From dave at boostpro.com  Mon Dec  1 18:50:58 2008
From: dave at boostpro.com (David Abrahams)
Date: Mon, 01 Dec 2008 12:50:58 -0500
Subject: [C++-sig] shared_ptr with custom destructor
In-Reply-To: <95291a80811241326g25a0a4f6tc819acce47928816@mail.gmail.com>
	(Renato Araujo's message of "Mon, 24 Nov 2008 18:26:09 -0300")
References: <95291a80811241326g25a0a4f6tc819acce47928816@mail.gmail.com>
Message-ID: <87hc5n7pj1.fsf@mcbain.luannocracy.com>


on Mon Nov 24 2008, "Renato Araujo" <renatox-AT-gmail.com> wrote:

> Hi all,
> I would like to specify during the class_ declaration use a shared_ptr
> with custom release function.
>
> shared_ptr has a constructor where you can pass a function to call in
> the release, I would like to use this in a generic form.
>
> I already tried some like that:
>
> shared_ptr<BaseClass> _custom_constructor()
> {
>       return shared_ptr<BaseClass>(new BaseClass(), _release_func);
> }
>
> class_ <BaseClass, shared_ptr<BaseClass> >
> ("BaseClass").def("__init__", make_constructor(_custom_constructor));
>
> but this not work to class derived from BaseClass: my

Derived in C++ or in Python?

> _custom_constructor is not called to others class.
>
> I would like something more generic where I can declare in class_
> template, something like that:
>
> class_<MyClass, boost::shared_ptr<MyClass, release_func> ("MyObject")
> .......
>
> There is something like that? How is the best way to do this?

Write a function template that you can use with make_constructor, e.g.:

      make_constructor(&_custom_constructor<DerivedClass>)

HTH,

-- 
Dave Abrahams
BoostPro Computing
http://www.boostpro.com

From alanbaljeu at yahoo.com  Mon Dec  1 22:29:49 2008
From: alanbaljeu at yahoo.com (Alan Baljeu)
Date: Mon, 1 Dec 2008 13:29:49 -0800 (PST)
Subject: [C++-sig] gccxml
Message-ID: <384576.93432.qm@web58307.mail.re3.yahoo.com>

I'm lost getting the right gccxml for use with pygccxml.  And of course that would be used in Pybindgen and/or py++.

The links point to a gccxml page which advertises version 0.6, but docs talk about versions 0.7 and 0.9.  Anyhow, I downloaded 0.6 and didn't have much success.  But then I'm not sure what I'm supposed to do to get a proper scan.  I also don't get where I would find 0.7 or 0.9 versions.  Ideally I'd just like a documented package that gives me everything together that I can test out.  

Please advise.

 Alan Baljeu
http://www.collaborative-systems.org
Intelligent software that works _with_ you.



      __________________________________________________________________
Get the name you've always wanted @ymail.com or @rocketmail.com! Go to http://ca.promos.yahoo.com/jacko/

From roman.yakovenko at gmail.com  Tue Dec  2 08:37:46 2008
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 2 Dec 2008 09:37:46 +0200
Subject: [C++-sig] gccxml
In-Reply-To: <384576.93432.qm@web58307.mail.re3.yahoo.com>
References: <384576.93432.qm@web58307.mail.re3.yahoo.com>
Message-ID: <7465b6170812012337gd1896ddyf47dcb956bfc5c02@mail.gmail.com>

On Mon, Dec 1, 2008 at 11:29 PM, Alan Baljeu <alanbaljeu at yahoo.com> wrote:
> I'm lost getting the right gccxml for use with pygccxml.  And of course that would be used in Pybindgen and/or py++.
>
> The links point to a gccxml page which advertises version 0.6, but docs talk about versions 0.7 and 0.9.  Anyhow, I downloaded 0.6 and didn't have much success.  But then I'm not sure what I'm supposed to do to get a proper scan.  I also don't get where I would find 0.7 or 0.9 versions.  Ideally I'd just like a documented package that gives me everything together that I can test out.

Unfortunately there is no GUI to install gccxml.
http://gccxml.org/HTML/Install.html contains set of simple
instructions to install latest GCCXML. It takes only few minutes to
install it.
If you use Ubuntu Linux, that you can install it using Synaptic.

HTH

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

From macke at yar.nu  Tue Dec  2 09:46:21 2008
From: macke at yar.nu (Marcus Lindblom)
Date: Tue, 02 Dec 2008 09:46:21 +0100
Subject: [C++-sig] [py++] Problem: empty xml file created
Message-ID: <4934F5DD.2060102@yar.nu>

Hi all,

I'm trying to get OpenSG's python-bindings to work on Windows (and 
contruibuting that back to the project), and trying to track down an 
exception that get's thrown when running it's gen_bindings.py. (see 
attached log.txt for output)

The source for gen_bindings.py is here, for reference:

https://realityforge.vrsource.org/trac/pyopensg/browser/branches/trunk_update/src/gen_bindings.py


 From what I can see, the exception is thrown, as can be seen, in 
read_gccxml_file in pygccxml/parser/source_reader.py. The generation of 
the temp xml file (self.create_xml_file( ffname )) creates an empty 
file, which then gets a syntax error when read back.

As I'm rather new to py++ (being old on c++ and average on python) I 
don't really know where to look when hunting for this bug.

The PyOpenSG-devs have only used linux, and have maybe seen this before, 
earlier, but can't recall what and when. They also have used a earlier 
py++-version, but I'm using 1.0. I might be shooting myself in the foot 
with that, but I hope not.

All help gratefully recieved. :)

Cheers,
/Marcus
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: log.txt
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081202/a507eb36/attachment.txt>

From roman.yakovenko at gmail.com  Tue Dec  2 10:08:34 2008
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 2 Dec 2008 11:08:34 +0200
Subject: [C++-sig] [py++] Problem: empty xml file created
In-Reply-To: <4934F5DD.2060102@yar.nu>
References: <4934F5DD.2060102@yar.nu>
Message-ID: <7465b6170812020108x3520eedeq37a802152802c156@mail.gmail.com>

2008/12/2 Marcus Lindblom <macke at yar.nu>:
> Hi all,
>
> I'm trying to get OpenSG's python-bindings to work on Windows (and
> contruibuting that back to the project), and trying to track down an
> exception that get's thrown when running it's gen_bindings.py. (see attached
> log.txt for output)
>
> The source for gen_bindings.py is here, for reference:
>
> https://realityforge.vrsource.org/trac/pyopensg/browser/branches/trunk_update/src/gen_bindings.py
>
>
> From what I can see, the exception is thrown, as can be seen, in
> read_gccxml_file in pygccxml/parser/source_reader.py. The generation of the
> temp xml file (self.create_xml_file( ffname )) creates an empty file, which
> then gets a syntax error when read back.
>
> As I'm rather new to py++ (being old on c++ and average on python) I don't
> really know where to look when hunting for this bug.
>
> The PyOpenSG-devs have only used linux, and have maybe seen this before,
> earlier, but can't recall what and when. They also have used a earlier
> py++-version, but I'm using 1.0. I might be shooting myself in the foot with
> that, but I hope not.

There were few changes here and there, but it should not be too
dificcult to upgrade,

The output contains the following command:

""C:\program files\bin\gccxml.exe"
/IJ:\3rd\OpenSG\opensg2_trunk\build.win32.cl.9.0\instlinks\include
/IC:\Program Files oost oost_1_36_0 -DBOOST_PYTHON_MAX_ARITY=21
-I"." -I"osg_module"
-I"J:\3rd\OpenSG\opensg2_trunk\build.win32.cl.9.0\instlinks\include"
-I"J:\3rd\OpenSG\opensg2_trunk\build.win32.cl.9.0\instlinks\include\OpenSG"
-I"j:\3rd\boost\boost_1_36_0" -I"c:\dev\python26\include"
"osg_module\osgmodule_mainheader.h"
-fxml="c:\users\marcusl\appdata\local\temp\tmp1e92ik.xml""

Can you run the command, from cmd and check that file was created and
contains data.

If I remember right, PyOpenSG project uses some tricky configuration
to suppress GCCXML warnings and\or errors.

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

From macke at yar.nu  Tue Dec  2 14:15:38 2008
From: macke at yar.nu (Marcus Lindblom)
Date: Tue, 02 Dec 2008 14:15:38 +0100
Subject: [C++-sig] [py++] Problem: empty xml file created
In-Reply-To: <7465b6170812020108x3520eedeq37a802152802c156@mail.gmail.com>
References: <4934F5DD.2060102@yar.nu>
	<7465b6170812020108x3520eedeq37a802152802c156@mail.gmail.com>
Message-ID: <493534FA.1020703@yar.nu>

Roman Yakovenko wrote:
> 2008/12/2 Marcus Lindblom <macke at yar.nu>:
>> Hi all,
>>
>> I'm trying to get OpenSG's python-bindings to work on Windows (and
>> contruibuting that back to the project), and trying to track down an
>> exception that get's thrown when running it's gen_bindings.py. (see attached
>> log.txt for output)

[snip]

>> The PyOpenSG-devs have only used linux, and have maybe seen this before,
>> earlier, but can't recall what and when. They also have used a earlier
>> py++-version, but I'm using 1.0. I might be shooting myself in the foot with
>> that, but I hope not.
> 
> There were few changes here and there, but it should not be too
> dificcult to upgrade,

That's what I'm hoping for. :)

> The output contains the following command:
> 
> ""C:\program files\bin\gccxml.exe"
> /IJ:\3rd\OpenSG\opensg2_trunk\build.win32.cl.9.0\instlinks\include
> /IC:\Program Files oost oost_1_36_0 -DBOOST_PYTHON_MAX_ARITY=21
> -I"." -I"osg_module"
> -I"J:\3rd\OpenSG\opensg2_trunk\build.win32.cl.9.0\instlinks\include"
> -I"J:\3rd\OpenSG\opensg2_trunk\build.win32.cl.9.0\instlinks\include\OpenSG"
> -I"j:\3rd\boost\boost_1_36_0" -I"c:\dev\python26\include"
> "osg_module\osgmodule_mainheader.h"
> -fxml="c:\users\marcusl\appdata\local\temp\tmp1e92ik.xml""
> 
> Can you run the command, from cmd and check that file was created and
> contains data.

Ok. It came right back at me:

pyopensg/src> ** the above **
Could not determine compiler setting.
Could not determine GCCXML_FLAGS setting.
pyopensg/src>

So, I changed it to run my compiled gccxml (pretty recent cvs head) and 
added "--gccxml-compiler cl" (I use VS9 and python 2.6) and fixed the 
include paths (pyopensg retrieves includepaths from the osg2-config 
python script, which used /I instead of /I and also had some non-escaped 
backslashes (turing 'c:\program files\boost' into 'c:\program files 
oost' as can be seen above :)

Running the modified cmd-line made gccxml run and print a whole lot, 
starting with:

j:/3rd/OpenSG/opensg2_trunk/build.win32.cl.9.0/instlinks/include/OpenSG/OSGConfig.h:135:2: 
error: #error Endian determination : could not guess your plattform

and ending up with an ICE.

It seems as I have to add a few defines and such to get the 
opensg-sources to identify the compiler correctly.

I can handle the osg2-config script myself, and probably figure out what 
defines are necessary, but where do I set compiler-arg + defines that 
are used when doing the cache-step that calls the above command?

Is this cache stuff in PyOpenSG or Py++? (because I can't seem to find 
any such call in gen_bindings.main() nor nearby related modules, unless 
I've missed something).

Cheers,
/Marcus

From kmihail at rocketmail.com  Tue Dec  2 14:33:14 2008
From: kmihail at rocketmail.com (Mihail Konstantinov)
Date: Tue, 2 Dec 2008 13:33:14 +0000 (GMT)
Subject: [C++-sig] boost.python: private copy constructor problem
Message-ID: <262531.34155.qm@web28402.mail.ukl.yahoo.com>

Hello,
I am still in the early process of learning boost.python.
I have reduced my problem to the following code:

#include <boost/python.hpp>
using namespace boost::python;
class A{
private:
  A(const A&){}; //no public copy constructor
};
class B: public A{
public:
  B(){};
};
BOOST_PYTHON_MODULE(boost_ext)
{
    class_<A>("A",init<>());
    class_<B, bases<A> >("B",init<>());
}

When running bjam on this example, I get
"boost.cpp:6: error: ?A::A(const A&)? is private"
as the first error among a long list of hints from the compiler.

How can I expose the classes with their constructors to python?

