 18 05:02:46 2006
From: mharidev at qualcomm.com (Haridev, Meghana)
Date: Mon, 17 Apr 2006 20:02:46 -0700
Subject: [C++-sig] Generation of wrapper for arguments of type T*&
Message-ID: <5383A81D5EB8F14FBDB487D9B5FBEA88C0E9A1@NAEX11.na.qualcomm.com>

Thanks Roman. 

 

> I plan to add this feature to the next release of pyplusplus.

> Right now you have few ways to go:

 

> 1. If you have only few such functions, then you can wrap them
manually.

 

Can I make pyplusplus just insert the "manual wrappers" code in the
generated file? Is there a simple example which illustrates a way in
which we can just write/insert pieces of code to the generated file? 

Or do I have to create a different file where I put the code for the
"manual wrappers" and then make pylusplus include that file in the
generated file?

 

>    after this you can add

>    .def( "for", &foo )

> 

>   code to the class, using add_code method

>   #I assume that you already has module_builder_t class instance

>   mb = module_builder_t( ... )

>   #Get declaration of Class_Z class

>   Class_Z = mb.class_( name="Class_Z" )

>  #get declaration of function "foo", If "foo" function is overloaded,
then

> 

>  #you can specify additional criteria: argument types

>   foo = Class_Z.member_function( name="foo" )

>   foo.exclude()

>   Class_Z.add_code( 'def( "foo", &foo )' )

 

This works for me....but how can I make pyplusplus insert this manual
code to the generated code? :-)

 

boost::python::tuple foo(Class_Z& self )

{

    Class_Y *ptr;

    Class_X objx = self.foo (ptr);

    return boost::python::make_tuple(objx, ptr);

}

 

> This is a simple way to do this.

> Lets start with this one, if it will not help you, we will proceed to

> other one.

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060417/f9a73ea5/attachment.htm>

From rodlist at hotmail.com  Tue Apr 18 05:29:55 2006
From: rodlist at hotmail.com (Rod Cloutier)
Date: Mon, 17 Apr 2006 23:29:55 -0400
Subject: [C++-sig] Exposing Abstract Classes and factories
Message-ID: <BAY108-DAV183F6AB2CD954B2F6181F2B4C40@phx.gbl>

Hi,

I am trying to use pyplusplus to expose python binding to a library. The library currently uses a lot of abstract classes and generate concrete object through factories. I am currently unable to use, in python, any of the object created through the factories. Is it possible to expose abstract classes in python and call concrete object, not exposed in python, through the interface ?

Here is a concise example of the problem:

Many thanks for your help


Rod


==================== Hello.h  =============================

#include <memory>

class Abstract
{
public:
 virtual void run() const = 0;
};

std::auto_ptr<Abstract> create();


==================== Hello.cpp  =============================

#include <iostream>
#include "hello.h"

class Concrete : public Abstract
{
 virtual void run( ) const
 {
  std::cout << "Concrete ran!";
 }
};

std::auto_ptr<Abstract> create()
{
 return std::auto_ptr<Abstract>( new Concrete() );
}


==================== PyPlusPlus Code  =============================

from pyplusplus import module_builder
from pyplusplus import code_creators


mb = module_builder.module_builder_t( [r"hello.h"]
                                      , gccxml_path=r"C:/GCC_XML/bin" 
                                      , working_directory=r"."
                                      , include_paths=['.']
                                      , define_symbols=[] )

mb.build_code_creator( module_name='pyplusplusTest' )
mb.write_module( './bindings.cpp' )


==================== Python Test Code =============================

import pyplusplusTest

instance = pyplusplusTest.create()
instance.run()


--------------------- Error obtained -----------------------
Traceback (most recent call last):
  File "D:\dev\cplusplus\pyplusplusTest\Debug\test.py", line 5, in ?
    instance.run()
Boost.Python.ArgumentError: Python argument types in
    Abstract.run(Abstract)
did not match C++ signature:
    run(struct Abstract_wrapper {lvalue})
    run(struct Abstract_wrapper {lvalue})
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060417/2eb810cb/attachment.htm>

From roman.yakovenko at gmail.com  Tue Apr 18 07:11:38 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 18 Apr 2006 08:11:38 +0300
Subject: [C++-sig] Generation of wrapper for arguments of type T*&
In-Reply-To: <5383A81D5EB8F14FBDB487D9B5FBEA88C0E9A1@NAEX11.na.qualcomm.com>
References: <5383A81D5EB8F14FBDB487D9B5FBEA88C0E9A1@NAEX11.na.qualcomm.com>
Message-ID: <7465b6170604172211y6684daf4g737b7061b4de57d0@mail.gmail.com>

On 4/18/06, Haridev, Meghana <mharidev at qualcomm.com> wrote:
> Thanks Roman.
>
> Can I make pyplusplus just insert the "manual wrappers" code in the
> generated file? Is there a simple example which illustrates a way in which
> we can just write/insert pieces of code to the generated file?

It is possible, but it still does not have good user interface.
Anyway, please take a look on pyplusplus/file_writers package.
It contains single_file.py and multiple_files.py modules. You can derive from
them and implement custom algorithm for dumping code to files.

> Or do I have to create a different file where I put the code for the "manual
> wrappers" and then make pylusplus include that file in the generated file?

You can do that.

mb = module_builder_t(...)
mb.build_code_creator( module_name="hw" )

my_include =  code_creators.include_t( header=header_file )
mb.code_creator.adopt_include( my_include )

The price you pay is compilation time. This file will be included in
every generated cpp
file.

>
> >    after this you can add
>
> >    .def( "for", &foo )
> >   code to the class, using add_code method
> >   #I assume that you already has module_builder_t class instance
> >   mb = module_builder_t( ... )
> >   #Get declaration of Class_Z class
> >   Class_Z = mb.class_( name="Class_Z" )
> >  #get declaration of function "foo", If "foo" function is overloaded, then
> >  #you can specify additional criteria: argument types
> >   foo = Class_Z.member_function( name="foo" )
> >   foo.exclude()
> >   Class_Z.add_code( 'def( "foo", &foo )' )

> This works for me....but how can I make pyplusplus insert this manual code
> to the generated code? :-)

Yesterday, I did not have access to source code, so the answer was incomplete.
Now I do have.

Class_Z.add_wrapper_code( "your code here" ).

That's all.

--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From roman.yakovenko at gmail.com  Tue Apr 18 07:18:30 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 18 Apr 2006 08:18:30 +0300
Subject: [C++-sig] Exposing Abstract Classes and factories
In-Reply-To: <BAY108-DAV183F6AB2CD954B2F6181F2B4C40@phx.gbl>
References: <BAY108-DAV183F6AB2CD954B2F6181F2B4C40@phx.gbl>
Message-ID: <7465b6170604172218q28006b28h1d452b9029acba41@mail.gmail.com>

On 4/18/06, Rod Cloutier <rodlist at hotmail.com> wrote:
>
> Hi,
>
> I am trying to use pyplusplus to expose python binding to a library. The
> library currently uses a lot of abstract classes and generate concrete
> object through factories. I am currently unable to use, in python, any of
> the object created through the factories. Is it possible to expose abstract
> classes in python and call concrete object, not exposed in python, through
> the interface ?
>
> Here is a concise example of the problem:
>
> Many thanks for your help

I am sorry, but my machine crashed yesterday, so I can not give you
the answer right now.
I hope I will be able to check this issue this evening or tomorrow morning.

