From Adrian.Baetu at plato-ag.com  Tue Nov  2 17:51:27 2010
From: Adrian.Baetu at plato-ag.com (AdrianB)
Date: Tue, 2 Nov 2010 16:51:27 +0000 (UTC)
Subject: [C++-sig] boost::python executes wrong implementation
Message-ID: <loom.20101102T172513-950@post.gmane.org>

Hello

The Situation:
There are three modules involved:
a) myapi.pyd
b) myapi.dll
c) myserver.dll

myapi is a VC++ Project which exposes few classes and methods to python.
  myapi.pyd is a UNICODE build, to be used in python.
  myapi.dll is a ANSI build, to be used embedded by the myserver.dll

myserver.dll runs python embedded, and links myapi.dll

UseCase in Python:
# ------------------
import myapi                                      # myapi.pyd
session = myapi.Session( <some credentials> )     # myapi.pyd::Session
# at this moment myserver.dll gets loaded into the python process
# along with its embedded python module, called "myembedded"

import myembedded                                 # myserver.dll
embedded_session = myembedded.Session()           # myapi.dll::Session

#!! The ISSUE !!
embedded_session.myEmbeddedImpl()
# should call myapi.dll::Session::myEmbeddedImpl()

# -- what happens? --
instead of executing 
   myapi.dll::Session::myEmbeddedImpl()
it executes
   myapi.pyd::Session::myEmbeddedImpl()!!!

Why? And how can I avoid this?

Anny idea and help is greatly appreciated,
best regards,
Adrian


From nat at lindenlab.com  Tue Nov  2 18:04:08 2010
From: nat at lindenlab.com (Nat Linden)
Date: Tue, 2 Nov 2010 13:04:08 -0400
Subject: [C++-sig] boost::python executes wrong implementation
In-Reply-To: <loom.20101102T172513-950@post.gmane.org>
References: <loom.20101102T172513-950@post.gmane.org>
Message-ID: <AANLkTim79oQNxq3RqN=6nsRdbtEop3LkGhmrBF3u+6si@mail.gmail.com>

On Tue, Nov 2, 2010 at 12:51 PM, AdrianB <Adrian.Baetu at plato-ag.com> wrote:

import myembedded                                 # myserver.dll
> embedded_session = myembedded.Session()           # myapi.dll::Session
>
> #!! The ISSUE !!
> embedded_session.myEmbeddedImpl()
> # should call myapi.dll::Session::myEmbeddedImpl()
>
> # -- what happens? --
> instead of executing
>   myapi.dll::Session::myEmbeddedImpl()
> it executes
>   myapi.pyd::Session::myEmbeddedImpl()!!!
>
> Why? And how can I avoid this?
>

Try renaming myapi.dll to myotherapi.dll, or some renaming like that?

Since myapi.pyd *is* a .dll file, my guess is that the OS loader thinks:
since we've already loaded myapi.pyd, there's no need to load a file called
myapi.dll.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20101102/01268ba0/attachment.html>

From simwarg at gmail.com  Tue Nov  2 19:04:32 2010
From: simwarg at gmail.com (Simon W)
Date: Tue, 2 Nov 2010 19:04:32 +0100
Subject: [C++-sig] Reference count on class object and import()
In-Reply-To: <AANLkTim4shmGA8Luzx6T=0D1NoivEM_sYycXJuos=D_f@mail.gmail.com>
References: <AANLkTinykSkbBDD=Xofg65QRtq5Qg1vBQbZ3mH-4NXsx@mail.gmail.com>
	<AANLkTim4shmGA8Luzx6T=0D1NoivEM_sYycXJuos=D_f@mail.gmail.com>
Message-ID: <AANLkTinJLjZBfoUjD099M_vCsLnsFXq7wMCsyKyCPSo6@mail.gmail.com>

Alright, it was a couple of days ago since I last had a look on this and it
turns out that issue described in my first mail still persists. Although I
had a typo. Swap 4 with 3 on the reference count. Also, when Im using
PyImport_Import() directly it also shows 2 reference counts?