Thank you
Mihail



      


From roman.yakovenko at gmail.com  Tue Dec  2 14:28:05 2008
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 2 Dec 2008 15:28:05 +0200
Subject: [C++-sig] [py++] Problem: empty xml file created
In-Reply-To: <493534FA.1020703@yar.nu>
References: <4934F5DD.2060102@yar.nu>
	<7465b6170812020108x3520eedeq37a802152802c156@mail.gmail.com>
	<493534FA.1020703@yar.nu>
Message-ID: <7465b6170812020528y2834955bp3a09b8c0a18026cc@mail.gmail.com>

On Tue, Dec 2, 2008 at 3:15 PM, Marcus Lindblom <macke at yar.nu> wrote:
> Ok. It came right back at me:
>
> pyopensg/src> ** the above **
> Could not determine compiler setting.
> Could not determine GCCXML_FLAGS setting.
> pyopensg/src>
>
> So, I changed it to run my compiled gccxml (pretty recent cvs head) and
> added "--gccxml-compiler cl" (I use VS9 and python 2.6) and fixed the
> include paths (pyopensg retrieves includepaths from the osg2-config python
> script, which used /I instead of /I and also had some non-escaped
> backslashes (turing 'c:\program files\boost' into 'c:\program files oost' as
> can be seen above :)
>
> Running the modified cmd-line made gccxml run and print a whole lot,
> starting with:
>
> j:/3rd/OpenSG/opensg2_trunk/build.win32.cl.9.0/instlinks/include/OpenSG/OSGConfig.h:135:2:
> error: #error Endian determination : could not guess your plattform
>
> and ending up with an ICE.
>
> It seems as I have to add a few defines and such to get the opensg-sources
> to identify the compiler correctly.
>
> I can handle the osg2-config script myself, and probably figure out what
> defines are necessary, but where do I set compiler-arg + defines that are
> used when doing the cache-step that calls the above command?

gen_bindings.py file contains the following lines:

813    mb = ModuleBuilder([main_header_filename],
814 	                      working_directory = ".",
815 	                      include_paths=inc_path,
816 	                      cache=cache_file+".module",
817 	                      define_symbols=defines,
818 	                      ignore_gccxml_output=True,
819 	                      optimize_queries=use_query_opt,
820 	                      #start_with_declarations=["OSG",],
821 	                      cflags=opensg_cflags + " " + boost_cflags)

opensg_cflags, boost_cflags variables contains the configuration flags

> Is this cache stuff in PyOpenSG or Py++? (because I can't seem to find any
> such call in gen_bindings.main() nor nearby related modules, unless I've
> missed something).

I don't understand the question, sorry :-(



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

From kmihail at rocketmail.com  Tue Dec  2 14:39:19 2008
From: kmihail at rocketmail.com (Mihail Konstantinov)
Date: Tue, 2 Dec 2008 13:39:19 +0000 (GMT)
Subject: [C++-sig] boost.python: private copy constructor problem
References: <262531.34155.qm@web28402.mail.ukl.yahoo.com>
Message-ID: <534641.33799.qm@web28407.mail.ukl.yahoo.com>

Sorry, the public A::A() constructor was missing. The corrected class A, which caused the described problem, is:

class A{
private:
  A(const A&){}; //no public copy constructor
public:
  A(){};
};



      


From seefeld at sympatico.ca  Tue Dec  2 14:43:12 2008
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Tue, 02 Dec 2008 08:43:12 -0500
Subject: [C++-sig] boost.python: private copy constructor problem
In-Reply-To: <262531.34155.qm@web28402.mail.ukl.yahoo.com>
References: <262531.34155.qm@web28402.mail.ukl.yahoo.com>
Message-ID: <49353B70.1070103@sympatico.ca>

Mihail Konstantinov wrote:
> Hello,
> I am still in the early process of learning boost.python.
> I have reduced my problem to the following code:
>
> #include <boost/python.hpp>
> using namespace boost::python;
> class A{
> private:
>   A(const A&){}; //no public copy constructor
> };
> class B: public A{
> public:
>   B(){};
> };
> BOOST_PYTHON_MODULE(boost_ext)
> {
>     class_<A>("A",init<>());
>     class_<B, bases<A> >("B",init<>());
> }
>
> When running bjam on this example, I get
> "boost.cpp:6: error: ?A::A(const A&)? is private"
> as the first error among a long list of hints from the compiler.
>   


In this case you want to tell Python that your object is non-copyable:

class_<A, noncopyable> a("A", init<>);
...

See 
http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/class.html#class_-spec 
for details.

HTH,
       Stefan

-- 

      ...ich hab' noch einen Koffer in Berlin...


From macke at yar.nu  Tue Dec  2 15:01:25 2008
From: macke at yar.nu (Marcus Lindblom)
Date: Tue, 02 Dec 2008 15:01:25 +0100
Subject: [C++-sig] [py++] Problem: empty xml file created
In-Reply-To: <7465b6170812020528y2834955bp3a09b8c0a18026cc@mail.gmail.com>
References: <4934F5DD.2060102@yar.nu>	<7465b6170812020108x3520eedeq37a802152802c156@mail.gmail.com>	<493534FA.1020703@yar.nu>
	<7465b6170812020528y2834955bp3a09b8c0a18026cc@mail.gmail.com>
Message-ID: <49353FB5.20809@yar.nu>

Roman Yakovenko wrote:
> On Tue, Dec 2, 2008 at 3:15 PM, Marcus Lindblom <macke at yar.nu> wrote:
>> Ok. It came right back at me:
>>
>> pyopensg/src> ** the above **
>> Could not determine compiler setting.
>> Could not determine GCCXML_FLAGS setting.
>> pyopensg/src>
>>
>> So, I changed it to run my compiled gccxml (pretty recent cvs head) and
>> added "--gccxml-compiler cl" (I use VS9 and python 2.6) and fixed the
>> include paths (pyopensg retrieves includepaths from the osg2-config python
>> script, which used /I instead of /I and also had some non-escaped
>> backslashes (turing 'c:\program files\boost' into 'c:\program files oost' as
>> can be seen above :)
>>
>> Running the modified cmd-line made gccxml run and print a whole lot,
>> starting with:
>>
>> j:/3rd/OpenSG/opensg2_trunk/build.win32.cl.9.0/instlinks/include/OpenSG/OSGConfig.h:135:2:
>> error: #error Endian determination : could not guess your plattform
>>
>> and ending up with an ICE.
>>
>> It seems as I have to add a few defines and such to get the opensg-sources
>> to identify the compiler correctly.
>>
>> I can handle the osg2-config script myself, and probably figure out what
>> defines are necessary, but where do I set compiler-arg + defines that are
>> used when doing the cache-step that calls the above command?
> 
> gen_bindings.py file contains the following lines:
> 
> 813    mb = ModuleBuilder([main_header_filename],
> 814 	                      working_directory = ".",
> 815 	                      include_paths=inc_path,
> 816 	                      cache=cache_file+".module",
> 817 	                      define_symbols=defines,
> 818 	                      ignore_gccxml_output=True,
> 819 	                      optimize_queries=use_query_opt,
> 820 	                      #start_with_declarations=["OSG",],
> 821 	                      cflags=opensg_cflags + " " + boost_cflags)
> 
> opensg_cflags, boost_cflags variables contains the configuration flags

Yes, but IIUC, this is way after the call to gccxml that I ran above, 
since the follwing is printed after the 'INFO gccxml cmd: ...' in the 
log-file in my first post:

566 print "Auto-building list of node cores using .fcd files... "

>> Is this cache stuff in PyOpenSG or Py++? (because I can't seem to find any
>> such call in gen_bindings.main() nor nearby related modules, unless I've
>> missed something).
> 
> I don't understand the question, sorry :-(

No worries. :)

The call to gccxml that you asked me to run does not produce a valid 
ouput file, because it's needs fixing.

So, I'm guessing a bit here, but since the log contains stuff about 
cache file load failure and cache creation, which does the gccxml-call 
that I need to augment with defines and --gccxml-compiler option, I 
assumed I need to fix the gccxml call in the cache-creation thingy, not 
the call to ModuleBuilder way below. (which tries to use the empty file 
that gccxml failed to create earlier.)

So, my question is, where does that gccxml call come from, and how do I 
change it's parameters? (Is even a py++ thing, or does it come from a 
pyopensg somewhere I've missed?)

(I don't think the gccxml-call happens because of the call to the 
ModuleBuilder at line 813, although I might be wrong?)

Does that make sense?

I'll tinker with the ModuleBuilder a bit anyway.. to see if it helps, 
but the order of output makes me think the problem is somewhere else.

Cheers,
/Marcus

P.S. Thanks for helping out! I'm ping-pong:ing a bit between the 
opensg-list and this one, so if I need to address my questions the other 
way, let me know.

From kmihail at rocketmail.com  Tue Dec  2 15:03:02 2008
From: kmihail at rocketmail.com (Mihail Konstantinov)
Date: Tue, 2 Dec 2008 14:03:02 +0000 (GMT)
Subject: [C++-sig] boost.python: private copy constructor problem
References: <262531.34155.qm@web28402.mail.ukl.yahoo.com>
	<49353B70.1070103@sympatico.ca>
Message-ID: <993348.632.qm@web28403.mail.ukl.yahoo.com>

> Stefan Seefeld wrote:

> 
> In this case you want to tell Python that your object is non-copyable:

I followed your suggestion and still get the same error message:
boost.cpp:6: error: ?A::A(const A&)? is private
(full bjam output is appended to this email)

Could you check this modified code? boost_ext.cpp:
#include <boost/python.hpp>
using namespace boost::python;

class A{
private:
  A(const A&){}; //no public copy constructor
public:
  A(){};
};

class B: public A{
public:
  B(){};
};

BOOST_PYTHON_MODULE(boost_ext)
{
    class_<A,boost::noncopyable> a("A",init<>());
    class_<B, bases<A> >("B",init<>());
}

Thank you
Mihail

Compilation output:

$bjam boost_ext
...patience...
...found 1202 targets...
...updating 2 targets...
gcc.compile.c++ bin/gcc-4.1.2/debug/boost.o
boost.cpp: In copy constructor ?B::B(const B&)?:
boost.cpp:11:   instantiated from ?boost::python::objects::value_holder<Value>::value_holder(PyObject*, A0) [with A0 = boost::reference_wrapper<const B>, Value = B]?
/home/mihail/temporary/boost_1_37_0/boost/python/object/make_instance.hpp:68:   instantiated from ?static Holder* boost::python::objects::make_instance<T, Holder>::construct(void*, PyObject*, boost::reference_wrapper<const T>) [with T = B, Holder = boost::python::objects::value_holder<B>]?
/home/mihail/temporary/boost_1_37_0/boost/python/object/make_instance.hpp:42:   instantiated from ?static PyObject* boost::python::objects::make_instance_impl<T, Holder, Derived>::execute(Arg&) [with Arg = const boost::reference_wrapper<const B>, T = B, Holder = boost::python::objects::value_holder<B>, Derived = boost::python::objects::make_instance<B, boost::python::objects::value_holder<B> >]?
/home/mihail/temporary/boost_1_37_0/boost/python/object/class_wrapper.hpp:29:   instantiated from ?static PyObject* boost::python::objects::class_cref_wrapper<Src, MakeInstance>::convert(const Src&) [with Src = B, MakeInstance = boost::python::objects::make_instance<B, boost::python::objects::value_holder<B> >]?
/home/mihail/temporary/boost_1_37_0/boost/python/converter/as_to_python_function.hpp:27:   instantiated from ?static PyObject* boost::python::converter::as_to_python_function<T, ToPython>::convert(const void*) [with T = B, ToPython = boost::python::objects::class_cref_wrapper<B, boost::python::objects::make_instance<B, boost::python::objects::value_holder<B> > >]?
/home/mihail/temporary/boost_1_37_0/boost/python/to_python_converter.hpp:87:   instantiated from ?boost::python::to_python_converter<T, Conversion, has_get_pytype>::to_python_converter() [with T = B, Conversion = boost::python::objects::class_cref_wrapper<B, boost::python::objects::make_instance<B, boost::python::objects::value_holder<B> > >, bool has_get_pytype = true]?
/home/mihail/temporary/boost_1_37_0/boost/python/object/class_wrapper.hpp:26:   instantiated from ?static void boost::python::objects::class_metadata<T, X1, X2, X3>::maybe_register_class_to_python(T2*, mpl_::false_) [with T2 = B, T = B, X1 = boost::python::bases<A, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_>, X2 = boost::python::detail::not_specified, X3 = boost::python::detail::not_specified]?
/home/mihail/temporary/boost_1_37_0/boost/python/object/class_metadata.hpp:229:   instantiated from ?static void boost::python::objects::class_metadata<T, X1, X2, X3>::register_aux2(T2*, Callback) [with T2 = B, Callback = boost::integral_constant<bool, false>, T = B, X1 = boost::python::bases<A, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_>, X2 = boost::python::detail::not_specified, X3 = boost::python::detail::not_specified]?
/home/mihail/temporary/boost_1_37_0/boost/python/object/class_metadata.hpp:219:   instantiated from ?static void boost::python::objects::class_metadata<T, X1, X2, X3>::register_aux(void*) [with T = B, X1 = boost::python::bases<A, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_>, X2 = boost::python::detail::not_specified, X3 = boost::python::detail::not_specified]?
/home/mihail/temporary/boost_1_37_0/boost/python/object/class_metadata.hpp:205:   instantiated from ?static void boost::python::objects::class_metadata<T, X1, X2, X3>::register_() [with T = B, X1 = boost::python::bases<A, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_>, X2 = boost::python::detail::not_specified, X3 = boost::python::detail::not_specified]?
/home/mihail/temporary/boost_1_37_0/boost/python/class.hpp:496:   instantiated from ?void boost::python::class_<T, X1, X2, X3>::initialize(const DefVisitor&) [with DefVisitor = boost::python::init_base<boost::python::init<mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_> >, W = B, X1 = boost::python::bases<A, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_>, X2 = boost::python::detail::not_specified, X3 = boost::python::detail::not_specified]?
/home/mihail/temporary/boost_1_37_0/boost/python/class.hpp:208:   instantiated from ?boost::python::class_<T, X1, X2, X3>::class_(const char*, const boost::python::init_base<DerivedT>&) [with DerivedT = boost::python::init<mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_>, W = B, X1 = boost::python::bases<A, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_>, X2 = boost::python::detail::not_specified, X3 = boost::python::detail::not_specified]?
boost.cpp:19:   instantiated from here
boost.cpp:6: error: ?A::A(const A&)? is private
boost.cpp:11: error: within this context
/home/mihail/temporary/boost_1_37_0/boost/python/object/value_holder.hpp: In constructor ?boost::python::objects::value_holder<Value>::value_holder(PyObject*, A0) [with A0 = boost::reference_wrapper<const B>, Value = B]?:
/home/mihail/temporary/boost_1_37_0/boost/python/object/value_holder.hpp:134: note: synthesized method ?B::B(const B&)? first required here 