> Rod

--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From fatfatson at gmail.com  Tue Apr 18 10:46:36 2006
From: fatfatson at gmail.com (=?GB2312?B?0rG+p7XL?=)
Date: Tue, 18 Apr 2006 16:46:36 +0800
Subject: [C++-sig] how can i import a module without executing any code?
Message-ID: <6d415ba30604180146q2601d128t2cdb70ccf07d4424@mail.gmail.com>

when i use PyImport_Import() to import a module writen in Python, the global
statement in which will by executed automatically,
while i don't want so,i'd like the python-programmer to write their initial
code in a "init()" function, and i will call it in C at the proper time
instead,
how can i make it so?
thanks very much,especially for endureing my bad english:)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060418/8e1af5f8/attachment.htm>

From fatfatson at gmail.com  Tue Apr 18 10:49:14 2006
From: fatfatson at gmail.com (=?GB2312?B?0rG+p7XL?=)
Date: Tue, 18 Apr 2006 16:49:14 +0800
Subject: [C++-sig] how can i import a module without executing any code?
In-Reply-To: <6d415ba30604180146q2601d128t2cdb70ccf07d4424@mail.gmail.com>
References: <6d415ba30604180146q2601d128t2cdb70ccf07d4424@mail.gmail.com>
Message-ID: <6d415ba30604180149q6cda621bs16791cb0aa9ef185@mail.gmail.com>

for additon: the exact meaning i say is : i expect a approch to ignore the
global statements in the .py module.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060418/faaf8867/attachment.htm>

From hans_meine at gmx.net  Tue Apr 18 18:22:41 2006
From: hans_meine at gmx.net (Hans Meine)
Date: Tue, 18 Apr 2006 18:22:41 +0200
Subject: [C++-sig] how can i import a module without executing any code?
In-Reply-To: <6d415ba30604180149q6cda621bs16791cb0aa9ef185@mail.gmail.com>
References: <6d415ba30604180146q2601d128t2cdb70ccf07d4424@mail.gmail.com>
	<6d415ba30604180149q6cda621bs16791cb0aa9ef185@mail.gmail.com>
Message-ID: <200604181822.41567.hans_meine@gmx.net>

Hi!

On Tuesday 18 April 2006 10:49, ??? wrote:
> for additon: the exact meaning i say is : i expect a approch to ignore the
> global statements in the .py module.

Then, what should be the result of the import?

def foo()
  pass

is also a "global statement", as is the definition of classes.

AFAICS, there is no such thing as what you're looking for.  You might try 
postponing your import though.

Ciao, /  /
     /--/
    /  / ANS


From roman.yakovenko at gmail.com  Tue Apr 18 22:53:38 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 18 Apr 2006 23:53:38 +0300
Subject: [C++-sig] Exposing Abstract Classes and factories
In-Reply-To: <BAY108-DAV183F6AB2CD954B2F6181F2B4C40@phx.gbl>
References: <BAY108-DAV183F6AB2CD954B2F6181F2B4C40@phx.gbl>
Message-ID: <7465b6170604181353j7782aa77k3c848220d15d2ee1@mail.gmail.com>

On 4/18/06, Rod Cloutier <rodlist at hotmail.com> wrote:
> Hi,
>
> ==================== Hello.h  =============================
>
> #include <memory>
>
> class Abstract
> {
> public:
>  virtual void run() const = 0;
> };
>
> std::auto_ptr<Abstract> create();
>
>
> ==================== Hello.cpp  =============================
>
> #include <iostream>
> #include "hello.h"
>
> class Concrete : public Abstract
> {
>  virtual void run( ) const
>  {
>   std::cout << "Concrete ran!";
>  }
> };
>
> std::auto_ptr<Abstract> create()
> {
>  return std::auto_ptr<Abstract>( new Concrete() );
> }
>
>

You found bug in pyplusplus. Thanks.

Description:
I renamed your classes to "abstract" and "concrete", and put them into
"factory" namespace

namespace bp = boost::python;

struct abstract_wrapper : factory::abstract, bp::wrapper< factory::abstract > {

    abstract_wrapper()
    : factory::abstract()
      , bp::wrapper< factory::abstract >()
    {}

    virtual int run(  ) const {
        bp::override run = this->get_override( "run" );
        return run(  );
    }

};

BOOST_PYTHON_MODULE(factory){
    if( true ){
        typedef bp::class_< abstract_wrapper, boost::noncopyable >
abstract_exposer_t;
        abstract_exposer_t abstract_exposer = abstract_exposer_t( "abstract" );
        bp::scope abstract_scope( abstract_exposer );
        abstract_exposer.def( "run"
                //, bp::pure_virtual( &::abstract_wrapper::run )
                //^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                , bp::pure_virtual( &::factory::abstract::run )
                , bp::default_call_policies() );
        bp::register_ptr_to_python< std::auto_ptr< factory::abstract > >();
    }

    bp::def( "create"
            , &factory::create
            , bp::default_call_policies() );
}

When generating "def" for pure virtual function "run", as argument I
provided address
to "run" function defined in wrapper class. This is just wrong. I need
to provide reference
to function defined in base class.

Fix. I don't have access to CVS right now, so I attach file to the mail.
Please put it in pyplusplus/code_creators directory.

Also, I added this use case to unittests.

--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: calldef.py
Type: text/x-python
Size: 40092 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060418/19218de6/attachment.py>

From dominic.sacre at gmx.de  Wed Apr 19 04:20:30 2006
From: dominic.sacre at gmx.de (Dominic =?iso-8859-1?q?Sacr=E9?=)
Date: Wed, 19 Apr 2006 04:20:30 +0200
Subject: [C++-sig] Wrapping pointers
In-Reply-To: <20060414010138.83299.qmail@web31506.mail.mud.yahoo.com>
References: <20060414010138.83299.qmail@web31506.mail.mud.yahoo.com>
Message-ID: <200604190420.30241.dominic.sacre@gmx.de>

On Friday, 14. April 2006 03:01, Ralf W. Grosse-Kunstleve wrote:
> CAVEAT: Raw pointers are intrinsically difficult to work with, terribly
> unsafe, and therefore not easily compatible with the safe Python layer.
> I avoid them completely (except in core code not exposed via public
> interfaces), i.e. I have no working experience related to your
> question.

Ok... Then what would you do? Does it make any difference whether I'm 
using pointers or references? I'm just wondering because obviously 
there's no such distinction in Python...

> FWIW: You may want to try
>
>   .add_property("p", make_getter(&A::p, xxx));

This doesn't help, I'm getting the same error as before.

All I want is to be able to do something like this in Python:

x = test.A()
x.p.foo()

The closest I could get was

class A {
    ...
    A* get_p() { return p; }
};

.def("get_p", &A::get_p, return_value_policy<reference_existing_object>())

but I'd rather be able to access p as a data member rather than a method. 
Any ideas?

Thanks,
    Dominic


> > > On 4/5/06, Dominic Sacr? <dominic.sacre at gmx.de> wrote:
> > > > Hi,
> > > >
> > > > I'm having trouble exporting C++ pointers to Python. My code
> > > > looks like this:
> > > >
> > > > class A
> > > > {
> > > >     A *p;
> > > >     ...
> > > > };
> > > >
> > > > BOOST_PYTHON_MODULE(test)
> > > > {
> > > >     class_<A>("A", init<>())
> > > >         .def_readonly("p", &A::p)
> > > >         ...
> > > >     ;
> > > > }
> > > >
> > > > Now every time I try to access A::p from Python I get this error:
> > > >
> > > > TypeError: No to_python (by-value) converter found for C++ type:
> > > > A*


From guillaume.desticourt at gmail.com  Wed Apr 19 10:56:33 2006
From: guillaume.desticourt at gmail.com (Guillaume Desticourt)
Date: Wed, 19 Apr 2006 09:56:33 +0100
Subject: [C++-sig] wrapping an abstract class with out parameters using
	boost.python
In-Reply-To: <m28xq5jhpo.fsf@boonen.local>
References: <609a651c0604110907y13d1ecd5qe2135642956a8866@mail.gmail.com>
	<m28xq5jhpo.fsf@boonen.local>
Message-ID: <609a651c0604190156x34745734k9f1d004f31e82619@mail.gmail.com>

On 4/16/06, David Abrahams <dave at boost-consulting.com> wrote:
>
> "Guillaume Desticourt" <guillaume.desticourt at gmail.com> writes:


[...]


> Please post a minimal, reproducible testcase that shows your problem.
> This should ideally be one C++ file (the extension module) and one
> Python file that I can invoke to exercise it and demonstrate the problem.


hello,

i didnt manage to reproduce the problem in a short source sample, but i ve
found  that i ve forgotten a base<...> in one of my class_<...>.
i am not sure of that because i ve tried and modified a lot of things, but
this may have been the cause of the problem.

regards,

--
Guillaume Desticourt
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060419/4e9a4a6c/attachment.htm>

From rodlist at hotmail.com  Wed Apr 19 14:59:31 2006
From: rodlist at hotmail.com (Rod Cloutier)
Date: Wed, 19 Apr 2006 12:59:31 +0000
Subject: [C++-sig] Exposing Abstract Classes and factories
In-Reply-To: <7465b6170604181353j7782aa77k3c848220d15d2ee1@mail.gmail.com>
Message-ID: <BAY108-F25A7D63AA40AB96681C269B4C50@phx.gbl>

Thank you Roman, it's working perfectly now!