On Tue, Oct 26, 2010 at 11:29 PM, Simon W <simwarg at gmail.com> wrote:

> Of course, I did find another reference that I didn't notice. Python
> reference works as expected .. hehe!
>
>
> On Mon, Oct 25, 2010 at 7:35 PM, Simon W <simwarg at gmail.com> wrote:
>
>> Hey again,
>>
>> I'm trying to implement a load / unload module functionallity. When I
>> delete the module in the __main__ dict the module's reference count is still
>> > 1. I've been trying to find if I reference it anywhere else in my code but
>> I can't find anything! When I did a check in vc++ debugger found something.
>> When I run this code:
>>
>> object obj = import(name.c_str());
>>
>> *obj*'s *ob_refcnt* member is 2 when I check in the vc++ debugger.
>> Shouldn't it be 1?
>>
>> After that line, I put it in the __main__ dict like:
>>
>> mMainNamespace[ name.c_str() ] = obj;
>>
>> The reference count shows 4, as expected.
>>
>> When I'm entering my unload() function when I want to remove the module I
>> do like:
>> void unload()
>> {
>>     object t = mMainNamespace[ name.c_str() ];
>>     // reference count in t is now 4 ?
>>
>>      mMainNamespace[ name.c_str() ].del(); // I delete but it's not
>> unloaded properly because it's still referenced somewhere ..
>> }
>>
>> What does import() do? It must save some reference somewhere?
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20101102/71e42985/attachment.html>

From talljimbo at gmail.com  Tue Nov  2 19:34:14 2010
From: talljimbo at gmail.com (Jim Bosch)
Date: Tue, 02 Nov 2010 11:34:14 -0700
Subject: [C++-sig] Reference count on class object and import()
In-Reply-To: <AANLkTinJLjZBfoUjD099M_vCsLnsFXq7wMCsyKyCPSo6@mail.gmail.com>
References: <AANLkTinykSkbBDD=Xofg65QRtq5Qg1vBQbZ3mH-4NXsx@mail.gmail.com>	<AANLkTim4shmGA8Luzx6T=0D1NoivEM_sYycXJuos=D_f@mail.gmail.com>
	<AANLkTinJLjZBfoUjD099M_vCsLnsFXq7wMCsyKyCPSo6@mail.gmail.com>
Message-ID: <4CD059A6.3060606@gmail.com>

On 11/02/2010 11:04 AM, Simon W wrote:
> Alright, it was a couple of days ago since I last had a look on this and
> it turns out that issue described in my first mail still persists.
> Although I had a typo. Swap 4 with 3 on the reference count. Also, when
> Im using PyImport_Import() directly it also shows 2 reference counts?
>

Python stores a global list of loaded modules, as a cache to speed up 
future imports.  You can look them up in sys.modules, though I don't 
think trying to delete things from that list is allowed.  To my 
knowledge, there is no way to unload a Python module, though one can of 
course force a reload.

Jim

From Adrian.Baetu at plato-ag.com  Wed Nov  3 09:21:48 2010
From: Adrian.Baetu at plato-ag.com (AdrianB)
Date: Wed, 3 Nov 2010 08:21:48 +0000 (UTC)
Subject: [C++-sig] boost::python executes wrong implementation
References: <loom.20101102T172513-950@post.gmane.org>
	<AANLkTim79oQNxq3RqN=6nsRdbtEop3LkGhmrBF3u+6si@mail.gmail.com>
Message-ID: <loom.20101103T091816-350@post.gmane.org>

Nat Linden <nat <at> lindenlab.com> writes:
> Try renaming myapi.dll ...

thanks for the hint.
my real life scenario uses:
MY-API.DLL
myapi.pyd

sorry for not mentioning this <:o)

best regards
Adrian


From charlessolar at gmail.com  Wed Nov  3 15:22:46 2010
From: charlessolar at gmail.com (Charles Solar)
Date: Wed, 3 Nov 2010 09:22:46 -0500
Subject: [C++-sig] Python iterable to std map
Message-ID: <AANLkTinGjArs50XOr2+FSDFvSuFJ3kKsbR0VP00PusJ5@mail.gmail.com>

I have a class interface I expose to python that has methods such as begin,
end, rbegin, rend.  Its meant to be a class that C++ can iterate over.  I
now want to make the object defined in python, so python can pass the C++
library an object of python's class and C++ can use it like it would any
real C++ object derived from the same interface.
Since python does not really have a concept of begin, end, rbegin, rend, I
thought a good idea would be to have the python class define a __iter__
method and use that method with stl_input_iterator to pull in the complete
list from python to use with begin, end, etc.