    "g++"  -ftemplate-depth-128 -O0 -fno-inline -Wall -g -fPIC    -I"/home/mihail/temporary/boost_1_37_0" -I"/usr/include/python2.5" -c -o "bin/gcc-4.1.2/debug/boost.o" "boost.cpp"

...failed gcc.compile.c++ bin/gcc-4.1.2/debug/boost.o....
...skipped <pbin/gcc-4.1.2/debug>boost_ext.so for lack of <pbin/gcc-4.1..2/debug>boost.o...
...failed updating 1 target...
...skipped 1 target....



      


From renatox at gmail.com  Tue Dec  2 15:11:09 2008
From: renatox at gmail.com (Renato Araujo)
Date: Tue, 2 Dec 2008 11:11:09 -0300
Subject: [C++-sig] boost.python: private copy constructor problem
In-Reply-To: <993348.632.qm@web28403.mail.ukl.yahoo.com>
References: <262531.34155.qm@web28402.mail.ukl.yahoo.com>
	<49353B70.1070103@sympatico.ca>
	<993348.632.qm@web28403.mail.ukl.yahoo.com>
Message-ID: <95291a80812020611g43190de5n47f0bd84c1c086a1@mail.gmail.com>

try use boost::noncopyable to B or implemente a copy constructor in B
without call A copy contructor.

BR



On Tue, Dec 2, 2008 at 11:03 AM, Mihail Konstantinov
<kmihail at rocketmail.com> wrote:
>> Stefan Seefeld wrote:
>
>>
>> In this case you want to tell Python that your object is non-copyable:
>
> I followed your suggestion and still get the same error message:
> boost.cpp:6: error: 'A::A(const A&)' is private
> (full bjam output is appended to this email)
>
> Could you check this modified code? boost_ext.cpp:
> #include <boost/python.hpp>
> using namespace boost::python;
>
> class A{
> private:
>  A(const A&){}; //no public copy constructor
> public:
>  A(){};
> };
>
> class B: public A{
> public:
>  B(){};
> };
>
> BOOST_PYTHON_MODULE(boost_ext)
> {
>    class_<A,boost::noncopyable> a("A",init<>());
>    class_<B, bases<A> >("B",init<>());
> }
>
> Thank you
> Mihail
>
> Compilation output:
>
> $bjam boost_ext
> ...patience...
> ...found 1202 targets...
> ...updating 2 targets...
> gcc.compile.c++ bin/gcc-4.1.2/debug/boost.o
> boost.cpp: In copy constructor 'B::B(const B&)':
> boost.cpp:11:   instantiated from 'boost::python::objects::value_holder<Value>::value_holder(PyObject*, A0) [with A0 = boost::reference_wrapper<const B>, Value = B]'
> /home/mihail/temporary/boost_1_37_0/boost/python/object/make_instance.hpp:68:   instantiated from 'static Holder* boost::python::objects::make_instance<T, Holder>::construct(void*, PyObject*, boost::reference_wrapper<const T>) [with T = B, Holder = boost::python::objects::value_holder<B>]'
> /home/mihail/temporary/boost_1_37_0/boost/python/object/make_instance.hpp:42:   instantiated from 'static PyObject* boost::python::objects::make_instance_impl<T, Holder, Derived>::execute(Arg&) [with Arg = const boost::reference_wrapper<const B>, T = B, Holder = boost::python::objects::value_holder<B>, Derived = boost::python::objects::make_instance<B, boost::python::objects::value_holder<B> >]'
> /home/mihail/temporary/boost_1_37_0/boost/python/object/class_wrapper.hpp:29:   instantiated from 'static PyObject* boost::python::objects::class_cref_wrapper<Src, MakeInstance>::convert(const Src&) [with Src = B, MakeInstance = boost::python::objects::make_instance<B, boost::python::objects::value_holder<B> >]'
> /home/mihail/temporary/boost_1_37_0/boost/python/converter/as_to_python_function.hpp:27:   instantiated from 'static PyObject* boost::python::converter::as_to_python_function<T, ToPython>::convert(const void*) [with T = B, ToPython = boost::python::objects::class_cref_wrapper<B, boost::python::objects::make_instance<B, boost::python::objects::value_holder<B> > >]'
> /home/mihail/temporary/boost_1_37_0/boost/python/to_python_converter.hpp:87:   instantiated from 'boost::python::to_python_converter<T, Conversion, has_get_pytype>::to_python_converter() [with T = B, Conversion = boost::python::objects::class_cref_wrapper<B, boost::python::objects::make_instance<B, boost::python::objects::value_holder<B> > >, bool has_get_pytype = true]'
> /home/mihail/temporary/boost_1_37_0/boost/python/object/class_wrapper.hpp:26:   instantiated from 'static void boost::python::objects::class_metadata<T, X1, X2, X3>::maybe_register_class_to_python(T2*, mpl_::false_) [with T2 = B, T = B, X1 = boost::python::bases<A, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_>, X2 = boost::python::detail::not_specified, X3 = boost::python::detail::not_specified]'
> /home/mihail/temporary/boost_1_37_0/boost/python/object/class_metadata.hpp:229:   instantiated from 'static void boost::python::objects::class_metadata<T, X1, X2, X3>::register_aux2(T2*, Callback) [with T2 = B, Callback = boost::integral_constant<bool, false>, T = B, X1 = boost::python::bases<A, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_>, X2 = boost::python::detail::not_specified, X3 = boost::python::detail::not_specified]'
> /home/mihail/temporary/boost_1_37_0/boost/python/object/class_metadata.hpp:219:   instantiated from 'static void boost::python::objects::class_metadata<T, X1, X2, X3>::register_aux(void*) [with T = B, X1 = boost::python::bases<A, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_>, X2 = boost::python::detail::not_specified, X3 = boost::python::detail::not_specified]'
> /home/mihail/temporary/boost_1_37_0/boost/python/object/class_metadata.hpp:205:   instantiated from 'static void boost::python::objects::class_metadata<T, X1, X2, X3>::register_() [with T = B, X1 = boost::python::bases<A, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_>, X2 = boost::python::detail::not_specified, X3 = boost::python::detail::not_specified]'
> /home/mihail/temporary/boost_1_37_0/boost/python/class.hpp:496:   instantiated from 'void boost::python::class_<T, X1, X2, X3>::initialize(const DefVisitor&) [with DefVisitor = boost::python::init_base<boost::python::init<mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_> >, W = B, X1 = boost::python::bases<A, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_>, X2 = boost::python::detail::not_specified, X3 = boost::python::detail::not_specified]'
> /home/mihail/temporary/boost_1_37_0/boost/python/class.hpp:208:   instantiated from 'boost::python::class_<T, X1, X2, X3>::class_(const char*, const boost::python::init_base<DerivedT>&) [with DerivedT = boost::python::init<mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_>, W = B, X1 = boost::python::bases<A, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_, mpl_::void_>, X2 = boost::python::detail::not_specified, X3 = boost::python::detail::not_specified]'
> boost.cpp:19:   instantiated from here
> boost.cpp:6: error: 'A::A(const A&)' is private
> boost.cpp:11: error: within this context
> /home/mihail/temporary/boost_1_37_0/boost/python/object/value_holder.hpp: In constructor 'boost::python::objects::value_holder<Value>::value_holder(PyObject*, A0) [with A0 = boost::reference_wrapper<const B>, Value = B]':
> /home/mihail/temporary/boost_1_37_0/boost/python/object/value_holder.hpp:134: note: synthesized method 'B::B(const B&)' first required here
>
>    "g++"  -ftemplate-depth-128 -O0 -fno-inline -Wall -g -fPIC    -I"/home/mihail/temporary/boost_1_37_0" -I"/usr/include/python2.5" -c -o "bin/gcc-4.1.2/debug/boost.o" "boost.cpp"
>
> ...failed gcc.compile.c++ bin/gcc-4.1.2/debug/boost.o....
> ...skipped <pbin/gcc-4.1.2/debug>boost_ext.so for lack of <pbin/gcc-4.1..2/debug>boost.o...
> ...failed updating 1 target...
> ...skipped 1 target....
>
>
>
>
>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig



-- 
Renato Araujo Oliveira Filho

From macke at yar.nu  Tue Dec  2 15:12:12 2008
From: macke at yar.nu (Marcus Lindblom)
Date: Tue, 02 Dec 2008 15:12:12 +0100
Subject: [C++-sig] [py++] Problem: empty xml file created
In-Reply-To: <49353FB5.20809@yar.nu>
References: <4934F5DD.2060102@yar.nu>	<7465b6170812020108x3520eedeq37a802152802c156@mail.gmail.com>	<493534FA.1020703@yar.nu>	<7465b6170812020528y2834955bp3a09b8c0a18026cc@mail.gmail.com>
	<49353FB5.20809@yar.nu>
Message-ID: <4935423C.4030203@yar.nu>

Marcus Lindblom wrote:

> (I don't think the gccxml-call happens because of the call to the 
> ModuleBuilder at line 813, although I might be wrong?)

Waitaminute. When I'm rerunning it now, the order makes more sense, and 
the cache-calls come up after the calls to ModuleBuilder. Probably the 
log-file merging of stdout and stderr was playing tricks on me.

I'll look augmenting the ModuleBuilder call, just as you said. Sorry for 
the confusion. :-/

cheers
/Marcus

From macke at yar.nu  Tue Dec  2 15:15:23 2008
From: macke at yar.nu (Marcus Lindblom)
Date: Tue, 02 Dec 2008 15:15:23 +0100
Subject: [C++-sig] boost.python: private copy constructor problem
In-Reply-To: <993348.632.qm@web28403.mail.ukl.yahoo.com>
References: <262531.34155.qm@web28402.mail.ukl.yahoo.com>	<49353B70.1070103@sympatico.ca>
	<993348.632.qm@web28403.mail.ukl.yahoo.com>
Message-ID: <493542FB.2090200@yar.nu>

Mihail Konstantinov wrote:
>> Stefan Seefeld wrote:
> 
>> In this case you want to tell Python that your object is non-copyable:
> 
> I followed your suggestion and still get the same error message:

[snip]

> /home/mihail/temporary/boost_1_37_0/boost/python/object/value_holder.hpp: In constructor ?boost::python::objects::value_holder<Value>::value_holder(PyObject*, A0) [with A0 = boost::reference_wrapper<const B>, Value = B]?:
> /home/mihail/temporary/boost_1_37_0/boost/python/object/value_holder.hpp:134: note: synthesized method ?B::B(const B&)? first required here 

Seems like B needs to be noncopyable too?

/Marcus

From roman.yakovenko at gmail.com  Tue Dec  2 15:16:36 2008
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 2 Dec 2008 16:16:36 +0200
Subject: [C++-sig] boost.python: private copy constructor problem
In-Reply-To: <95291a80812020611g43190de5n47f0bd84c1c086a1@mail.gmail.com>
References: <262531.34155.qm@web28402.mail.ukl.yahoo.com>
	<49353B70.1070103@sympatico.ca>
	<993348.632.qm@web28403.mail.ukl.yahoo.com>
	<95291a80812020611g43190de5n47f0bd84c1c086a1@mail.gmail.com>
Message-ID: <7465b6170812020616r85fa951pe421b7a5bfdeaa33@mail.gmail.com>

On Tue, Dec 2, 2008 at 4:11 PM, Renato Araujo <renatox at gmail.com> wrote:
> try use boost::noncopyable to B or implemente a copy constructor in B
> without call A copy contructor.