Rodrigue Cloutier




From rodlist at hotmail.com  Wed Apr 19 15:04:56 2006
From: rodlist at hotmail.com (Rod Cloutier)
Date: Wed, 19 Apr 2006 13:04:56 +0000
Subject: [C++-sig] void pointers as arguments and return values ( using
	pyplusplus )
Message-ID: <BAY108-F2555CFFA034ACE8F925810B4C50@phx.gbl>

Hi,

What is the status of void pointer handling in pyplusplus ? I have been 
trying to follow the multiple threads but I seem to get lost in all the 
different patches.

So far, if I understand correctly, the cvs trunk has the patch. I should be 
able to get the latest version, compile boost.python and everything should 
work fine without any changes to pyplusplus. Am I right ?

By the way, I did all those steps and it didn't work so I am trying to 
remove all known facts before looking for my problem...

Thanks


Rodrigue Cloutier




From guillaume.desticourt at gmail.com  Wed Apr 19 16:51:13 2006
From: guillaume.desticourt at gmail.com (Guillaume Desticourt)
Date: Wed, 19 Apr 2006 15:51:13 +0100
Subject: [C++-sig] wrapping member functions with incomplete types parameters
Message-ID: <609a651c0604190751u2fd3ccb6r1f0255268133d2be@mail.gmail.com>

hello,

i ve got a problem trying to wrap a constructor whose argument is a
reference to an incomplete type:
i have found this link that highlights the issue, but sadly doesn t come
with a solution  :)
http://mail.python.org/pipermail/c++-sig/2003-January/003352.html

i ve written some code which illustrates the problem:

ultimatetrader% cat test.cc

#include <boost/python/module.hpp>
#include <boost/python.hpp>

using namespace boost::python;


typedef struct unknown problem_t;

class Pouet
{
public:
  Pouet(problem_t& pb) :
    pb_(pb)
  {
  }
private:
  problem_t& pb_;

};


BOOST_PYTHON_MODULE(pyunknown)
{
  class_<Pouet>("PyPouet", init<problem_t&>())
    ;
}

and the error message:
test.cc:26:   instantiated from here
/usr/include/boost/python/type_id.hpp:79: error: invalid use of undefined
type
    'unknown'
test.cc:8: error: forward declaration of 'unknown'

thanks you for your help,


--
Guillaume Desticourt
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060419/a009cf4c/attachment.htm>

From roman.yakovenko at gmail.com  Wed Apr 19 18:49:07 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Wed, 19 Apr 2006 19:49:07 +0300
Subject: [C++-sig] void pointers as arguments and return values ( using
	pyplusplus )
In-Reply-To: <BAY108-F2555CFFA034ACE8F925810B4C50@phx.gbl>
References: <BAY108-F2555CFFA034ACE8F925810B4C50@phx.gbl>
Message-ID: <7465b6170604190949r437f987an7c17388b009dbb1a@mail.gmail.com>

On 4/19/06, Rod Cloutier <rodlist at hotmail.com> wrote:
> Hi,
>
> What is the status of void pointer handling in pyplusplus ? I have been
> trying to follow the multiple threads but I seem to get lost in all the
> different patches.

As for me they just work. I run a tester ( call_policies_tester.py )
that checks this
issue.

> So far, if I understand correctly, the cvs trunk has the patch. I should be
> able to get the latest version, compile boost.python and everything should
> work fine without any changes to pyplusplus. Am I right ?

Right

> By the way, I did all those steps and it didn't work so I am trying to
> remove all known facts before looking for my problem...

Good luck. Also my advise is: take a look on boost.python tests and
pyplusplus unit tests.
( Search for void* ptr and try to run relevant tests )

Also if you find some issues with pyplusplus, please report them to
pygccxml-development
list: http://lists.sourceforge.net/lists/listinfo/pygccxml-development

> Thanks
>
> Rodrigue Cloutier

Thanks
--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From dave at boost-consulting.com  Thu Apr 20 01:25:26 2006
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 19 Apr 2006 16:25:26 -0700
Subject: [C++-sig] none
In-Reply-To: <223C228C-DC27-41BE-8E92-E439736F29CB@earthbrowser.com> (Matthew
	Giger's message of "Wed, 19 Apr 2006 12:12:36 -0700")
References: <223C228C-DC27-41BE-8E92-E439736F29CB@earthbrowser.com>
Message-ID: <u8xq1ywpl.fsf@boost-consulting.com>


Hi Matthew,

It's better to post these things to the C++-sig:
http://www.boost.org/more/mailing_lists.htm#cplussig

Matthew Giger <mgiger at earthbrowser.com> writes:

> Just thought I'd pass along a fix for a crashing bug to the
> builtin_converters.cpp code:
>
> Change:
>
> std::wstring result(::PyObject_Length(intermediate), L' ');
> int err = PyUnicode_AsWideChar(
> 	(PyUnicodeObject *)intermediate
> 	, result.size() ? &result[0] : 0
> 	, result.size());
>
> if (err == -1)
> 	throw_error_already_set();
>
>
> To:
>
> std::wstring result;
> int len = ::PyObject_Length(intermediate);
> if(len > 0)
> {
>   result.resize(len, L' ');
>   int err = PyUnicode_AsWideChar(
> 	 (PyUnicodeObject *)intermediate
> 	 , result.size() ? &result[0] : 0
> 	 , result.size());
>
>   if (err == -1)
> 	  throw_error_already_set();
> }
>

Can you explain why this is necessary and what causes the crash you're
reporting?

Also, can you submit a minimal failing testcase?  Normally I don't
make fixes withouut an example I can check in as a regression test so
that the fix never gets undone.

> Thanks for writing a great library.

You're welcome.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com


From rwgk at yahoo.com  Wed Apr 19 23:05:22 2006
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Wed, 19 Apr 2006 14:05:22 -0700 (PDT)
Subject: [C++-sig] Wrapping pointers
In-Reply-To: <200604190420.30241.dominic.sacre@gmx.de>
Message-ID: <20060419210522.92608.qmail@web31512.mail.mud.yahoo.com>

--- Dominic Sacr??? <dominic.sacre at gmx.de> wrote:
> .def("get_p", &A::get_p, return_value_policy<reference_existing_object>())
> 
> but I'd rather be able to access p as a data member rather than a method. 
> Any ideas?

I'd try:

  .add_property("p", make_function(&A::get_p,
return_value_policy<reference_existing_object>()))

See also: http://www.boost.org/libs/python/doc/v2/make_function.html


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 


From dave at boost-consulting.com  Thu Apr 20 03:18:16 2006
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 19 Apr 2006 18:18:16 -0700
Subject: [C++-sig] none
In-Reply-To: <F3F545DB-9249-4CCA-BEF1-9EC9737419BA@earthbrowser.com> (Matthew
	Giger's message of "Wed, 19 Apr 2006 14:28:25 -0700")
References: <223C228C-DC27-41BE-8E92-E439736F29CB@earthbrowser.com>
	<u8xq1ywpl.fsf@boost-consulting.com>
	<F3F545DB-9249-4CCA-BEF1-9EC9737419BA@earthbrowser.com>
Message-ID: <uy7y1xcx3.fsf@boost-consulting.com>

Matthew Giger <mgiger at earthbrowser.com> writes:

> Sorry I didn't go through normal channels.  

Okay, but I would appreciate it if you'd do that now.

> I don't have an easy way
> to provide a regression test since I currently build the boost
> libraries into my Mac Xcode project and not through the standard
> build interface.

How you build it isn't really important.  The standard regression test
is a C++ file containing an extension module and a Python file
containing a driver for the test case.

> The Python code that crashed it was like:
>
> uni_str = u'some string'
> uni_len = len(uni_str)
> uni_str = uni_str[uni_len:]

So far I see no obvious crash, nor even any PYthon code that would
cause Boost.Python to get involved.

> The reasoning is fairly straightforward. If PyObject_Length() returns
> 0, no memory is allocated for the wstring, thus passing in a pointer
> to the first element if the wstring (never recommended, but we all do
> it). 