What I have currently is a python converter for tuples of size 2 to std
pairs, then the wrapper class for the interface calls
> struct IFooHolder_wrapper : IFooHolder, bp::wrapper< IFooHolder >
> {
> mutable std::map< std::string, boost::shared_ptr< IFoo > > _map;
> const_iterator begin() const
> {
>    std::copy( bp::stl_input_iterator< std::pair< std::string,
boost::shared_ptr< IFoo > > >( bp::object( bp::handle<>( bp::borrowed(
bp::detail::wrapper_base_::get_owner(*this) ) ) ) ),
bp::stl_input_iterator<      std::pair< std::string, boost::shared_ptr< IFoo
> > >(), std::inserter( _map, _map.begin() ) );
>    return _map.begin();
> }

to copy all the iterations into a map the wrapper holds.  Then the begin,
end, rbegin, rend methods return iterations from that map.

My main concern is firstly if there is a better way.  More elegant solutions
like the map indexing suite do not really work on this problem, and google
can only help so much.  I fear there is not much documentation from others
trying to do the same thing.
I especially do not like "bp::object( bp::handle<>( bp::borrowed(
bp::detail::wrapper_base_::get_owner(*this) ) ) )"
>From my understanding of stl_input_iterator, it needs a bp::object that has
an "__iter__" attribute.  I want to use this->get_override( "__iter__" ) but
that does not work.

So if someone knows a better way to build a bp::object from a bp::wrapper
class that would be a good start.  I should mention my solution compiles but
does not work.  Python crashes when it tries to do the std::copy

Thanks for any input
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20101103/408e19c5/attachment.html>

From cyberaishu at gmail.com  Fri Nov  5 10:35:07 2010
From: cyberaishu at gmail.com (Aishwarya Venkataraman)
Date: Fri, 5 Nov 2010 02:35:07 -0700
Subject: [C++-sig] Unable to find -lboost_python
Message-ID: <AANLkTi=YKBH=0PBz6TqhEv_BEX0EPgj4rfzgwynUAJif@mail.gmail.com>

Hello,

I want to install pydoop to get python talk to hadoop/ hdfs file system.
This requires me to have boost installed. I have installed boost using
macports.

python setup.py build_ext -L/opt/local/lib/ -I/opt/local/include/

When I run the above command to install pydoop, it gives me the following
error:

g++ -L/Applications/djangostack-1.2.3-0/common/lib -bundle -undefined
dynamic_lookup build/temp.macosx-10.4-i386-2.6/src/pipes.o
build/temp.macosx-10.4-i386-2.6/src/pipes_context.o
build/temp.macosx-10.4-i386-2.6/src/pipes_test_support.o
build/temp.macosx-10.4-i386-2.6/src/pipes_serial_utils.o
build/temp.macosx-10.4-i386-2.6/src/exceptions.o
build/temp.macosx-10.4-i386-2.6/src/_pipes_main.o
build/temp.macosx-10.4-i386-2.6/src/SerialUtils.o
build/temp.macosx-10.4-i386-2.6/src/HadoopPipes.o
build/temp.macosx-10.4-i386-2.6/src/StringUtils.o -L/opt/local/lib/
-lpthread *-lboost_python* -o
build/lib.macosx-10.4-i386-2.6/pydoop/_pipes.so
*ld: library not found for -lboost_python*

Where exactly is lboost_python located on a mac? How do I find this library
?