He is right. The following code was generated by Py++:

#include "boost/python.hpp"

namespace bp = boost::python;

BOOST_PYTHON_MODULE(pyplusplus){
    bp::class_< A, boost::noncopyable >( "A", bp::init< >() );

    bp::class_< B, bp::bases< A >, boost::noncopyable >( "B", bp::init< >() );
}

Py++ has some GUI too (
http://language-binding.net/pyplusplus/documentation/tutorials/pyplusplus_gui.html
) - no need to learn API
-- 
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/

From seefeld at sympatico.ca  Tue Dec  2 15:19:03 2008
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Tue, 02 Dec 2008 09:19:03 -0500
Subject: [C++-sig] boost.python: private copy constructor problem
In-Reply-To: <993348.632.qm@web28403.mail.ukl.yahoo.com>
References: <262531.34155.qm@web28402.mail.ukl.yahoo.com>	<49353B70.1070103@sympatico.ca>
	<993348.632.qm@web28403.mail.ukl.yahoo.com>
Message-ID: <493543D7.90904@sympatico.ca>

Mihail Konstantinov wrote:
>
> gcc.compile.c++ bin/gcc-4.1.2/debug/boost.o
> boost.cpp: In copy constructor ?B::B(const B&)?:
>   

As this error message suggests: as B derives from A, it is (by default) 
non-copyable, too. That you have to tell python, too, thus:

class_<A, noncopyable> a("A", init<>);
class_<A, bases<A>, noncopyable> b("B", init<>);
...

HTH,
       Stefan
 

-- 

      ...ich hab' noch einen Koffer in Berlin...


From kmihail at rocketmail.com  Tue Dec  2 15:25:47 2008
From: kmihail at rocketmail.com (Mihail Konstantinov)
Date: Tue, 2 Dec 2008 14:25:47 +0000 (GMT)
Subject: [C++-sig] boost.python: private copy constructor problem
References: <262531.34155.qm@web28402.mail.ukl.yahoo.com>
	<49353B70.1070103@sympatico.ca>
	<993348.632.qm@web28403.mail.ukl.yahoo.com>
	<95291a80812020611g43190de5n47f0bd84c1c086a1@mail.gmail.com>
Message-ID: <68970.45747.qm@web28404.mail.ukl.yahoo.com>

> try use boost::noncopyable to B or implemente a copy constructor in B

> without call A copy contructor.

Thank you. This works:
BOOST_PYTHON_MODULE(boost_ext)
{
    class_<A, boost::noncopyable> a("A",init<>());
    class_<B, bases<A>, boost::noncopyable >("B",init<>());
}

One more question about the syntax. You cited http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/class.html#class_-spec which specifies the syntax as "class_<T,Bases,HeldType,NonCopyable>" so that I thought I have to include all 4 template arguments. But when I used instead of the above (working!) solution:
    class_<A, bases<>, A, boost::noncopyable> a("A",init<>());
    class_<B, bases<A>, B, boost::noncopyable >("B",init<>());
I get the error:
>/home/mihail/temporary/boost_1_37_0/boost/python/object/value_holder.hpp:159: error: no matching function for call to
>?B::B(PyObject*&)?
>boost.cpp:13: note: candidates are: B::B()
>boost.cpp:11: note:                 B::B(const B&)

Why does the abbreviated template argument list work, and why not the complete argument list?

Thank you
Mihail



      


From seefeld at sympatico.ca  Tue Dec  2 15:33:27 2008
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Tue, 02 Dec 2008 09:33:27 -0500
Subject: [C++-sig] boost.python: private copy constructor problem
In-Reply-To: <68970.45747.qm@web28404.mail.ukl.yahoo.com>
References: <262531.34155.qm@web28402.mail.ukl.yahoo.com>	<49353B70.1070103@sympatico.ca>	<993348.632.qm@web28403.mail.ukl.yahoo.com>	<95291a80812020611g43190de5n47f0bd84c1c086a1@mail.gmail.com>
	<68970.45747.qm@web28404.mail.ukl.yahoo.com>
Message-ID: <49354737.4040703@sympatico.ca>

Mihail Konstantinov wrote:
> One more question about the syntax. You cited 
> http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/class.html#class_-spec 
> which specifies the syntax as "class_<T,Bases,HeldType,NonCopyable>" 
> so that I thought I have to include all 4 template arguments. But when 
> I used instead of the above (working!) solution:
>     class_<A, bases<>, A, boost::noncopyable> a("A",init<>());
>     class_<B, bases<A>, B, boost::noncopyable >("B",init<>());
> I get the error:
>   
>> /home/mihail/temporary/boost_1_37_0/boost/python/object/value_holder.hpp:159: error: no matching function for call to
>> ?B::B(PyObject*&)?
>> boost.cpp:13: note: candidates are: B::B()
>> boost.cpp:11: note:                 B::B(const B&)
>>     
>
> Why does the abbreviated template argument list work, and why not the complete argument list?
>   

As the docs say, all but the first template argument are optional. And, 
thanks to some meta-programming magic, their order doesn't matter.
The above link also discusses requirements on the HeldType, such as "its 
exposed constructor(s) must accept an initial |PyObject*| argument..."

Again, may not actually need to care about the HeldType if the defaults 
work fine for you.

HTH,
       Stefan

-- 

      ...ich hab' noch einen Koffer in Berlin...


From alanbaljeu at yahoo.com  Tue Dec  2 16:26:21 2008
From: alanbaljeu at yahoo.com (Alan Baljeu)
Date: Tue, 2 Dec 2008 07:26:21 -0800 (PST)
Subject: [C++-sig] gccxml
References: <384576.93432.qm@web58307.mail.re3.yahoo.com>
	<7465b6170812012337gd1896ddyf47dcb956bfc5c02@mail.gmail.com>
Message-ID: <255255.48164.qm@web58306.mail.re3.yahoo.com>

This is getting out of hand :-(

All I wanted to do was parse some simple C++ function declarations to generate a corresponding py-callable wrapper.  To do this I need to get pygccxml.  Then I need gccxml to use pygccxml.  To get that I need WinCVS or some equivalent (which version, and learn how CVS works) or Cygwin.  Then I need to choose the preferred version of gccxml.  And then I need to obtain CMake to install it?

I think I'd rather just do my own thing, except that C++ is so warty it's difficult to come up with anything reasonable that will serve the purpose.

Alan Baljeu



----- Original Message ----
From: Roman Yakovenko <roman.yakovenko at gmail.com>
To: Development of Python/C++ integration <cplusplus-sig at python.org>
Sent: Tuesday, December 2, 2008 2:37:46 AM
Subject: Re: [C++-sig] gccxml

On Mon, Dec 1, 2008 at 11:29 PM, Alan Baljeu <alanbaljeu at yahoo.com> wrote:
> I'm lost getting the right gccxml for use with pygccxml.  And of course that would be used in Pybindgen and/or py++.
>
> The links point to a gccxml page which advertises version 0.6, but docs talk about versions 0.7 and 0.9.  Anyhow, I downloaded 0.6 and didn't have much success.  But then I'm not sure what I'm supposed to do to get a proper scan.  I also don't get where I would find 0.7 or 0.9 versions.  Ideally I'd just like a documented package that gives me everything together that I can test out.

Unfortunately there is no GUI to install gccxml.
http://gccxml.org/HTML/Install.html contains set of simple
instructions to install latest GCCXML. It takes only few minutes to
install it.
If you use Ubuntu Linux, that you can install it using Synaptic.

HTH

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



      __________________________________________________________________
Instant Messaging, free SMS, sharing photos and more... Try the new Yahoo! Canada Messenger at http://ca.beta.messenger.yahoo.com/

From seefeld at sympatico.ca  Tue Dec  2 16:41:45 2008
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Tue, 02 Dec 2008 10:41:45 -0500
Subject: [C++-sig] gccxml
In-Reply-To: <255255.48164.qm@web58306.mail.re3.yahoo.com>
References: <384576.93432.qm@web58307.mail.re3.yahoo.com>	<7465b6170812012337gd1896ddyf47dcb956bfc5c02@mail.gmail.com>
	<255255.48164.qm@web58306.mail.re3.yahoo.com>
Message-ID: <49355739.40106@sympatico.ca>

Alan Baljeu wrote:
> This is getting out of hand :-(
>   

> I think I'd rather just do my own thing, except that C++ is so warty it's difficult to come up with anything reasonable that will serve the purpose.
>   

How many functions and objects do you want to export, anyway ? I have so 
far always preferred to write my boost.python code manually. Not only 
because installing the prerequisites for the alternate tools is so 
cumbersome, but mainly because I want the python API to look a little 
more pythonic than the equivalent C++ code I'm wrapping.

You can always move 'back' to auto-generated wrapper code later if you 
find it is worth it.

Regards,
       Stefan

-- 

      ...ich hab' noch einen Koffer in Berlin...


From roman.yakovenko at gmail.com  Tue Dec  2 16:58:33 2008
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 2 Dec 2008 17:58:33 +0200
Subject: [C++-sig] gccxml
In-Reply-To: <255255.48164.qm@web58306.mail.re3.yahoo.com>
References: <384576.93432.qm@web58307.mail.re3.yahoo.com>
	<7465b6170812012337gd1896ddyf47dcb956bfc5c02@mail.gmail.com>
	<255255.48164.qm@web58306.mail.re3.yahoo.com>
Message-ID: <7465b6170812020758r1d85a7f6oc13bb06b118b47c3@mail.gmail.com>

On Tue, Dec 2, 2008 at 5:26 PM, Alan Baljeu <alanbaljeu at yahoo.com> wrote:
> This is getting out of hand :-(
>
> All I wanted to do was parse some simple C++ function declarations to generate a corresponding py-callable wrapper.  To do this I need to get pygccxml.  Then I need gccxml to use pygccxml.  To get that I need WinCVS or some equivalent (which version, and learn how CVS works) or Cygwin.  Then I need to choose the preferred version of gccxml.  And then I need to obtain CMake to install it?
>
> I think I'd rather just do my own thing, except that C++ is so warty it's difficult to come up with anything reasonable that will serve the purpose.

How much effort did you put to improve the situation? Did you
contribute to GCCXML project in someway or another?

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

From alanbaljeu at yahoo.com  Tue Dec  2 17:14:34 2008
From: alanbaljeu at yahoo.com (Alan Baljeu)
Date: Tue, 2 Dec 2008 08:14:34 -0800 (PST)
Subject: [C++-sig] gccxml
References: <384576.93432.qm@web58307.mail.re3.yahoo.com>
	<7465b6170812012337gd1896ddyf47dcb956bfc5c02@mail.gmail.com>
	<255255.48164.qm@web58306.mail.re3.yahoo.com>
	<7465b6170812020758r1d85a7f6oc13bb06b118b47c3@mail.gmail.com>
Message-ID: <136637.36945.qm@web58304.mail.re3.yahoo.com>

I would make a contribution to some projects.  GCC-XML is not one of them, seeing how it's not even worth my effort to try to install and configure this stuff.  If I were involved with these packages, I would make it easy for others to get involved.  As it isn't easy I'm going to look elsewhere.

 Alan Baljeu




----- Original Message ----
From: Roman Yakovenko <roman.yakovenko at gmail.com>
To: Development of Python/C++ integration <cplusplus-sig at python.org>
Sent: Tuesday, December 2, 2008 10:58:33 AM
Subject: Re: [C++-sig] gccxml

On Tue, Dec 2, 2008 at 5:26 PM, Alan Baljeu <alanbaljeu at yahoo.com> wrote:
> This is getting out of hand :-(
>
> All I wanted to do was parse some simple C++ function declarations to generate a corresponding py-callable wrapper.  To do this I need to get pygccxml.  Then I need gccxml to use pygccxml.  To get that I need WinCVS or some equivalent (which version, and learn how CVS works) or Cygwin.  Then I need to choose the preferred version of gccxml.  And then I need to obtain CMake to install it?
>
> I think I'd rather just do my own thing, except that C++ is so warty it's difficult to come up with anything reasonable that will serve the purpose.

How much effort did you put to improve the situation? Did you
contribute to GCCXML project in someway or another?

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



      __________________________________________________________________
Yahoo! Canada Toolbar: Search from anywhere on the web, and bookmark your favourite sites. Download it now at
http://ca.toolbar.yahoo.com.