I don't see any such pointer passing.

  result.size() ? &result[0] : 0

ensures that no such pointer is passed.

> Logically, the result.size() being 0 should bypass the dereference

Yes, and it does, unless your compiler is broken.  And even if your
compiler *is* broken, I don't know of an implementation where
&result[0] would ever cause a crash.

> and PyUnicode_AsWideChar should recognize that it
> shouldn't do anything with a 0 length conversion. I've looked in the
> latest python source and PyUnicode_AsWideChar looks ok, but on the
> Mac we are using Python 2.3...  

And?  Python 2.3 is broken, and dereferences the pointer even if the
provided length is zero?  That seems unlikely to me; have you
confirmed it?

> Once I made this fix, it no longer crashed. 

That doesn't mean much.  Your code could be broken in other ways.
That's why a minimal test case is so important.

> It couldn't hurt to explicitly check the length for 0 rather than
> relying on wstring and Python to be smart about it.

I do explicitly check.  I'm not sure what smartness you're referring
to.

> Again, sorry this isn't through normal channels. I will do that if
> and when I have another bug I find.

Thanks.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com


From dominic.sacre at gmx.de  Thu Apr 20 00:44:41 2006
From: dominic.sacre at gmx.de (Dominic =?iso-8859-1?q?Sacr=E9?=)
Date: Thu, 20 Apr 2006 00:44:41 +0200
Subject: [C++-sig] Wrapping pointers
In-Reply-To: <20060419210522.92608.qmail@web31512.mail.mud.yahoo.com>
References: <20060419210522.92608.qmail@web31512.mail.mud.yahoo.com>
Message-ID: <200604200044.41722.dominic.sacre@gmx.de>

On Wednesday, 19. April 2006 23:05, Ralf W. Grosse-Kunstleve wrote:
>   .add_property("p", make_function(&A::get_p,
> return_value_policy<reference_existing_object>()))

Thanks, that did the trick!

    Dominic


From fatfatson at gmail.com  Thu Apr 20 08:25:14 2006
From: fatfatson at gmail.com (=?GB2312?B?0rG+p7XL?=)
Date: Thu, 20 Apr 2006 14:25:14 +0800
Subject: [C++-sig] how to set a global object for python environment?
Message-ID: <6d415ba30604192325h2fbf20d1s1bcf7b1c95372f93@mail.gmail.com>

i have created some global object in C++,
now i hope them could be visited from python,
as in ASP,some one implemented the "Request","Response",then we use it in
vbs.

what i want exactly is that: expose a created object to python,but not a
class definition, and when the object is modified in python, we can feel
this in C++.

is there a  simple approach?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060420/3441eb9e/attachment.htm>

From roman.yakovenko at gmail.com  Thu Apr 20 10:19:31 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Thu, 20 Apr 2006 11:19:31 +0300
Subject: [C++-sig] how to set a global object for python environment?
In-Reply-To: <6d415ba30604192325h2fbf20d1s1bcf7b1c95372f93@mail.gmail.com>
References: <6d415ba30604192325h2fbf20d1s1bcf7b1c95372f93@mail.gmail.com>
Message-ID: <7465b6170604200119x5bbd94ebmab06daf7b3a8e7f2@mail.gmail.com>

On 4/20/06, ??? <fatfatson at gmail.com> wrote:
>
> i have created some global object in C++,
> now i hope them could be visited from python,
> as in ASP,some one implemented the "Request","Response",then we use it in
> vbs.
>
> what i want exactly is that: expose a created object to python,but not a
> class definition, and when the object is modified in python, we can feel
> this in C++.
>
> is there a  simple approach?

yes and now:

1. You can expose class without exposing class definition
2.  boost::python::scope().attr( "xyz" ) = <your class instance>;

--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/

From roman.yakovenko at gmail.com  Thu Apr 20 10:20:04 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Thu, 20 Apr 2006 11:20:04 +0300
Subject: [C++-sig] how to set a global object for python environment?
In-Reply-To: <7465b6170604200119x5bbd94ebmab06daf7b3a8e7f2@mail.gmail.com>
References: <6d415ba30604192325h2fbf20d1s1bcf7b1c95372f93@mail.gmail.com>
	<7465b6170604200119x5bbd94ebmab06daf7b3a8e7f2@mail.gmail.com>
Message-ID: <7465b6170604200120k5dc3a201xd60152a741d154ea@mail.gmail.com>

On 4/20/06, Roman Yakovenko <roman.yakovenko at gmail.com> wrote:
> On 4/20/06, ??? <fatfatson at gmail.com> wrote:
> >
> > i have created some global object in C++,
> > now i hope them could be visited from python,
> > as in ASP,some one implemented the "Request","Response",then we use it in
> > vbs.
> >
> > what i want exactly is that: expose a created object to python,but not a
> > class definition, and when the object is modified in python, we can feel
> > this in C++.
> >
> > is there a  simple approach?
>
> yes and now:
   ^^^^^^^^^^^^^
Sorry, I meant no

>
> 1. You can expose class without exposing class definition
> 2.  boost::python::scope().attr( "xyz" ) = <your class instance>;
>
> --
> Roman Yakovenko
> C++ Python language binding
> http://www.language-binding.net/
>


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/

From abhi at qualcomm.com  Fri Apr 21 03:10:16 2006
From: abhi at qualcomm.com (Abhi)
Date: Thu, 20 Apr 2006 18:10:16 -0700
Subject: [C++-sig] pyplusplus annoyance
Message-ID: <F82F3445DCF24C8156FE8DE4@[10.30.128.28]>

I noticed that whenever I run py++, the generated file always contains 2 
#includes for the original c++ header files.

#include "boost/python.hpp"
#ifdef _MSC_VER
    #pragma hdrstop
#endif //_MSC_VER

#include "test.h"

#include "test.h"



Is it possible to suppress this (ie have only one #include generated for 
each different file)

thanks
- Abhi



From abhi at qualcomm.com  Fri Apr 21 03:21:17 2006
From: abhi at qualcomm.com (Abhi)
Date: Thu, 20 Apr 2006 18:21:17 -0700
Subject: [C++-sig] Suppressing assignment operator in Boost.Python
Message-ID: <EAB997874A65D9461A0047B5@[10.30.128.28]>

Consider a simplified example:

// Example
#include <iostream>
class A
{
public:
  int x;
  A() : x(5) {}

private:
  const A& operator=(const A& a)
  {
    std::cout << "A::=" << std::endl;
    x = a.x;
    return *this;
  }
  A(const A& a);

};

class B
{
public:
  A aObj;
  B()
  {
    aObj.x = 6;
  }
};

Notice that the assignment operator is private, hence the aObj inside B 
needs to be manipulated in place.

The boost.python binding for this looks like (and this is what py++ also 
generates out-of-the-box):

#include "test.cpp"


namespace bp = boost::python;



BOOST_PYTHON_MODULE(Boost_YMI){
  bp::class_< B, boost::noncopyable>( "B" )
        .def( bp::init< >()[bp::default_call_policies()] )
       .def_readwrite("aObj", &B::aObj)
    ;


    bp::class_< A, boost::noncopyable >( "A" )
        .def( bp::init< >()[bp::default_call_policies()] )
        .def_readwrite( "x", &A::x );
}


But this does not compile since Boost.Python generates an assignment 
operator for the attribute aObj (which has a private assignment operator).

Is this anyway to tell Boost.Python to not generate the assignment operator 
for aObj of B.