Thanks,
Aishwarya
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20101105/6d02ee0e/attachment.html>

From berthet at hi.is  Fri Nov  5 11:48:51 2010
From: berthet at hi.is (Jean Claude Berthet)
Date: Fri, 05 Nov 2010 10:48:51 +0000
Subject: [C++-sig] Unable to find -lboost_python
In-Reply-To: <AANLkTi=YKBH=0PBz6TqhEv_BEX0EPgj4rfzgwynUAJif@mail.gmail.com>
References: <AANLkTi=YKBH=0PBz6TqhEv_BEX0EPgj4rfzgwynUAJif@mail.gmail.com>
Message-ID: <4CD3E113.2010606@hi.is>

By default it is in /usr/local/lib/ but you may have to ask for to be 
compiled explicitly:
./bjam --with-python

On 11/5/10 9:35 AM, Aishwarya Venkataraman wrote:
> Hello,
>
> I want to install pydoop to get python talk to hadoop/ hdfs file 
> system. This requires me to have boost installed. I have installed 
> boost using macports.
>
> python setup.py build_ext -L/opt/local/lib/ -I/opt/local/include/
>
> When I run the above command to install pydoop, it gives me the 
> following error:
>
> g++ -L/Applications/djangostack-1.2.3-0/common/lib -bundle -undefined 
> dynamic_lookup build/temp.macosx-10.4-i386-2.6/src/pipes.o 
> build/temp.macosx-10.4-i386-2.6/src/pipes_context.o 
> build/temp.macosx-10.4-i386-2.6/src/pipes_test_support.o 
> build/temp.macosx-10.4-i386-2.6/src/pipes_serial_utils.o 
> build/temp.macosx-10.4-i386-2.6/src/exceptions.o 
> build/temp.macosx-10.4-i386-2.6/src/_pipes_main.o 
> build/temp.macosx-10.4-i386-2.6/src/SerialUtils.o 
> build/temp.macosx-10.4-i386-2.6/src/HadoopPipes.o 
> build/temp.macosx-10.4-i386-2.6/src/StringUtils.o -L/opt/local/lib/ 
> -lpthread *-lboost_python* -o 
> build/lib.macosx-10.4-i386-2.6/pydoop/_pipes.so
> *ld: library not found for -lboost_python*
>
> Where exactly is lboost_python located on a mac? How do I find this 
> library ?
>
> Thanks,
> Aishwarya
>
>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20101105/33b07bc3/attachment.html>

From j.reid at mail.cryst.bbk.ac.uk  Fri Nov  5 15:42:16 2010
From: j.reid at mail.cryst.bbk.ac.uk (John Reid)
Date: Fri, 05 Nov 2010 14:42:16 +0000
Subject: [C++-sig] 2 "using python" declarations in user-config.jam not
	working
Message-ID: <ib1548$ra0$1@dough.gmane.org>

Hi,

I can't get boost.build configured for 2 different python versions, of 
which one is a debug build. I follow the instructions here:
http://www.boost.org/doc/libs/1_44_0/libs/python/doc/building.html

My user-config.jam has entries like this
# default python - this picks up a standard Ubuntu 2.6 install
using python ;

# debug python build - this picks up my debug build from source
using python : 2.7 : /home/john/local/python-dbg/bin/python2.7
         :
         :
         : <python-debugging>on
         ;


Unfortunately when I build my project
bjam -q -j3 debug

which is configured thus:
project Proj
	: requirements
		<variant>debug:<python-debugging>on
	: usage-requirements
	: default-build debug
	;

It uses a boost.python built against the default python. This 
libboost_python.so.1.44.0 has an undefined symbol 
PyUnicodeUCS4_AsUTF8String.

If I comment out the line in user-config.jam
using python ;

so that all my builds use the debug 2.7 python it works ok but 
presumably then I have a problem with release builds.

This looks like a bug to me. How am I supposed to configure this in 
user-config.jam?

Thanks,
John.



From j.reid at mail.cryst.bbk.ac.uk  Fri Nov  5 15:52:08 2010
From: j.reid at mail.cryst.bbk.ac.uk (John Reid)
Date: Fri, 05 Nov 2010 14:52:08 +0000
Subject: [C++-sig] 2 "using python" declarations in user-config.jam not
 working
In-Reply-To: <ib1548$ra0$1@dough.gmane.org>
References: <ib1548$ra0$1@dough.gmane.org>
Message-ID: <ib15mo$top$1@dough.gmane.org>