From gjcarneiro at gmail.com  Tue Dec  2 17:47:42 2008
From: gjcarneiro at gmail.com (Gustavo Carneiro)
Date: Tue, 2 Dec 2008 16:47:42 +0000
Subject: [C++-sig] gccxml
In-Reply-To: <136637.36945.qm@web58304.mail.re3.yahoo.com>
References: <384576.93432.qm@web58307.mail.re3.yahoo.com>
	<7465b6170812012337gd1896ddyf47dcb956bfc5c02@mail.gmail.com>
	<255255.48164.qm@web58306.mail.re3.yahoo.com>
	<7465b6170812020758r1d85a7f6oc13bb06b118b47c3@mail.gmail.com>
	<136637.36945.qm@web58304.mail.re3.yahoo.com>
Message-ID: <a467ca4f0812020847k58b7e1f5v1c0a4998f7ccd4f3@mail.gmail.com>

2008/12/2 Alan Baljeu <alanbaljeu at yahoo.com>

> I would make a contribution to some projects.  GCC-XML is not one of them,
> seeing how it's not even worth my effort to try to install and configure
> this stuff.  If I were involved with these packages, I would make it easy
> for others to get involved.  As it isn't easy I'm going to look elsewhere.


Unfortunately you will soon find out there is no real alternative to GCC-XML
if you have to do serious parsing of C++ headers.  The time invested in
getting GCC-XML to work for you will pay off in the end in case your project
has multi-year scope and large API set to wrap.  If on the other hand your
project is small, you can get by if you just manually write your wrappers
(both boost.python and pybindgen support this manual wrapping method).

Either way, know that wrapping C/C++ for Python is usually hard for
beginners, but don't give up, it gets easier with practice :-)

-- 
Gustavo J. A. M. Carneiro
INESC Porto, Telecommunications and Multimedia Unit
"The universe is always one step beyond logic." -- Frank Herbert
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081202/b5980542/attachment.htm>

From kmihail at rocketmail.com  Tue Dec  2 18:04:58 2008
From: kmihail at rocketmail.com (Mihail Konstantinov)
Date: Tue, 2 Dec 2008 17:04:58 +0000 (GMT)
Subject: [C++-sig] boost.python: implicit type conversion
Message-ID: <220240.21313.qm@web28405.mail.ukl.yahoo.com>

Dear all,
thanks a lot for your immediate responses that lead to a fast solution for my previous problem.

Here comes the next one.

In the example below both classes A and B are derived from std::string. Class B can be implicitly converted to A by means of 'B::operator const A()'. So the function printme(const A&) can also be called with an argument of class B (see main function).

My aim is to be able to call in python:
>>>import boost_ext
>>>b=boost_ext.B("hello world")
>>>boost_ext.printme(b)
but it produces the error:
> <class 'Boost.Python.ArgumentError'>: Python argument types in  boost_ext.printme(B)
> did not match C++ signature:
>    printme(A)

Can I modify the declaration inside BOOST_PYTHON_MODULE so that the implicit conversion is recognized and the above call boost_ext.printme(b) succeeds?

Thank you
Mihail

BTW, in the code below I'm using a preprocessor definition __BJAM__ which I had to define explicitly in the Jamroot. Is there some variable defined by default which I can use to distinguish between bjam builds and other (testing) calls to g++?

boost.cpp:
#ifdef __BJAM__ // I define __BJAM__ in the Jamroot file..
  #include <boost/python.hpp>
#endif
#include <string>
#include <iostream>

class A: public std::string{
public:
  A(const std::string&s,int i):std::string(s){}
};

class B: public std::string{
public:
  B(const std::string&s):std::string(s){}
  operator const A(){return A(*this,-1);}
};

void printme(const A &a){std::cout<<a<<std::endl;}

int main(){
  B b("hello world");
  printme(b);
}

#ifdef __BJAM__
BOOST_PYTHON_MODULE(boost_ext)
{
    using namespace boost::python;
    class_<std::string>("string");
    class_<A,bases<std::string> > ("A",init<const std::string&,int>());
    class_<B,bases<std::string> > ("B",init<const std::string&>());
    def("printme",&printme);
}
#endif



      


From mathieu.malaterre at gmail.com  Tue Dec  2 18:37:18 2008
From: mathieu.malaterre at gmail.com (Mathieu Malaterre)
Date: Tue, 2 Dec 2008 18:37:18 +0100
Subject: [C++-sig] gccxml
In-Reply-To: <a467ca4f0812020847k58b7e1f5v1c0a4998f7ccd4f3@mail.gmail.com>
References: <384576.93432.qm@web58307.mail.re3.yahoo.com>
	<7465b6170812012337gd1896ddyf47dcb956bfc5c02@mail.gmail.com>
	<255255.48164.qm@web58306.mail.re3.yahoo.com>
	<7465b6170812020758r1d85a7f6oc13bb06b118b47c3@mail.gmail.com>
	<136637.36945.qm@web58304.mail.re3.yahoo.com>
	<a467ca4f0812020847k58b7e1f5v1c0a4998f7ccd4f3@mail.gmail.com>
Message-ID: <bf0c3b3f0812020937nd5b5497xbc761a780afda23b@mail.gmail.com>

2008/12/2 Gustavo Carneiro <gjcarneiro at gmail.com>:
> 2008/12/2 Alan Baljeu <alanbaljeu at yahoo.com>
>>
>> I would make a contribution to some projects.  GCC-XML is not one of them,
>> seeing how it's not even worth my effort to try to install and configure
>> this stuff.  If I were involved with these packages, I would make it easy
>> for others to get involved.  As it isn't easy I'm going to look elsewhere.
>
> Unfortunately you will soon find out there is no real alternative to GCC-XML
> if you have to do serious parsing of C++ headers.  The time invested in
> getting GCC-XML to work for you will pay off in the end in case your project
> has multi-year scope and large API set to wrap.

Switch to a debian based system:

 $ sudo apt-get install gccxml

Approx ~1min of time invested (+ time to write this email).

-- 
Mathieu

From gjcarneiro at gmail.com  Tue Dec  2 19:06:23 2008
From: gjcarneiro at gmail.com (Gustavo Carneiro)
Date: Tue, 2 Dec 2008 18:06:23 +0000
Subject: [C++-sig] gccxml
In-Reply-To: <bf0c3b3f0812020937nd5b5497xbc761a780afda23b@mail.gmail.com>
References: <384576.93432.qm@web58307.mail.re3.yahoo.com>
	<7465b6170812012337gd1896ddyf47dcb956bfc5c02@mail.gmail.com>
	<255255.48164.qm@web58306.mail.re3.yahoo.com>
	<7465b6170812020758r1d85a7f6oc13bb06b118b47c3@mail.gmail.com>
	<136637.36945.qm@web58304.mail.re3.yahoo.com>
	<a467ca4f0812020847k58b7e1f5v1c0a4998f7ccd4f3@mail.gmail.com>
	<bf0c3b3f0812020937nd5b5497xbc761a780afda23b@mail.gmail.com>
Message-ID: <a467ca4f0812021006x327260e2s3c3bd0d9465e9cc2@mail.gmail.com>

2008/12/2 Mathieu Malaterre <mathieu.malaterre at gmail.com>

> 2008/12/2 Gustavo Carneiro <gjcarneiro at gmail.com>:
> > 2008/12/2 Alan Baljeu <alanbaljeu at yahoo.com>
> >>
> >> I would make a contribution to some projects.  GCC-XML is not one of
> them,
> >> seeing how it's not even worth my effort to try to install and configure
> >> this stuff.  If I were involved with these packages, I would make it
> easy
> >> for others to get involved.  As it isn't easy I'm going to look
> elsewhere.
> >
> > Unfortunately you will soon find out there is no real alternative to
> GCC-XML
> > if you have to do serious parsing of C++ headers.  The time invested in
> > getting GCC-XML to work for you will pay off in the end in case your
> project
> > has multi-year scope and large API set to wrap.
>
> Switch to a debian based system:
>
>  $ sudo apt-get install gccxml
>
> Approx ~1min of time invested (+ time to write this email).


Sure, but how current is this debian gccxml?  Is it a CVS snapshot, or the
years old official release that does not work well with pygccxml?

-- 
Gustavo J. A. M. Carneiro
INESC Porto, Telecommunications and Multimedia Unit
"The universe is always one step beyond logic." -- Frank Herbert
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081202/8e3e9eb8/attachment-0001.htm>

From mathieu.malaterre at gmail.com  Tue Dec  2 19:17:42 2008
From: mathieu.malaterre at gmail.com (Mathieu Malaterre)
Date: Tue, 2 Dec 2008 19:17:42 +0100
Subject: [C++-sig] gccxml
In-Reply-To: <a467ca4f0812021006x327260e2s3c3bd0d9465e9cc2@mail.gmail.com>
References: <384576.93432.qm@web58307.mail.re3.yahoo.com>
	<7465b6170812012337gd1896ddyf47dcb956bfc5c02@mail.gmail.com>
	<255255.48164.qm@web58306.mail.re3.yahoo.com>
	<7465b6170812020758r1d85a7f6oc13bb06b118b47c3@mail.gmail.com>
	<136637.36945.qm@web58304.mail.re3.yahoo.com>
	<a467ca4f0812020847k58b7e1f5v1c0a4998f7ccd4f3@mail.gmail.com>
	<bf0c3b3f0812020937nd5b5497xbc761a780afda23b@mail.gmail.com>
	<a467ca4f0812021006x327260e2s3c3bd0d9465e9cc2@mail.gmail.com>
Message-ID: <bf0c3b3f0812021017x33590952n8a3c80ed1d6f3092@mail.gmail.com>

2008/12/2 Gustavo Carneiro <gjcarneiro at gmail.com>:
>
>
> 2008/12/2 Mathieu Malaterre <mathieu.malaterre at gmail.com>
>>
>> 2008/12/2 Gustavo Carneiro <gjcarneiro at gmail.com>:
>> > 2008/12/2 Alan Baljeu <alanbaljeu at yahoo.com>
>> >>
>> >> I would make a contribution to some projects.  GCC-XML is not one of
>> >> them,
>> >> seeing how it's not even worth my effort to try to install and
>> >> configure
>> >> this stuff.  If I were involved with these packages, I would make it
>> >> easy
>> >> for others to get involved.  As it isn't easy I'm going to look
>> >> elsewhere.
>> >
>> > Unfortunately you will soon find out there is no real alternative to
>> > GCC-XML
>> > if you have to do serious parsing of C++ headers.  The time invested in
>> > getting GCC-XML to work for you will pay off in the end in case your
>> > project
>> > has multi-year scope and large API set to wrap.
>>
>> Switch to a debian based system:
>>
>>  $ sudo apt-get install gccxml
>>
>> Approx ~1min of time invested (+ time to write this email).
>
> Sure, but how current is this debian gccxml?  Is it a CVS snapshot, or the
> years old official release that does not work well with pygccxml?
>

From:
http://packages.debian.org/lenny/gccxml

...
Package: gccxml (0.9.0+cvs20080525-1)
...

I do not know if you call that /recent/ or not...

-- 
Mathieu

From gjcarneiro at gmail.com  Tue Dec  2 19:36:57 2008
From: gjcarneiro at gmail.com (Gustavo Carneiro)
Date: Tue, 2 Dec 2008 18:36:57 +0000
Subject: [C++-sig] gccxml
In-Reply-To: <bf0c3b3f0812021017x33590952n8a3c80ed1d6f3092@mail.gmail.com>
References: <384576.93432.qm@web58307.mail.re3.yahoo.com>
	<7465b6170812012337gd1896ddyf47dcb956bfc5c02@mail.gmail.com>
	<255255.48164.qm@web58306.mail.re3.yahoo.com>
	<7465b6170812020758r1d85a7f6oc13bb06b118b47c3@mail.gmail.com>
	<136637.36945.qm@web58304.mail.re3.yahoo.com>
	<a467ca4f0812020847k58b7e1f5v1c0a4998f7ccd4f3@mail.gmail.com>
	<bf0c3b3f0812020937nd5b5497xbc761a780afda23b@mail.gmail.com>
	<a467ca4f0812021006x327260e2s3c3bd0d9465e9cc2@mail.gmail.com>
	<bf0c3b3f0812021017x33590952n8a3c80ed1d6f3092@mail.gmail.com>
Message-ID: <a467ca4f0812021036g281652b9p50fd6c7fda866f21@mail.gmail.com>

2008/12/2 Mathieu Malaterre <mathieu.malaterre at gmail.com>

> 2008/12/2 Gustavo Carneiro <gjcarneiro at gmail.com>:
> >
> >
> > 2008/12/2 Mathieu Malaterre <mathieu.malaterre at gmail.com>
> >>
> >> 2008/12/2 Gustavo Carneiro <gjcarneiro at gmail.com>:
> >> > 2008/12/2 Alan Baljeu <alanbaljeu at yahoo.com>
> >> >>
> >> >> I would make a contribution to some projects.  GCC-XML is not one of
> >> >> them,
> >> >> seeing how it's not even worth my effort to try to install and
> >> >> configure
> >> >> this stuff.  If I were involved with these packages, I would make it
> >> >> easy
> >> >> for others to get involved.  As it isn't easy I'm going to look
> >> >> elsewhere.
> >> >
> >> > Unfortunately you will soon find out there is no real alternative to
> >> > GCC-XML
> >> > if you have to do serious parsing of C++ headers.  The time invested
> in
> >> > getting GCC-XML to work for you will pay off in the end in case your
> >> > project
> >> > has multi-year scope and large API set to wrap.
> >>
> >> Switch to a debian based system:
> >>
> >>  $ sudo apt-get install gccxml
> >>
> >> Approx ~1min of time invested (+ time to write this email).
> >
> > Sure, but how current is this debian gccxml?  Is it a CVS snapshot, or
> the
> > years old official release that does not work well with pygccxml?
> >
>
> From:
> http://packages.debian.org/lenny/gccxml
>
> ...
> Package: gccxml (0.9.0+cvs20080525-1)
> ...
>
> I do not know if you call that /recent/ or not...