One workaround (which is not at all elegant!) is for me to write the 
following hand-crafted methods:

void set(B& self, A& a)
{
  self.aObj.x = a.x;
}

const A& get(B& self)
{
  return self.aObj;
}

 and then not expose the aObj as an attribute but rather through these 
methods in python:
    bp::class_< B, boost::noncopyable>( "B" )
        .def( bp::init< >()[bp::default_call_policies()] )
        .def("set", &set)
        .def("get", &get, bp::return_internal_reference<>() )



thanks
- Abhi



From roman.yakovenko at gmail.com  Fri Apr 21 10:10:23 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Fri, 21 Apr 2006 11:10:23 +0300
Subject: [C++-sig] pyplusplus annoyance
In-Reply-To: <F82F3445DCF24C8156FE8DE4@10.30.128.28>
References: <F82F3445DCF24C8156FE8DE4@10.30.128.28>
Message-ID: <7465b6170604210110k65c0759dm97e9834d057a5f01@mail.gmail.com>

On 4/21/06, Abhi <abhi at qualcomm.com> wrote:
> I noticed that whenever I run py++, the generated file always contains 2
> #includes for the original c++ header files.
>
> #include "boost/python.hpp"
> #ifdef _MSC_VER
>     #pragma hdrstop
> #endif //_MSC_VER
>
> #include "test.h"
>
> #include "test.h"
>
>
>
> Is it possible to suppress this (ie have only one #include generated for
> each different file)

I'll be glad to fix this issue. Can you send your script that use py++
and small header file, that will reproduce the problem.

> thanks

Also, can you post all issues regarding py++ to its mailing list:
https://lists.sourceforge.net/lists/listinfo/pygccxml-development

> - Abhi

Thanks
--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From Nicolas.Rougier at loria.fr  Fri Apr 21 10:58:24 2006
From: Nicolas.Rougier at loria.fr (Nicolas Rougier)
Date: Fri, 21 Apr 2006 10:58:24 +0200
Subject: [C++-sig] Request for comment on "tutorial"
Message-ID: <20060421105824.12238c92@dagda.loria.fr>



Hi all,

I've tried to gather what I've learned so far about boost python and I
wrote a simple example to illustrate some mechanisms. Basically, I tried
to illustrate:

 *   - inheritance
 *   - class having back references onto the PyObject that holds
 *      the underlying C++ object
 *   - vector indexing suite
 *      (that gives access to C++ vector-based object "a la python"
 *   - dotted notation for children
 *      (child access via father.child or father[0])
 *   - simple deep clone mechanism for containers
 *   - creation of python objects on c++ side
 *   - docstrings

Can you have a look at the code to check I did it the proper way
(and if you might find it useful at all) ?


Nicolas
-------------- next part --------------
A non-text attachment was scrubbed...
Name: core.cc
Type: text/x-c++src
Size: 5413 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060421/6944a830/attachment.cc>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: core.py
Type: text/x-python
Size: 1746 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060421/6944a830/attachment.py>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Makefile
Type: application/octet-stream
Size: 100 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060421/6944a830/attachment.obj>

From roman.yakovenko at gmail.com  Fri Apr 21 13:07:02 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Fri, 21 Apr 2006 14:07:02 +0300
Subject: [C++-sig] Suppressing assignment operator in Boost.Python
In-Reply-To: <EAB997874A65D9461A0047B5@10.30.128.28>
References: <EAB997874A65D9461A0047B5@10.30.128.28>
Message-ID: <7465b6170604210407u45666844scc3b962f7d7f6a3c@mail.gmail.com>

On 4/21/06, Abhi <abhi at qualcomm.com> wrote:
> namespace bp = boost::python;
>
>
> BOOST_PYTHON_MODULE(Boost_YMI){
>   bp::class_< B, boost::noncopyable>( "B" )
>         .def( bp::init< >()[bp::default_call_policies()] )
>        .def_readwrite("aObj", &B::aObj)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

This one has been fixed and unit test has been added.
Now pyplusplus will generate def_readonly.
You will not be able to assign new object to aObj, but you will be
able to modify
it's state.

Quick work around is next:
mb = module_builder_t( .... )
cls_b = mb.class_( 'B' )
cls_b.member_variable( 'aObj' ).exclude()
cls_b.add_code( 'def_readonly( "aObj", &B::aObj )' )

>     ;
>
>
>     bp::class_< A, boost::noncopyable >( "A" )
>         .def( bp::init< >()[bp::default_call_policies()] )
>         .def_readwrite( "x", &A::x );
> }
>
>
> But this does not compile since Boost.Python generates an assignment
> operator for the attribute aObj (which has a private assignment operator).
>
> Is this anyway to tell Boost.Python to not generate the assignment operator
> for aObj of B.

I am not sure that I fully understand you, but proposed solution
should work for you.

> One workaround (which is not at all elegant!) is for me to write the
> following hand-crafted methods:
>
> void set(B& self, A& a)
> {
>   self.aObj.x = a.x;
> }
>
> const A& get(B& self)
> {
>   return self.aObj;
> }
>
>  and then not expose the aObj as an attribute but rather through these
> methods in python:
>     bp::class_< B, boost::noncopyable>( "B" )
>         .def( bp::init< >()[bp::default_call_policies()] )
>         .def("set", &set)
>         .def("get", &get, bp::return_internal_reference<>() )

I don't think you need this.

>
> thanks
> - Abhi


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From abhi at qualcomm.com  Fri Apr 21 16:36:01 2006
From: abhi at qualcomm.com (Abhi)
Date: Fri, 21 Apr 2006 07:36:01 -0700
Subject: [C++-sig] Suppressing assignment operator in Boost.Python
In-Reply-To: <7465b6170604210407u45666844scc3b962f7d7f6a3c@mail.gmail.com>
References: <EAB997874A65D9461A0047B5@10.30.128.28>
	<7465b6170604210407u45666844scc3b962f7d7f6a3c@mail.gmail.com>
Message-ID: <C7AB9D0210165D1EEFFA80F4@[10.50.0.159]>



--On Friday, April 21, 2006 2:07 PM +0300 Roman Yakovenko 
<roman.yakovenko at gmail.com> wrote:

> On 4/21/06, Abhi <abhi at qualcomm.com> wrote:
>> namespace bp = boost::python;
>>
>>
>> BOOST_PYTHON_MODULE(Boost_YMI){
>>   bp::class_< B, boost::noncopyable>( "B" )
>>         .def( bp::init< >()[bp::default_call_policies()] )
>>        .def_readwrite("aObj", &B::aObj)
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>
> This one has been fixed and unit test has been added.
> Now pyplusplus will generate def_readonly.
> You will not be able to assign new object to aObj, but you will be
> able to modify
> it's state.

I want to be able to modify it in place.
Like this:
-----------------------------C++
// Example
#include <iostream>
class A
{
public:
  int x;
  A() : x(5) {}

private:
  const A& operator=(const A& a)
  {
    std::cout << "A::=" << std::endl;
    x = a.x;
    return *this;
  }
  A(const A& a);

};

class B
{
public:
  A aObj;
  B()
  {
    aObj.x = 6;
  }
};
-------------------------------
In python:

---------------.py
b = B()
a = A()

# No assignment operator for A
# so we cannot do
# b.aObj = someNewA
# but we can modify it in place
b.aObj.x  = 10;
# and also read it in place
y = b.aObj.x
--------------------

How do I generate Boost.Python bindings in py++ for doing this?
And also what does the .def look like (ie the Boost.Python binding look 
like). I could just do the code.wrapper in py++ for it.

thanks
- Abhi



>
> Quick work around is next:
> mb = module_builder_t( .... )
> cls_b = mb.class_( 'B' )
> cls_b.member_variable( 'aObj' ).exclude()
> cls_b.add_code( 'def_readonly( "aObj", &B::aObj )' )
>
>>     ;
>>
>>
>>     bp::class_< A, boost::noncopyable >( "A" )
>>         .def( bp::init< >()[bp::default_call_policies()] )
>>         .def_readwrite( "x", &A::x );
>> }
>>
>>
>> But this does not compile since Boost.Python generates an assignment
>> operator for the attribute aObj (which has a private assignment
>> operator).
>>
>> Is this anyway to tell Boost.Python to not generate the assignment
>> operator for aObj of B.
>
> I am not sure that I fully understand you, but proposed solution
> should work for you.
>
>> One workaround (which is not at all elegant!) is for me to write the
>> following hand-crafted methods:
>>
>> void set(B& self, A& a)
>> {
>>   self.aObj.x = a.x;
>> }
>>
>> const A& get(B& self)
>> {
>>   return self.aObj;
>> }
>>
>>  and then not expose the aObj as an attribute but rather through these
>> methods in python:
>>     bp::class_< B, boost::noncopyable>( "B" )
>>         .def( bp::init< >()[bp::default_call_policies()] )
>>         .def("set", &set)
>>         .def("get", &get, bp::return_internal_reference<>() )
>
> I don't think you need this.
>
>>
>> thanks
>> - Abhi
>
>
> --
> Roman Yakovenko
> C++ Python language binding
> http://www.language-binding.net/