On 05/11/10 14:42, John Reid wrote:
> Hi,
>
> I can't get boost.build configured for 2 different python versions, of
> which one is a debug build. I follow the instructions here:
> http://www.boost.org/doc/libs/1_44_0/libs/python/doc/building.html
>
> My user-config.jam has entries like this
> # default python - this picks up a standard Ubuntu 2.6 install
> using python ;
>
> # debug python build - this picks up my debug build from source
> using python : 2.7 : /home/john/local/python-dbg/bin/python2.7
> :
> :
> : <python-debugging>on
> ;
>
>
> Unfortunately when I build my project
> bjam -q -j3 debug
>
> which is configured thus:
> project Proj
> : requirements
> <variant>debug:<python-debugging>on
> : usage-requirements
> : default-build debug
> ;
>
> It uses a boost.python built against the default python. This
> libboost_python.so.1.44.0 has an undefined symbol
> PyUnicodeUCS4_AsUTF8String.
>
> If I comment out the line in user-config.jam
> using python ;
>
> so that all my builds use the debug 2.7 python it works ok but
> presumably then I have a problem with release builds.
>
> This looks like a bug to me. How am I supposed to configure this in
> user-config.jam?

I have a work-around. If I always remember to put "python=2.7" for the 
debug builds as a bjam option it works ok. E.g.:
bjamc -q -j3 -d+2 debug python=2.7

>
> Thanks,
> John.


From Adrian.Baetu at plato-ag.com  Mon Nov  8 10:34:41 2010
From: Adrian.Baetu at plato-ag.com (AdrianB)
Date: Mon, 8 Nov 2010 09:34:41 +0000 (UTC)
Subject: [C++-sig] boost::python executes wrong implementation
References: <loom.20101102T172513-950@post.gmane.org>
Message-ID: <loom.20101108T103214-678@post.gmane.org>

HI,

still hopeing somone could comme up with anny idea for:
http://article.gmane.org/gmane.comp.python.c%2B%2B/15036

regards Adrian



From simwarg at gmail.com  Tue Nov  9 21:31:26 2010
From: simwarg at gmail.com (Simon W)
Date: Tue, 9 Nov 2010 21:31:26 +0100
Subject: [C++-sig] Expose c++ class as python module
Message-ID: <AANLkTinxruovzb2M6r5cpdcJbnCiR9y204_s+yv=0-3r@mail.gmail.com>

Hi,
I want to export my c++ class and make an instance of it and then expose it
to python as if it were a python module. Is it possible?

// export and create instance
class_<myClass> [...];

namespace["instance"] = ptr( instanceOfMyClass );


Then in python:

import instance

instance.someFunction()

Thank you! /Simon
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20101109/3ba0837f/attachment.html>

From talljimbo at gmail.com  Tue Nov  9 22:09:33 2010
From: talljimbo at gmail.com (Jim Bosch)
Date: Tue, 09 Nov 2010 13:09:33 -0800
Subject: [C++-sig] Expose c++ class as python module
In-Reply-To: <AANLkTinxruovzb2M6r5cpdcJbnCiR9y204_s+yv=0-3r@mail.gmail.com>
References: <AANLkTinxruovzb2M6r5cpdcJbnCiR9y204_s+yv=0-3r@mail.gmail.com>
Message-ID: <4CD9B88D.5070009@gmail.com>

On 11/09/2010 12:31 PM, Simon W wrote:
> Hi,
> I want to export my c++ class and make an instance of it and then expose
> it to python as if it were a python module. Is it possible?
>
> // export and create instance
> class_<myClass> [...];
>
> namespace["instance"] = ptr( instanceOfMyClass );
>
>
> Then in python:
>
> import instance
>
> instance.someFunction()
>

This is not possible to do directly, as boost::python::class_ makes a 
Python type object, and that's fundamentally different from a Python 
module object.

You can simulate this by doing tricks on the Python side, however:

  - Make a regular Boost.Python module.

  - In your module-that-mimics-a-class (pure Python), instantiate your 
instance in Python as a hidden module-level variable.

  - Extract bound methods from the instance and assign them to 
module-level variables.