Goodie!  That should work with pygccxml 0.9.5, in theory.  And to my
surprise Ubuntu 8.04 now has that same snapshot as well.

Kids these days have it too easy... :P

Unfortunately Debian/Ubuntu have no pygccxml package, what a shame! :-(

-- 
Gustavo J. A. M. Carneiro
INESC Porto, Telecommunications and Multimedia Unit
"The universe is always one step beyond logic." -- Frank Herbert
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081202/8fe0d9eb/attachment.htm>

From roman.yakovenko at gmail.com  Tue Dec  2 19:54:45 2008
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 2 Dec 2008 20:54:45 +0200
Subject: [C++-sig] boost.python: implicit type conversion
In-Reply-To: <220240.21313.qm@web28405.mail.ukl.yahoo.com>
References: <220240.21313.qm@web28405.mail.ukl.yahoo.com>
Message-ID: <7465b6170812021054g6099040j29d742e3b41ec266@mail.gmail.com>

On Tue, Dec 2, 2008 at 7:04 PM, Mihail Konstantinov
<kmihail at rocketmail.com> wrote:
> Dear all,
> thanks a lot for your immediate responses that lead to a fast solution for my previous problem.
>
> Here comes the next one.
>
> In the example below both classes A and B are derived from std::string.

It is not a good idea. Python string is immutable. May be you should
consider custom converter.
http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/faq.html#custom_string

>Class B can be implicitly converted to A by means of 'B::operator const A()'. So the function printme(const A&) can also be called with an argument of class B (see main function).
>
> My aim is to be able to call in python:
>>>>import boost_ext
>>>>b=boost_ext.B("hello world")
>>>>boost_ext.printme(b)
> but it produces the error:
>> <class 'Boost.Python.ArgumentError'>: Python argument types in  boost_ext.printme(B)
>> did not match C++ signature:
>>    printme(A)
>
> Can I modify the declaration inside BOOST_PYTHON_MODULE so that the implicit conversion is recognized and the above call boost_ext.printme(b) succeeds?

Yes: http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/implicit.html

> Thank you
> Mihail
>
> BTW, in the code below I'm using a preprocessor definition __BJAM__ which I had to define explicitly in the Jamroot. Is there some variable defined by default which I can use to distinguish between bjam builds and other (testing) calls to g++?

You can separate your source code to different files.


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

From kmihail at rocketmail.com  Tue Dec  2 20:20:44 2008
From: kmihail at rocketmail.com (Mihail Konstantinov)
Date: Tue, 2 Dec 2008 19:20:44 +0000 (GMT)
Subject: [C++-sig] boost.python: implicit type conversion
References: <220240.21313.qm@web28405.mail.ukl.yahoo.com>
	<7465b6170812021054g6099040j29d742e3b41ec266@mail.gmail.com>
Message-ID: <479726.59172.qm@web28413.mail.ukl.yahoo.com>

> Roman Yakovenko wrote:

> 
> It is not a good idea. Python string is immutable. May be you should
> consider custom converter.
> http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/faq.html#custom_string

I don't see why it is not a good idea. I am nowhere accessing any string derived classes in a mutable fashion, am I?
Also I have no option to change the class hierarchy. The code fragments I am posting are simplified from a large project that I try to pythonify, reduced to what I think is the cause of the problems. I want to avoid modifying the underlying project under all circumstances.

> > Can I modify the declaration inside BOOST_PYTHON_MODULE so that the implicit 
> conversion is recognized and the above call boost_ext.printme(b) succeeds?
> 
> Yes: http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/implicit.html

This looks very helpful. But I don't understand the C++ module definition example on that page. It ends with

>    implicitly_convertible<X,int>();
>    implicitly_convertible<int,X>();
>}
I see that an int can be implicitly converted to X. But how can an X be converted to an int? I can't believe that boost.python searches all implemented functions to find that make_x is capable of converting an int to X?

Thank you
Mihail


      


From kmihail at rocketmail.com  Tue Dec  2 20:31:02 2008
From: kmihail at rocketmail.com (Mihail Konstantinov)
Date: Tue, 2 Dec 2008 19:31:02 +0000 (GMT)
Subject: [C++-sig] boost.python: implicit type conversion
References: <220240.21313.qm@web28405.mail.ukl.yahoo.com>
	<7465b6170812021054g6099040j29d742e3b41ec266@mail.gmail.com>
	<479726.59172.qm@web28413.mail.ukl.yahoo.com>
Message-ID: <620254.92061.qm@web28411.mail.ukl.yahoo.com>

> >    implicitly_convertible();

> >    implicitly_convertible();
> >}
> I see that an int can be implicitly converted to X. But how can an X be 
> converted to an int? I can't believe that boost.python searches all implemented 
> functions to find that make_x is capable of converting an int to X?

Found it. X::operator int() const. Apparently I need some sleep.

Thank you
Mihail



      


From roman.yakovenko at gmail.com  Tue Dec  2 20:32:47 2008
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 2 Dec 2008 21:32:47 +0200
Subject: [C++-sig] boost.python: implicit type conversion
In-Reply-To: <479726.59172.qm@web28413.mail.ukl.yahoo.com>
References: <220240.21313.qm@web28405.mail.ukl.yahoo.com>
	<7465b6170812021054g6099040j29d742e3b41ec266@mail.gmail.com>
	<479726.59172.qm@web28413.mail.ukl.yahoo.com>
Message-ID: <7465b6170812021132p4bb988aaj4c52e0393b2ff0c@mail.gmail.com>

On Tue, Dec 2, 2008 at 9:20 PM, Mihail Konstantinov
<kmihail at rocketmail.com> wrote:
>> Roman Yakovenko wrote:
>> It is not a good idea. Python string is immutable. May be you should
>> consider custom converter.
>> http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/faq.html#custom_string
>
> I don't see why it is not a good idea. I am nowhere accessing any string derived classes in a mutable fashion, am I?

May in the code you posted you don't. May be you don't change it in
the whole project. But if you do, it breaks too many Python
assumptions.

> Also I have no option to change the class hierarchy. The code fragments I am posting are simplified from a large project that I try to pythonify, reduced to what I think is the cause of the problems. I want to avoid modifying the underlying project under all circumstances.

I understand. I suggest you to reconsider registration of std::string
as a base class of your classes.

Also, Boost.Python has built-in functionality to convert Python string
and std::basic_string. I don't know what effect class_< std::string >
registration will have.

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

From seefeld at sympatico.ca  Tue Dec  2 20:32:57 2008
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Tue, 02 Dec 2008 14:32:57 -0500
Subject: [C++-sig] boost.python: implicit type conversion
In-Reply-To: <479726.59172.qm@web28413.mail.ukl.yahoo.com>
References: <220240.21313.qm@web28405.mail.ukl.yahoo.com>	<7465b6170812021054g6099040j29d742e3b41ec266@mail.gmail.com>
	<479726.59172.qm@web28413.mail.ukl.yahoo.com>
Message-ID: <49358D69.4080000@sympatico.ca>

Mihail Konstantinov wrote:
>> Roman Yakovenko wrote:
>>     
>
>   
>> It is not a good idea. Python string is immutable. May be you should
>> consider custom converter.
>> http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/faq.html#custom_string
>>     
>
> I don't see why it is not a good idea. I am nowhere accessing any string derived classes in a mutable fashion, am I?
> Also I have no option to change the class hierarchy. The code fragments I am posting are simplified from a large project that I try to pythonify, reduced to what I think is the cause of the problems. I want to avoid modifying the underlying project under all circumstances.
>   

You are attempting to make Python aware of the inheritance relationship 
between std::string and your own class. That is what doesn't work, not 
your own deriving from std::string.
In order to make 'bases<std::string>' work, you would have to export the 
std::string class to Python first. But as that isn't what you actually 
want, you may as well remove the 'bases<std::string>' from your code, 
and instead rely on (implicit) conversion.


>   
>>> Can I modify the declaration inside BOOST_PYTHON_MODULE so that the implicit 
>>>       
>> conversion is recognized and the above call boost_ext.printme(b) succeeds?
>>
>> Yes: http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/implicit.html
>>     
>
> This looks very helpful. But I don't understand the C++ module definition example on that page. It ends with
>
>   
>>    implicitly_convertible<X,int>();
>>    implicitly_convertible<int,X>();
>> }
>>     
> I see that an int can be implicitly converted to X. But how can an X be converted to an int? I can't believe that boost.python searches all implemented functions to find that make_x is capable of converting an int to X?
>   

I believe it simply expects an 'operator int ()' to be available in that 
case.

Regards,
       Stefan

-- 

      ...ich hab' noch einen Koffer in Berlin...


From Matthew.Scouten at tradingtechnologies.com  Tue Dec  2 21:23:39 2008
From: Matthew.Scouten at tradingtechnologies.com (Matthew Scouten (TT))
Date: Tue, 2 Dec 2008 14:23:39 -0600
Subject: [C++-sig] Pickle an enum from c++
Message-ID: <32490DFF7774554A85D65D23A9F0F938099CB568@chiex01>

I have an enum in my c++ library. In fact I have a crap load of them.
169 to be exact. 

 

I need them to be pickle-able.  The enum_ object does not have a
.def_pickle function, so I cannot use the pickle suite. I tried to just
up and pickle it but that does not work either.

 

 enum box {cardboard, cereal, toy};

 

BOOST_PYTHON_MODULE(busybox)

{

 

    enum_<box>("box")

        .value("cardboard",cardboard )

        .value("cereal",cereal )

        .value("toy", toy);

 

}

 

C:\Documents and Settings\mscouten>cd C:\tt-dev\manticore\

 

C:\tt-dev\manticore>call setpath.bat

Python 2.5.1 (r251:54863, Apr 18 2007, 08:51:08) [MSC v.1310 32 bit
(Intel)] on win32

Type "help", "copyright", "credits" or "license" for more information.

>>> import pickle, busybox

>>> 

>>> b = busybox.box.toy

>>> 

>>> bs = pickle.dumps(b)

Traceback (most recent call last):

  File "<stdin>", line 1, in <module>

  File "C:\Python25\lib\pickle.py", line 1366, in dumps

    Pickler(file, protocol).dump(obj)

  File "C:\Python25\lib\pickle.py", line 224, in dump

    self.save(obj)

  File "C:\Python25\lib\pickle.py", line 331, in save

    self.save_reduce(obj=obj, *rv)

  File "C:\Python25\lib\pickle.py", line 401, in save_reduce

    save(args)

  File "C:\Python25\lib\pickle.py", line 286, in save

    f(self, obj) # Call unbound method with explicit self

  File "C:\Python25\lib\pickle.py", line 562, in save_tuple

    save(element)

  File "C:\Python25\lib\pickle.py", line 286, in save

    f(self, obj) # Call unbound method with explicit self

  File "C:\Python25\lib\pickle.py", line 748, in save_global

    (obj, module, name))

pickle.PicklingError: Can't pickle <class '__main__.busybox.box'>: it's
not found as __main__.busybox.box

>>> 

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081202/174af0cc/attachment.htm>

From kmihail at rocketmail.com  Wed Dec  3 11:03:42 2008
From: kmihail at rocketmail.com (Mihail Konstantinov)
Date: Wed, 3 Dec 2008 10:03:42 +0000 (GMT)
Subject: [C++-sig] pure class, to expose or not to expose?
Message-ID: <721544.61664.qm@web28401.mail.ukl.yahoo.com>

Hello,
I need help again.

In the example below I either get the error message during compilation:
> ...
>/home/mihail/temporary/boost_1_37_0/boost/python/object/value_holder.hpp:66: error: cannot declare field 
>?boost::python::objects::value_holder<A0>::m_held? to be of abstract type ?A0?
>boost.cpp:5: note:   because the following virtual functions are pure within ?A0?:
>boost.cpp:7: note:     virtual void A0::pure()

or, if I don't expose class A0 (as it is currently commented out), when importing the module in python:
> <type 'exceptions.RuntimeError'>: extension class wrapper for base class A0 has not been created yet

How can I fix the problem? I don't need to expose A0, well, exposing pure classes doesn't make sense anyway, does it?
But I need to be able to call in python boost_ext.B(boost_ext.A()), so that I think I have to declare the A0<-A inheritance relationship somehow.