From roman.yakovenko at gmail.com  Fri Apr 21 18:59:00 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Fri, 21 Apr 2006 19:59:00 +0300
Subject: [C++-sig] Suppressing assignment operator in Boost.Python
In-Reply-To: <C7AB9D0210165D1EEFFA80F4@10.50.0.159>
References: <EAB997874A65D9461A0047B5@10.30.128.28>
	<7465b6170604210407u45666844scc3b962f7d7f6a3c@mail.gmail.com>
	<C7AB9D0210165D1EEFFA80F4@10.50.0.159>
Message-ID: <7465b6170604210959r7701bd34vd3607ecef12b28da@mail.gmail.com>

On 4/21/06, Abhi <abhi at qualcomm.com> wrote:
>
>
> --On Friday, April 21, 2006 2:07 PM +0300 Roman Yakovenko
> <roman.yakovenko at gmail.com> wrote:
>
> > On 4/21/06, Abhi <abhi at qualcomm.com> wrote:
> >> namespace bp = boost::python;
> >>
> >>
> >> BOOST_PYTHON_MODULE(Boost_YMI){
> >>   bp::class_< B, boost::noncopyable>( "B" )
> >>         .def( bp::init< >()[bp::default_call_policies()] )
> >>        .def_readwrite("aObj", &B::aObj)
> > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> >
> > This one has been fixed and unit test has been added.
> > Now pyplusplus will generate def_readonly.
> > You will not be able to assign new object to aObj, but you will be
> > able to modify
> > it's state.
>
> I want to be able to modify it in place.
> Like this:
> -----------------------------C++
> // Example
> #include <iostream>
> class A
> {
> public:
>   int x;
>   A() : x(5) {}
>
> private:
>   const A& operator=(const A& a)
>   {
>     std::cout << "A::=" << std::endl;
>     x = a.x;
>     return *this;
>   }
>   A(const A& a);
>
> };
>
> class B
> {
> public:
>   A aObj;
>   B()
>   {
>     aObj.x = 6;
>   }
> };
> -------------------------------
> In python:
>
> ---------------.py
> b = B()
> a = A()
>
> # No assignment operator for A
> # so we cannot do
> # b.aObj = someNewA
> # but we can modify it in place
> b.aObj.x  = 10;
> # and also read it in place
> y = b.aObj.x
> --------------------
>
> How do I generate Boost.Python bindings in py++ for doing this?
> And also what does the .def look like (ie the Boost.Python binding look
> like). I could just do the code.wrapper in py++ for it.

Please read my answer to your mail. I explained what should you do in
order to get
desired result.

> thanks
> - Abhi


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From abhi at qualcomm.com  Fri Apr 21 19:18:27 2006
From: abhi at qualcomm.com (Abhi)
Date: Fri, 21 Apr 2006 10:18:27 -0700
Subject: [C++-sig] Suppressing assignment operator in Boost.Python
In-Reply-To: <7465b6170604210959r7701bd34vd3607ecef12b28da@mail.gmail.com>
References: <EAB997874A65D9461A0047B5@10.30.128.28>	
	<7465b6170604210407u45666844scc3b962f7d7f6a3c@mail.gmail.com>	
	<C7AB9D0210165D1EEFFA80F4@10.50.0.159>
	<7465b6170604210959r7701bd34vd3607ecef12b28da@mail.gmail.com>
Message-ID: <A18138C3B54AA81D74F04935@[10.30.5.3]>


I realized that .def_readonly avoids the assignment of the attribute, but 
not its data members. Hence, it allows modification in place.

Thanks Roman
- Abhi


--On Friday, April 21, 2006 7:59 PM +0300 Roman Yakovenko 
<roman.yakovenko at gmail.com> wrote:

> On 4/21/06, Abhi <abhi at qualcomm.com> wrote:
>>
>>
>> --On Friday, April 21, 2006 2:07 PM +0300 Roman Yakovenko
>> <roman.yakovenko at gmail.com> wrote:
>>
>> > On 4/21/06, Abhi <abhi at qualcomm.com> wrote:
>> >> namespace bp = boost::python;
>> >>
>> >>
>> >> BOOST_PYTHON_MODULE(Boost_YMI){
>> >>   bp::class_< B, boost::noncopyable>( "B" )
>> >>         .def( bp::init< >()[bp::default_call_policies()] )
>> >>        .def_readwrite("aObj", &B::aObj)
>> > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>> >
>> > This one has been fixed and unit test has been added.
>> > Now pyplusplus will generate def_readonly.
>> > You will not be able to assign new object to aObj, but you will be
>> > able to modify
>> > it's state.
>>
>> I want to be able to modify it in place.
>> Like this:
>> -----------------------------C++
>> // Example
>> # include <iostream>
>> class A
>> {
>> public:
>>   int x;
>>   A() : x(5) {}
>>
>> private:
>>   const A& operator=(const A& a)
>>   {
>>     std::cout << "A::=" << std::endl;
>>     x = a.x;
>>     return *this;
>>   }
>>   A(const A& a);
>>
>> };
>>
>> class B
>> {
>> public:
>>   A aObj;
>>   B()
>>   {
>>     aObj.x = 6;
>>   }
>> };
>> -------------------------------
>> In python:
>>
>> ---------------.py
>> b = B()
>> a = A()
>>
>> # No assignment operator for A
>> # so we cannot do
>> # b.aObj = someNewA
>> # but we can modify it in place
>> b.aObj.x  = 10;
>> # and also read it in place
>> y = b.aObj.x
>> --------------------
>>
>> How do I generate Boost.Python bindings in py++ for doing this?
>> And also what does the .def look like (ie the Boost.Python binding look
>> like). I could just do the code.wrapper in py++ for it.
>
> Please read my answer to your mail. I explained what should you do in
> order to get
> desired result.
>
>> thanks
>> - Abhi
>
>
> --
> Roman Yakovenko
> C++ Python language binding
> http://www.language-binding.net/






From roman.yakovenko at gmail.com  Sat Apr 22 12:05:16 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Sat, 22 Apr 2006 13:05:16 +0300
Subject: [C++-sig] Wrapping pointers
In-Reply-To: <200604200044.41722.dominic.sacre@gmx.de>
References: <20060419210522.92608.qmail@web31512.mail.mud.yahoo.com>
	<200604200044.41722.dominic.sacre@gmx.de>
Message-ID: <7465b6170604220305t3ed839b4wb6c3a7c0e60bd5e8@mail.gmail.com>

On 4/20/06, Dominic Sacr? <dominic.sacre at gmx.de> wrote:
> On Wednesday, 19. April 2006 23:05, Ralf W. Grosse-Kunstleve wrote:
> >   .add_property("p", make_function(&A::get_p,
> > return_value_policy<reference_existing_object>()))
>
> Thanks, that did the trick!
>
>     Dominic

I am working on pyplusplus to generate right code for this situation.
The case discussed here is already implemented ( not committed yet ).
pyplusplus generates class wrapper and adds "get" method to it.
I attached the file that demonstrates the use case.

My question about "set" functionality. I am not sure what call
policies I should set by
default? default or with_custodian_and_ward_postcall< 0, 1 >
I think, that first one is a mistake, while the second one is correct.
I just want to be sure.

Also, right now I don't see differences between regular member
variable and static
member variable as for the call policies. Am I right?

Thanks

--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: mvpointer.cpp
Type: text/x-c++src
Size: 3448 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060422/a42806c1/attachment.cpp>

From capak at inputwish.com  Mon Apr 24 10:34:00 2006
From: capak at inputwish.com (Libor Capak)
Date: Mon, 24 Apr 2006 10:34:00 +0200
Subject: [C++-sig] singleton wrapping
Message-ID: <20060424083400.GA28600@irma.inputwish.local>

hello,

i'm quite new to boost::python. i'd like to ask how you solve
singleton wrapping.
i have simple template:

template<typename T> class Singleton
{
  static T* singleton;
public:
  //.. blabla ..
  static T& instance()
  {
     mustBe(singleton); return *singleton;
  }
  // .. blabla ..
};

and simple class, for example:

class A : public Singleton<A>
{
public:
  void foo();
};

and now to export into python module i'm using wrapper class:

class A_wrap
{
  static void foo() { A.instance().foo(); }
};

BOOST_PYTHON_MODULE(test)
{
  class_<A_wrap>("A", no_init)
    .def("foo", &A_wrap::foo)
    .staticmethod("foo")
    ;
}

is there any trick how to skip writing class A_wrap and still 
use method A::foo() as staticmethod from python?