Jim Bosch


From josh.davidson at lmco.com  Mon Nov 29 07:36:21 2010
From: josh.davidson at lmco.com (Davidson, Josh)
Date: Sun, 28 Nov 2010 23:36:21 -0700
Subject: [C++-sig] pygccxml and inner classes
Message-ID: <3602620DAE03994D80A0E6387C1A3B83B9F54870@HDXMSP8.us.lmco.com>

I'm wondering if this behavior is correct when using pygccxml.  Consider the following two sets of simple class definitions:

-------------- SET 1 ----------------------
class inner {
protected:
    unsigned a;
    unsigned b;
    unsigned c;
};

class Cool {
public:
    Cool(){}

protected:
    int x;
    int y;
    int z;

    inner i;
};

-------------- SET 2 ----------------------
class Cool {
public:
    Cool();

protected:
    class inner {
    protected:
        unsigned a;
        unsigned b;
        unsigned c;
    };

    int x;
    int y;
    int z;

    inner i;
};

-------------- END ----------------------

So you'll notice that the only difference between 1 & 2 is that 2, defines "inner" as an inner class.  Now, if I parse a header file containing those classes, obtain a reference to the pygccxml for class Cool and print it's variables like so:

for var in classType.variables():
  print "var", var.name, var.type

This is what I get:

SET 1
var x int
var y int
var z int
var i inner

SET 2
var x int
var y int
var z int
var i Cool::inner
var a unsigned int
var b unsigned int
var c unsigned int


As you can see, when I is an inner class, the inner class's members show up in the parent class in addition to the inner class itself.  Is this behavior correct?  In SET 2, I would not expect to see a, b, and c show up as members of class Cool.  If this behavior is correct, is there a way to filter out those members that belong to the inner class?



From roman.yakovenko at gmail.com  Mon Nov 29 21:22:39 2010
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Mon, 29 Nov 2010 22:22:39 +0200
Subject: [C++-sig] pygccxml and inner classes
In-Reply-To: <3602620DAE03994D80A0E6387C1A3B83B9F54870@HDXMSP8.us.lmco.com>
References: <3602620DAE03994D80A0E6387C1A3B83B9F54870@HDXMSP8.us.lmco.com>
Message-ID: <AANLkTik1wBe71xAXbRVdZDT0qAVZ_OA81sR9ibPaJLf0@mail.gmail.com>

On Mon, Nov 29, 2010 at 8:36 AM, Davidson, Josh <josh.davidson at lmco.com>wrote:

> I'm wondering if this behavior is correct when using pygccxml.  Consider
> the following two sets of simple class definitions:
>
> -------------- SET 1 ----------------------
> class inner {
> protected:
>    unsigned a;
>    unsigned b;
>    unsigned c;
> };
>
> class Cool {
> public:
>    Cool(){}
>
> protected:
>    int x;
>    int y;
>    int z;
>
>    inner i;
> };
>
> -------------- SET 2 ----------------------
> class Cool {
> public:
>    Cool();
>
> protected:
>    class inner {
>    protected:
>        unsigned a;
>        unsigned b;
>        unsigned c;
>    };
>
>    int x;
>    int y;
>    int z;
>
>    inner i;
> };
>
> -------------- END ----------------------
>
> So you'll notice that the only difference between 1 & 2 is that 2, defines
> "inner" as an inner class.  Now, if I parse a header file containing those
> classes, obtain a reference to the pygccxml for class Cool and print it's
> variables like so:
>
> for var in classType.variables():
>  print "var", var.name, var.type
>
> This is what I get:
>
> SET 1
> var x int
> var y int
> var z int
> var i inner
>
> SET 2
> var x int
> var y int
> var z int
> var i Cool::inner
> var a unsigned int
> var b unsigned int
> var c unsigned int
>
>
> As you can see, when I is an inner class, the inner class's members show up
> in the parent class in addition to the inner class itself.  Is this behavior
> correct?  In SET 2, I would not expect to see a, b, and c show up as members
> of class Cool.  If this behavior is correct, is there a way to filter out
> those members that belong to the inner class?
>
>
This is the current package behavior. Another interesting use case, you can
bring, is based classes. You will not see the variables from the base
classes.

Take a look on this document:
http://pygccxml.svn.sourceforge.net/viewvc/pygccxml/pygccxml_dev/docs/query_interface.rest?revision=1727&view=markup
and
then on pygccxml/declarations/scopedef.py - variables method.


In your case, it is very simple to get the desired variables:

class_vars = classType.variables( lambda var: var.parent is classType )


HTH
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20101129/8093e44f/attachment.html>