Thank you
Mihail

boost.cpp:

// A is a concrete implementation of the abstract class A0.
// the constructor of B<T> takes the abstract class A0
class A0{
public:
  virtual void pure()=0;
  virtual ~A0(){};
};

class A: public A0{
public:
  void pure(){};
  ~A(){};
};

template<class T> class B{
  T value;
public:
  B(const A0&){};
};

typedef B<double> Lumi;

BOOST_PYTHON_MODULE(boost_ext)
{
    using namespace boost::python;
    //class_<A0>("A0");
    class_<A,bases<A0> >("A");
    class_<Lumi>("Lumi",init<const A0&>());
}



      


From renatox at gmail.com  Wed Dec  3 19:04:31 2008
From: renatox at gmail.com (Renato Araujo)
Date: Wed, 3 Dec 2008 15:04:31 -0300
Subject: [C++-sig] PyObject RefCount
Message-ID: <95291a80812031004v7d1276afl66a86b09b19996d2@mail.gmail.com>

Hi all,

I'm trying check  if have someone using my object in python, but I got
some problems I think I'm not doing this in the correct way:

I have a function like that:

voi MyClass::~MyClass()
{
    for(int i=0; i < childCount(); i++)
    {
        BaseClass *child = child_at(i);

        python::object py_obj(pyhton::ptr(child));  //I can't this
without ptr because my object no have copy constructor
        //Whats happen here??  ^^^
                  - How this work when the object already exists? and
whats happen when the object not exists?
                  - There is a better way to verify if already have a
python object associate with my child c++ object?

        if (py_obj.ptr()->ob_refcnt == 1)
             delete child;
         else
          {
              //do nothing python GC will destroy this
           }
     }
}

How I can fix this?? How is the best way to do this?

Thanks


-- 
Renato Araujo Oliveira Filho

From seefeld at sympatico.ca  Wed Dec  3 20:16:44 2008
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Wed, 03 Dec 2008 14:16:44 -0500
Subject: [C++-sig] PyObject RefCount
In-Reply-To: <95291a80812031004v7d1276afl66a86b09b19996d2@mail.gmail.com>
References: <95291a80812031004v7d1276afl66a86b09b19996d2@mail.gmail.com>
Message-ID: <4936DB1C.7010104@sympatico.ca>

Renato Araujo wrote:
> Hi all,
>
> I'm trying check  if have someone using my object in python, but I got
> some problems I think I'm not doing this in the correct way:
>   
First of all, I'm not sure I understand what you are trying to achieve. 
Why do you want to manually delete reference-counted objects, instead of 
letting Python's runtime do this ?

> I have a function like that:
>
> voi MyClass::~MyClass()
> {
>     for(int i=0; i < childCount(); i++)
>     {
>         BaseClass *child = child_at(i);
>
>         python::object py_obj(pyhton::ptr(child));  //I can't this
> without ptr because my object no have copy constructor
>         //Whats happen here??  ^^^
>   

By default, the object would be passed by-value, i.e. a copy would be 
made, and assigned to the python wrapper object.
python::ptr() is basically a marker to tell boost.python to copy 
by-reference (by-pointer, as it stands) instead.

>                   - How this work when the object already exists? and
> whats happen when the object not exists?
>   
In both cases, I believe a new python wrapper object is created. This, 
of course, is problematic if the object is already ref-counted 
elsewhere, since now you have got two reference counters, both 
attempting to delete the object once they drop down to 0.

>                   - There is a better way to verify if already have a
> python object associate with my child c++ object?
>   

I don't think there is any way to get from the wrapped object to the 
wrapper, as I don't know of any place that would provide this mapping.

Regards,
       Stefan

-- 

      ...ich hab' noch einen Koffer in Berlin...


From renatox at gmail.com  Wed Dec  3 20:39:12 2008
From: renatox at gmail.com (Renato Araujo)
Date: Wed, 3 Dec 2008 16:39:12 -0300
Subject: [C++-sig] PyObject RefCount
In-Reply-To: <4936DB1C.7010104@sympatico.ca>
References: <95291a80812031004v7d1276afl66a86b09b19996d2@mail.gmail.com>
	<4936DB1C.7010104@sympatico.ca>
Message-ID: <95291a80812031139l55285ab7y3dc79d05a5a0207a@mail.gmail.com>

Thanks for your points, I will try explain my problem better:

In my library I have a parent->child control then when the parent is
deleted all children is deleted too.

but in python I would like change this behavior, if there is some
python object using my child I will not delete this object.

The problem here is, some children is created in c++ code without
communicate to python. Then during the parent destructor I would like
to know which child are create in c++ and which is created in python,
for check this I'm using refcount, if recount > 1 then there is some
python object using this else I can delete this.

But after some tests this way to create python object using
"obj(ptr(my_object))" if the object is new (not exists in python) the
refcount is 1 and if there is only 1 python object using this the
refcount is 1 too, then this cause me a problem.




On Wed, Dec 3, 2008 at 4:16 PM, Stefan Seefeld <seefeld at sympatico.ca> wrote:
> Renato Araujo wrote:
>>
>> Hi all,
>>
>> I'm trying check  if have someone using my object in python, but I got
>> some problems I think I'm not doing this in the correct way:
>>
>
> First of all, I'm not sure I understand what you are trying to achieve. Why
> do you want to manually delete reference-counted objects, instead of letting
> Python's runtime do this ?

Renato Araujo Oliveira Filho

From seefeld at sympatico.ca  Wed Dec  3 20:53:04 2008
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Wed, 03 Dec 2008 14:53:04 -0500
Subject: [C++-sig] PyObject RefCount
In-Reply-To: <95291a80812031139l55285ab7y3dc79d05a5a0207a@mail.gmail.com>
References: <95291a80812031004v7d1276afl66a86b09b19996d2@mail.gmail.com>	<4936DB1C.7010104@sympatico.ca>
	<95291a80812031139l55285ab7y3dc79d05a5a0207a@mail.gmail.com>
Message-ID: <4936E3A0.1030208@sympatico.ca>

Renato Araujo wrote:
> Thanks for your points, I will try explain my problem better:
>
> In my library I have a parent->child control then when the parent is
> deleted all children is deleted too.
>
> but in python I would like change this behavior, if there is some
> python object using my child I will not delete this object.
>   

I see. What if your parent held its children via shared_ptr<> objects ? 
If you then export the child's type with shared_ptr<> as HeldType, i.e.

class_<BaseClass, shared_ptr<BaseClass>, ...>

you should get what you want, since now all the communication across the 
language boundary is using shared_ptr<>, and the ref counts are 
correctly shared. With that, when the Python wrapper itself gets deleted 
by the Python runtime, it will decrement the shared_ptr<> counter you 
passed it. As your parents do the same, they can happily coexist with 
python references to the same objects.

HTH,
       Stefan

-- 

      ...ich hab' noch einen Koffer in Berlin...


From paniq at paniq.org  Wed Dec  3 22:27:51 2008
From: paniq at paniq.org (Leonard Ritter)
Date: Wed, 3 Dec 2008 22:27:51 +0100
Subject: [C++-sig] wrapper and shared_ptr - how to combine?
In-Reply-To: <87skpaldg7.fsf@mcbain.luannocracy.com>
References: <32c88000811150811p545ccfaaqc8a566ee1301dbc1@mail.gmail.com>
	<491EF6B0.3000208@sympatico.ca>
	<32c88000811151007r1ef9d422x2d9a6a80ca993002@mail.gmail.com>
	<32c88000811151148j737fc2a2m9d6a2905b3ea63a@mail.gmail.com>
	<32c88000811151258h7ba10e1aif18ab013db4434df@mail.gmail.com>
	<87skpaldg7.fsf@mcbain.luannocracy.com>
Message-ID: <32c88000812031327i420c8880s6dbd27dd1c638057@mail.gmail.com>

On Sat, Nov 29, 2008 at 5:11 PM, David Abrahams <dave at boostpro.com> wrote:

> Yoiks!  You clearly need gSTLFilt.  Please post such things as
> attachments or in some other way that doesn't insert linebreaks.


oh, awesome suggestion, thank you very much! gcc/eclipse outputs that stuff
this way, and gmail inserts the linebreaks.


> So the problem is that you're trying to wrap PyActor in a way that
> implies it is default-constructible.  But it's not; it's an abstract
> class.  You need no_init.  See
>

thank you, i will remember your advice when i have that issue again. i
solved the problem by writing the add/remove methods in python, which does
fine so far.

of all binding sets available, boost.python, although intellectually
challenging, leaves no wishes open and provides the most coherent way to do
C++ bindings. you have done extraordinary work there.

what will you do when PyPy grows mature? ;)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081203/cb3b9e89/attachment.htm>

From kekela at gmx.net  Fri Dec  5 11:48:02 2008
From: kekela at gmx.net (kekela at gmx.net)
Date: Fri, 05 Dec 2008 11:48:02 +0100
Subject: [C++-sig] ImportError...'No such file' etc. when importing
 hello_ext from tutorial
Message-ID: <20081205104802.172990@gmx.net>

Hi,

I'm new to boost and aiming to use boost:python and I am experiencing difficulties with the boost:python tutorial. While running bjam plain (i.e. with test targets and without --preserve-test-target or -n -a) creates nice test results (i.e. hello, hello.output and hello.test are created with "hello, world\n\nEXIT STATUS: 0" as well as "passed"), I cannot import hello_ext in python, when bjamming without test targets etc.

But then again, maybe I shouldn't try to import hello_ext.so but have some kind of hello_ext.pyd?

Well, what happens is this:

[XXXX release]$ ls
hello_ext.so  hello.o
[XXXX release]$ python
Python 2.4.3 (#1, May 24 2008, 13:57:05) 
[GCC 4.1.2 20070626 (Red Hat 4.1.2-14)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import hello_ext
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
ImportError: libboost_python-gcc41-1_37.so.1.37.0: cannot open shared object file: No such file or directory


Actually I wasn't able to locate any libboost* files yet, but since the tests are passed, I'm probably doing something wrong. Please guide me!

Thank you very much.

From kekela at gmx.net  Fri Dec  5 13:39:44 2008
From: kekela at gmx.net (kekela at gmx.net)
Date: Fri, 05 Dec 2008 13:39:44 +0100
Subject: [C++-sig] ImportError...'No such file' etc. when importing
 hello_ext from tutorial
In-Reply-To: <d73be26c0812050256u26986ea3j92c6cadd3c084d6e@mail.gmail.com>
References: <20081205104802.172990@gmx.net>
	<d73be26c0812050256u26986ea3j92c6cadd3c084d6e@mail.gmail.com>
Message-ID: <20081205123944.274100@gmx.net>

> Maybe your library path is incorrect. You may try (with bash)
> export LD_LIBRARY_PATH=path_to_your_libboost_python-gcc41-1_37.so.1.37.0

Now it works. Well, yes, there were 3 issues. First, I had skipped the "make install" part in "getting started with boost", i.e. I didn't have any libboost* files (of which I know explicitly). Second, LD_LIBRARY_PATH was not set and, third, libboost_python-gcc41-1_37.so.1.37.0 didn't exist after making. I just copied "libboost_python-gcc41-mt-1_37.so.1.37.0" to "libboost_python-gcc41-1_37.so.1.37.0" in $LD_LIBRARY_PATH/.

Thank you very much for your quick response.

From pyplusplus at assumetheposition.nl  Fri Dec  5 13:39:23 2008
From: pyplusplus at assumetheposition.nl (pyplusplus at assumetheposition.nl)
Date: Fri, 5 Dec 2008 13:39:23 +0100 (CET)
Subject: [C++-sig] Custom from-python converter (almost working)
Message-ID: <34319.146.50.22.50.1228480763.squirrel@www.assumetheposition.nl>

I'm trying to add some custom from-python converters to my bindings (see
code at the end of the mail).
These basically should convert Python 3-tuples of floats to a C++ vec3f
instance. I can't seem to find reference docs on classes related to
conversion, e.g. boost::python::converter::registry, so after some
googling and experimenting I think the code at the end of this mail is a
pretty clean implementation of what I want, and it almost fully works:

Python 2.4.3 (#1, Jun 13 2006, 16:41:18)
[GCC 4.0.2 20051125 (Red Hat 4.0.2-8)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import doh
>>> doh.f((1,2,3))
1.000000, 2.000000, 3.000000
>>> doh.p((1,2,3))
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
Boost.Python.ArgumentError: Python argument types in
    doh.p(tuple)
did not match C++ signature:
    p(vec3f*)
>>>

So the implicit conversion to a pointer to a vec3f doesn't work. After
some more googling I came across the thread [1], which seemed to suggest
that there is another way of registering a converter, i.e.