Libor Capak



From roman.yakovenko at gmail.com  Mon Apr 24 10:17:45 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Mon, 24 Apr 2006 11:17:45 +0300
Subject: [C++-sig] singleton wrapping
In-Reply-To: <20060424083400.GA28600@irma.inputwish.local>
References: <20060424083400.GA28600@irma.inputwish.local>
Message-ID: <7465b6170604240117r610948e4v25ca0782e5829570@mail.gmail.com>

On 4/24/06, Libor Capak <capak at inputwish.com> wrote:
> hello,
> ...

> is there any trick how to skip writing class A_wrap and still
> use method A::foo() as staticmethod from python?

I think next link can help you:
http://boost.org/libs/python/doc/v2/scope.html#scope-spec

And next line can do the trick, but I am not sure

scope().attr("A_singleton") = &A::instance();



> Libor Capak


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From juho.makinen at gmail.com  Tue Apr 25 21:39:08 2006
From: juho.makinen at gmail.com (=?ISO-8859-1?Q?Juho_M=E4kinen?=)
Date: Tue, 25 Apr 2006 22:39:08 +0300
Subject: [C++-sig] Howto expose exception classes using boost.python?
Message-ID: <4a41ceba0604251239x340ee52am7aef5c86bc6c9cb2@mail.gmail.com>

I have stumbled across this problem when I started looking the right
way how to expose a bunch of C++ Exception classes to python.

I have a "master" exception class, which is delivered from
std::exception. All other exception classes in our project are base on
this.

As far as I know, Python requires that the exceptions must be
delivered from PyExc_Exception, or a class which is delivered from
PyExc_Exception, but I haven't found any way how to expose a C++ class
which delivers from a python class. There's also PyErr_NewException()
available within Python C API.

This would be the ideal way how I'd like to use our exceptions.

PyErr_SetObject(PyExc_MyException,  InstanceOfMyException)
-or in python-
raise mymodule.MyException(arg1, arg2)

I know that I can expose C++ exceptions with boost.python (the
PodBayDoorException example) using
boost::python::register_exception_translator, but I want to be able to
distinguish my different C++ exceptions at the python side. (eg. It's
not enough that all my exceptions are translated into
PyExc_RuntimeError, or some other standard python exception)

All ideas how to expose exception hierarchy from C++ to Python are appreciated.

Thanks in advance,

 - Juho M?kinen


From craigwharris at comcast.net  Thu Apr 27 01:28:49 2006
From: craigwharris at comcast.net (craig harris)
Date: Wed, 26 Apr 2006 16:28:49 -0700
Subject: [C++-sig] Best way to use python as an  API inside a Qt/C++ app?
Message-ID: <20060426232901.8DC8F1E4005@bag.python.org>

I am using Qt as the GUI for a C++ app.  I would also like to use python as
an API for the application.  For instance when a menu item is pressed, a
stored python command is executed.

Is this wise?  And what is the best approach (tool) to implementing this?
PyQt, boost.python, other?

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060426/fa95bd9d/attachment.htm>

From seefeld at sympatico.ca  Thu Apr 27 02:20:50 2006
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Wed, 26 Apr 2006 20:20:50 -0400
Subject: [C++-sig] Best way to use python as an API inside a Qt/C++ app?
In-Reply-To: <20060426232901.8DC8F1E4005@bag.python.org>
References: <20060426232901.8DC8F1E4005@bag.python.org>
Message-ID: <44500E62.5060309@sympatico.ca>

craig harris wrote:
> I am using Qt as the GUI for a C++ app.  I would also like to use python 
> as an API for the application.  For instance when a menu item is 
> pressed, a stored python command is executed.
> 
> Is this wise?

That sounds very reasonable to me.

> And what is the best approach (tool) to implementing 
> this?  PyQt, boost.python, other?

That depends on what you mean by 'implement'. Executing some
python code via a menu is trivial. What is much less trivial is
to expose your application's state to python so you can access
and modify it from these commands. As far as Qt is concerned, there
already are a number of bindings (sip for Qt3, and boost.python for
Qt4). How to reflect your own code into python is a different issue.
Boost.python is definitely a good choice, though if you use a lot of
Qt internally it would be best to stay with the tool used to generate
the Qt bindings, and simply add your own classes.

Regards,
		Stefan



From roman.yakovenko at gmail.com  Thu Apr 27 07:44:01 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Thu, 27 Apr 2006 08:44:01 +0300
Subject: [C++-sig] Howto expose exception classes using boost.python?
In-Reply-To: <4a41ceba0604251239x340ee52am7aef5c86bc6c9cb2@mail.gmail.com>
References: <4a41ceba0604251239x340ee52am7aef5c86bc6c9cb2@mail.gmail.com>
Message-ID: <7465b6170604262244x5971cc89k89283af7c5f47837@mail.gmail.com>

On 4/25/06, Juho M?kinen <juho.makinen at gmail.com> wrote:
> I have stumbled across this problem when I started looking the right
> way how to expose a bunch of C++ Exception classes to python.
>
> I have a "master" exception class, which is delivered from
> std::exception. All other exception classes in our project are base on
> this.
>
> As far as I know, Python requires that the exceptions must be
> delivered from PyExc_Exception, or a class which is delivered from
> PyExc_Exception,

Are you sure? Where did you read it?

> but I haven't found any way how to expose a C++ class
> which delivers from a python class.

I think such functionality does not exists in boost.python library.
But I could be wrong.

>There's also PyErr_NewException() available within Python C API.
>
> This would be the ideal way how I'd like to use our exceptions.
>
> PyErr_SetObject(PyExc_MyException,  InstanceOfMyException)
> -or in python-
> raise mymodule.MyException(arg1, arg2)

Here is relevant piece of Python documentation:

void PyErr_SetObject( PyObject *type, PyObject *value)

This function is similar to PyErr_SetString() but lets you specify an
arbitrary Python object for the ``value'' of the exception.

I think that "an arbitrary" is the key to your solution.

In the past I created custom exception translator.
( I can not find code, so I will just write an algorithm )

Register all your exception classes as usual.

Write next function:

template<class MyCppException>
void translate_it( const MyCppException& err ){
    boost::python::object bp_err( err );

    extract PyObjectType and PyObject from py_err
    this exact code I don't remember, but it is not to difficult to find out.

    pay attention to reference count ( borrow or handle )

    PyErr_SetObject( py_type, py_error );
}

For every exception register it's translation using translate_it.

>From now your C++/Python code can throw/raise exceptions as usual.
They will be translated to Python.

The only (very) bad thing is that the code that treats exceptions now
should treat 2 exception kinds.

>
> I know that I can expose C++ exceptions with boost.python (the
> PodBayDoorException example) using
> boost::python::register_exception_translator, but I want to be able to
> distinguish my different C++ exceptions at the python side. (eg. It's
> not enough that all my exceptions are translated into
> PyExc_RuntimeError, or some other standard python exception)
>
> All ideas how to expose exception hierarchy from C++ to Python are appreciated.
>
> Thanks in advance,

I hope this will help you, also I am sure other people do know better solution.
In any case if you find a solution can you post it? I will add it to pyplusplus
( http://www.language-binding.net/pyplusplus/pyplusplus.html )

>  - Juho M?kinen


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From roman.yakovenko at gmail.com  Thu Apr 27 12:50:20 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Thu, 27 Apr 2006 13:50:20 +0300
Subject: [C++-sig] Fwd: Virtual Functions with Default Implementations
In-Reply-To: <7465b6170604270349o39af72bev199f5f6b0fa5d23@mail.gmail.com>
References: <7465b6170604270349o39af72bev199f5f6b0fa5d23@mail.gmail.com>
Message-ID: <7465b6170604270350q7d65c78j5872235307c5933e@mail.gmail.com>

Hi.
It seems to me that I miss something very important
( or very obvious )

Here is a relevant piece of code from tutorials( http://tinyurl.com/hw5hb )

struct Base
{
    virtual ~Base() {}
    virtual int f() { return 0; }
};

struct BaseWrap : Base, wrapper<Base>
{
    int f()
    {
        if (override f = this->get_override("f"))
            return f(); // *note*
        return Base::f();
    }

    int default_f() { return this->Base::f(); }
};

class_<BaseWrap, boost::noncopyable>("Base")
    .def("f", &Base::f, &BaseWrap::default_f) ;

Why boost.python requests to register BaseWrap::default_f?
It seems to me that BaseWrap::f implementation deals with this.

So what did I miss?
--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From iluetkeb at gmail.com  Thu Apr 27 14:15:26 2006
From: iluetkeb at gmail.com (=?ISO-8859-1?Q?Ingo_L=FCtkebohle?=)
Date: Thu, 27 Apr 2006 14:15:26 +0200
Subject: [C++-sig] status of signatures?
Message-ID: <fdd727f80604270515m3172c448pa043dd42ddee0792@mail.gmail.com>

Hi,

what is the status of the boost-signatures patch (originally by Ralf
IIRC)?  Is it being taken forwarded?  Is it planned for inclusion into
boost::python sometime soon?

I think more complete documentation generation like implemented by
that patch is one of the most important things still missing!  Having
read through the list I recognize there are awkward cases with
unwieldy signatures but are they a showstopper?  I figure that most
easy cases might be covered already.

Best Regards

Ingo


From roman.yakovenko at gmail.com  Thu Apr 27 14:34:42 2006
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Thu, 27 Apr 2006 15:34:42 +0300
Subject: [C++-sig] status of signatures?
In-Reply-To: <fdd727f80604270515m3172c448pa043dd42ddee0792@mail.gmail.com>
References: <fdd727f80604270515m3172c448pa043dd42ddee0792@mail.gmail.com>
Message-ID: <7465b6170604270534w1cf51004ide38ac4d15721e69@mail.gmail.com>

On 4/27/06, Ingo L?tkebohle <iluetkeb at gmail.com> wrote:
> Hi,
>
> what is the status of the boost-signatures patch (originally by Ralf
> IIRC)?  Is it being taken forwarded?  Is it planned for inclusion into
> boost::python sometime soon?

I could be wrong, but it is already in CVS. Take a look on cvs (
boost.python ) news.html file.


> Best Regards
>
> Ingo
--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/


From amohr at pixar.com  Thu Apr 27 18:34:05 2006
From: amohr at pixar.com (Alex Mohr)
Date: Thu, 27 Apr 2006 09:34:05 -0700
Subject: [C++-sig] Fwd: Virtual Functions with Default Implementations
In-Reply-To: <7465b6170604270350q7d65c78j5872235307c5933e@mail.gmail.com>
References: <7465b6170604270349o39af72bev199f5f6b0fa5d23@mail.gmail.com>
	<7465b6170604270350q7d65c78j5872235307c5933e@mail.gmail.com>
Message-ID: <4450F27D.3030003@pixar.com>

> Why boost.python requests to register BaseWrap::default_f?
> It seems to me that BaseWrap::f implementation deals with this.
> 
> So what did I miss?

I think it's so you can call Base's f() from your derived, overridden f 
without an infinite loop.  Something like this:

class Dervied(Base):
     def f(self):
         return Base.f(self) * 2

Here I don't want my Base.f(self) call to dispatch back to me, as this 
would cause an infinite recursion.

Alex


From padilla.henry at gmail.com  Fri Apr 28 19:21:28 2006
From: padilla.henry at gmail.com (Henry Padilla)
Date: Fri, 28 Apr 2006 13:21:28 -0400
Subject: [C++-sig] VC 2005 support in boost.python
Message-ID: <c5da86060604281021q1958ffbs3cdcf4fa6c183367@mail.gmail.com>

I play Civilization 4, that uses Boost.  We can't compile the SDK because
there's some incompatibility issues with VC 2005.

I was wondering if anyone still had incompatibility issues with VC 2005 or
if the macro BOOST_WORKAROUND was still needed.

Thanks for the help.
Tom P.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060428/4a7f070e/attachment.htm>

From Paul_Kunz at slac.stanford.edu  Fri Apr 28 19:43:30 2006
From: Paul_Kunz at slac.stanford.edu (Paul F. Kunz)
Date: Fri, 28 Apr 2006 10:43:30 -0700
Subject: [C++-sig] VC 2005 support in boost.python
In-Reply-To: <c5da86060604281021q1958ffbs3cdcf4fa6c183367@mail.gmail.com>
	(padilla.henry@gmail.com)
References: <c5da86060604281021q1958ffbs3cdcf4fa6c183367@mail.gmail.com>
Message-ID: <200604281743.k3SHhUuj017163@noric04.slac.stanford.edu>

>>>>> On Fri, 28 Apr 2006 13:21:28 -0400, "Henry Padilla" <padilla.henry at gmail.com> said:

> I play Civilization 4, that uses Boost.  We can't compile the SDK
> because there's some incompatibility issues with VC 2005.

> I was wondering if anyone still had incompatibility issues with VC
> 2005 or if the macro BOOST_WORKAROUND was still needed.

> Thanks for the help.  Tom P.

   At least for me, the Boost.Python libraries work with VC 2005.



From padilla.henry at gmail.com  Fri Apr 28 19:52:51 2006
From: padilla.henry at gmail.com (Henry Padilla)
Date: Fri, 28 Apr 2006 13:52:51 -0400
Subject: [C++-sig] VC 2005 support in boost.python
In-Reply-To: <200604281743.k3SHhUuj017163@noric04.slac.stanford.edu>
References: <c5da86060604281021q1958ffbs3cdcf4fa6c183367@mail.gmail.com>
	<200604281743.k3SHhUuj017163@noric04.slac.stanford.edu>
Message-ID: <c5da86060604281052u432d383lef98713f914f9bbf@mail.gmail.com>

The specific error we get is "error C2665:
'boost::**python::detail::make_getter'
: none of the 3 overloads could convert all the argument types".

Don't know if this is fixed and compatible with the Civ4 SDK.

Tom P.




On 4/28/06, Paul F. Kunz <Paul_Kunz at slac.stanford.edu> wrote:
>
> >>>>> On Fri, 28 Apr 2006 13:21:28 -0400, "Henry Padilla" <
> padilla.henry at gmail.com> said:
>
> > I play Civilization 4, that uses Boost.  We can't compile the SDK
> > because there's some incompatibility issues with VC 2005.
>
> > I was wondering if anyone still had incompatibility issues with VC
> > 2005 or if the macro BOOST_WORKAROUND was still needed.
>
> > Thanks for the help.  Tom P.
>
>   At least for me, the Boost.Python libraries work with VC 2005.
>
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20060428/1cbd9603/attachment.htm>

From Nicolas.Rougier at loria.fr  Sat Apr 29 17:18:47 2006
From: Nicolas.Rougier at loria.fr (Nicolas P. Rougier)
Date: Sat, 29 Apr 2006 17:18:47 +0200
Subject: [C++-sig] Simple boost example available
Message-ID: <1146323927.16912.2.camel@localhost.localdomain>


Hi all,

I put on http://www.loria.fr/~rougier/software/boost/ a simple example
showing some boost python features in action. Hope this can help
someone.


Nicolas Rougier.




From dave at boost-consulting.com  Sun Apr 30 06:38:40 2006
From: dave at boost-consulting.com (David Abrahams)
Date: Sun, 30 Apr 2006 00:38:40 -0400
Subject: [C++-sig] Howto expose exception classes using boost.python?
References: <4a41ceba0604251239x340ee52am7aef5c86bc6c9cb2@mail.gmail.com>
	<7465b6170604262244x5971cc89k89283af7c5f47837@mail.gmail.com>
Message-ID: <uejzfn0dr.fsf@boost-consulting.com>

"Roman Yakovenko" <roman.yakovenko at gmail.com> writes:

>> As far as I know, Python requires that the exceptions must be
>> delivered from PyExc_Exception, or a class which is delivered from
>> PyExc_Exception,
>
> Are you sure? Where did you read it?

Officially, it does require that exceptions be derived from
PyExc_Exception (I don't remember where that's documented), but
unofficially, you can throw anything :)

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com