"""
    Don't try to convert to pointers; there are no pointers in
    Boost.Python; only lvalues and rvalues ;-)

      void* extract_foo(PyObject* op)
      {
          return FooObject_ptr(op);
      }

      boost::python::converter::registry::insert(
                &extractor_foo, boost::python::type_id<FOO>());
"""

In the February 2002 progress report [2] (does that reflect the current
situation in 1.37 w.r.t. to/from-python converters, b.t.w?) it says:

"""There are basically two categories of from_python conversions:
those which lvalues stored within or held by the Python object
(essentially extractions), like what happens when an instance of a
C++ class exposed with class_ is used as the target of a wrapped
member function), and those in which a new rvalue gets created, as
when a Python Float is converted to a C++ complex<double> or a Python
tuple is converted to a C++ std::vector<>"""

It seems my non-working case of conversion to a vec3f* is of the latter
category.

Can I assume that the two different ways of registering a converter
reflect these categories?

If so, then I don't see a way to make this work, as I need actual
conversion, not extraction, i.e. I need to create a new object that
doesn't exist yet. Or should I register my converter in both ways to the
registry?

Any help is greatly appreciated,

Regards,
Paul

[1] http://mail.python.org/pipermail/cplusplus-sig/2006-November/011269.html
[2] http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/feb2002.html




#include <boost/python.hpp>

namespace bp = boost::python;

struct vec3f
{
    vec3f(): x(0), y(0), z(0) {}

    vec3f(float xx, float yy, float zz)
    {
        x = xx; y = yy; z = zz;
    }

    float x, y, z;
};

struct convert_py_tuple_to_vec3f
{
    convert_py_tuple_to_vec3f()
    {
        bp::converter::registry::push_back(
            &convertible,
            &construct,
            bp::type_id<vec3f>());
    }

    // Check if given Python object is convertible to a vec3f.
    // If so, return obj, otherwise return 0
    static void* convertible(PyObject* obj)
    {
        if (PyTuple_Check(obj))
            return obj;

        return 0;
    }

    // Construct a vec3f object from the given Python object, and
    // store it in the stage1 (?) data.
    static void construct(PyObject* obj,
bp::converter::rvalue_from_python_stage1_data* data)
    {
        // Fill in values
        float x, y, z;
        if (!PyArg_ParseTuple(obj, "fff", &x, &y, &z))
        {
            // Raise exception, error will have been set by PyArg_ParseTuple
            boost::python::throw_error_already_set();
        }

        typedef bp::converter::rvalue_from_python_storage<vec3f>
vec3f_storage;
        void* const storage =
reinterpret_cast<vec3f_storage*>(data)->storage.bytes;
        vec3f *v = new (storage) vec3f(x, y, z);

        data->convertible = storage;
    }
};

void
f(vec3f v)
{
    printf("%f, %f, %f\n", v.x, v.y, v.z);
}

void
p(vec3f *v)
{
    printf("%f, %f, %f\n", v->x, v->y, v->z);
}

BOOST_PYTHON_MODULE(doh)
{
    bp::class_< vec3f >("vec3f")
        .def(bp::init<float, float, float>())
    ;

    bp::def("f", &f);
    bp::def("p", &p);

    convert_py_tuple_to_vec3f();
}



From macke at yar.nu  Fri Dec  5 16:01:09 2008
From: macke at yar.nu (Marcus Lindblom)
Date: Fri, 05 Dec 2008 16:01:09 +0100
Subject: [C++-sig] [py++] Problem: empty xml file created
In-Reply-To: <4935423C.4030203@yar.nu>
References: <4934F5DD.2060102@yar.nu>	<7465b6170812020108x3520eedeq37a802152802c156@mail.gmail.com>	<493534FA.1020703@yar.nu>	<7465b6170812020528y2834955bp3a09b8c0a18026cc@mail.gmail.com>	<49353FB5.20809@yar.nu>
	<4935423C.4030203@yar.nu>
Message-ID: <49394235.5030909@yar.nu>

Marcus Lindblom wrote:
> Marcus Lindblom wrote:
> 
>> (I don't think the gccxml-call happens because of the call to the 
>> ModuleBuilder at line 813, although I might be wrong?)
> 
> Waitaminute. When I'm rerunning it now, the order makes more sense, and 
> the cache-calls come up after the calls to ModuleBuilder. Probably the 
> log-file merging of stdout and stderr was playing tricks on me.
> 
> I'll look augmenting the ModuleBuilder call, just as you said. Sorry for 
> the confusion. :-/

I've gotten it to work. :) Now I just have to try to make it parse 
successfully, which it doesn't. :-/

How does gccxml work in compiler-work-alike mode?

I've built with VS9, and am going to compile the wrappers with VS9, so I 
suppose I have to tell gccxml and the code to behave like that, with 
defines and all? (Or should that happen automatically?)

Namely, I get some compile errors from gccxml on code that VS9 accepted, 
so I'm trying to figure out how to approach the problem, before going 
into specifics.

Cheers,
/Marcus

From rwgk at yahoo.com  Fri Dec  5 18:42:45 2008
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Fri, 5 Dec 2008 09:42:45 -0800 (PST)
Subject: [C++-sig] Custom from-python converter (almost working)
References: <34319.146.50.22.50.1228480763.squirrel@www.assumetheposition.nl>
Message-ID: <190802.95882.qm@web111403.mail.gq1.yahoo.com>

> void

> p(vec3f *v)
> {
>     printf("%f, %f, %f\n", v->x, v->y, v->z);
> }

This is indeed not supported, since it would require "converters with write-back".

vec3f const* v

should work, but

ve3f* v

doesn't because the missing const indicates that you want to modify the pointee in place.
For a Python tuple this can definitely not work since tuples are immutable.
It could in theory work for a Python list, but Boost.Python doesn't support this.
A few years ago we had extensive discussions about this, but it was too hard to
implement.

Ralf

P.S.: I'm using this seasoned file for all "list/tuple/iter <-> small C++ array" conversions:

http://cctbx.svn.sourceforge.net/viewvc/cctbx/trunk/scitbx/boost_python/container_conversions.h?view=markup

Probably, all you need is

  tuple_mapping_fixed_capacity<vec3f>

From roman.yakovenko at gmail.com  Fri Dec  5 20:09:49 2008
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Fri, 5 Dec 2008 21:09:49 +0200
Subject: [C++-sig] [py++] Problem: empty xml file created
In-Reply-To: <49394235.5030909@yar.nu>
References: <4934F5DD.2060102@yar.nu>
	<7465b6170812020108x3520eedeq37a802152802c156@mail.gmail.com>
	<493534FA.1020703@yar.nu>
	<7465b6170812020528y2834955bp3a09b8c0a18026cc@mail.gmail.com>
	<49353FB5.20809@yar.nu> <4935423C.4030203@yar.nu>
	<49394235.5030909@yar.nu>
Message-ID: <7465b6170812051109x7d24c823l8ddcf89bcd8ff0c@mail.gmail.com>

On Fri, Dec 5, 2008 at 5:01 PM, Marcus Lindblom <macke at yar.nu> wrote:
> How does gccxml work in compiler-work-alike mode?

I am not sure what you mean by this. GCC-XML doesn't tries to emulate
other compilers. The configuration is used to find system and C++
header files.

> I've built with VS9, and am going to compile the wrappers with VS9, so I
> suppose I have to tell gccxml and the code to behave like that, with defines
> and all? (Or should that happen automatically?)

In your case, it doesn't really matter. Just select the compiler
GCCXML supports better. I think this is one of Visual Studio 2003 or
2005.

> Namely, I get some compile errors from gccxml on code that VS9 accepted, so
> I'm trying to figure out how to approach the problem, before going into
> specifics.

The bottom line - you will have to change the code. GCCXML defines
__GCCXML__ define. (http://gccxml.org/HTML/Running.html ). So you can
introduce the changes for this compiler only.

HTH

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

From macke at yar.nu  Fri Dec  5 16:01:09 2008
From: macke at yar.nu (Marcus Lindblom)
Date: Fri, 05 Dec 2008 16:01:09 +0100
Subject: [C++-sig] [py++] Problem: empty xml file created
In-Reply-To: <4935423C.4030203@yar.nu>
References: <4934F5DD.2060102@yar.nu>	<7465b6170812020108x3520eedeq37a802152802c156@mail.gmail.com>	<493534FA.1020703@yar.nu>	<7465b6170812020528y2834955bp3a09b8c0a18026cc@mail.gmail.com>	<49353FB5.20809@yar.nu>
	<4935423C.4030203@yar.nu>
Message-ID: <49394235.5030909@yar.nu>

Marcus Lindblom wrote:
> Marcus Lindblom wrote:
> 
>> (I don't think the gccxml-call happens because of the call to the 
>> ModuleBuilder at line 813, although I might be wrong?)
> 
> Waitaminute. When I'm rerunning it now, the order makes more sense, and 
> the cache-calls come up after the calls to ModuleBuilder. Probably the 
> log-file merging of stdout and stderr was playing tricks on me.
> 
> I'll look augmenting the ModuleBuilder call, just as you said. Sorry for 
> the confusion. :-/

I've gotten it to work. :) Now I just have to try to make it parse 
successfully, which it doesn't. :-/

How does gccxml work in compiler-work-alike mode?

I've built with VS9, and am going to compile the wrappers with VS9, so I 
suppose I have to tell gccxml and the code to behave like that, with 
defines and all? (Or should that happen automatically?)

Namely, I get some compile errors from gccxml on code that VS9 accepted, 
so I'm trying to figure out how to approach the problem, before going 
into specifics.

Cheers,
/Marcus


From adrien at saladin.fr  Fri Dec  5 22:54:41 2008
From: adrien at saladin.fr (Adrien Saladin)
Date: Fri, 5 Dec 2008 22:54:41 +0100
Subject: [C++-sig] ImportError...'No such file' etc. when importing
	hello_ext from tutorial
In-Reply-To: <20081205123944.274100@gmx.net>
References: <20081205104802.172990@gmx.net>
	<d73be26c0812050256u26986ea3j92c6cadd3c084d6e@mail.gmail.com>
	<20081205123944.274100@gmx.net>
Message-ID: <d73be26c0812051354n27a506cak261f2a694bb13289@mail.gmail.com>

> Now it works. Well, yes, there were 3 issues. First, I had skipped the "make install" part in "getting started with boost", i.e. I didn't have any libboost* files (of which I know explicitly). Second, LD_LIBRARY_PATH was not set and, third, libboost_python-gcc41-1_37.so.1.37.0 didn't exist after making. I just copied "libboost_python-gcc41-mt-1_37.so.1.37.0" to "libboost_python-gcc41-1_37.so.1.37.0" in $LD_LIBRARY_PATH/.
>
> Thank you very much for your quick response.


You're welcome.
I remember that you mentioned bjam, and if you like it then continue
with it, but don't trust the documentation website: many other tools
can compile and link with boost.python, not only bjam. I'm using SCons
but make and cmake works as well.

I'm telling that because I really dislike bjam syntax and when I began
with boost.python I thought that boost.python was not for me just
because of this building tool.

Also if you begin with boost.python you must have a look at pyplusplus
(or Py++), it's a very convenient tool that parses your C++ headers
and generates boost.python code almost automatically. I'm using it for
two years now and I'm really happy with it.

There is also one problem with boost.python in general: library tends
to become bigger and bigger with the number of functions and classes
that you interface. Compile-time may also become an issue at some
point. My library (~7000 lines of C++, 15 classes) now takes 50sec on
a 8 cores 3Ghz computer. About 45sec are for the boost.python
bindings...
I'm now considering switching to pybindgen, but it's still a very new
project that lacks some features of boost.python.

Just curious, for what type of project do you wish to learn boost.python ?

Best,
Adrien

From macke at yar.nu  Sat Dec  6 10:52:50 2008
From: macke at yar.nu (Marcus Lindblom)
Date: Sat, 06 Dec 2008 10:52:50 +0100
Subject: [C++-sig] [py++] Problem: empty xml file created
In-Reply-To: <7465b6170812051109x7d24c823l8ddcf89bcd8ff0c@mail.gmail.com>
References: <4934F5DD.2060102@yar.nu>	<7465b6170812020108x3520eedeq37a802152802c156@mail.gmail.com>	<493534FA.1020703@yar.nu>	<7465b6170812020528y2834955bp3a09b8c0a18026cc@mail.gmail.com>	<49353FB5.20809@yar.nu>
	<4935423C.4030203@yar.nu>	<49394235.5030909@yar.nu>
	<7465b6170812051109x7d24c823l8ddcf89bcd8ff0c@mail.gmail.com>
Message-ID: <493A4B72.8030707@yar.nu>

Roman Yakovenko wrote:
> On Fri, Dec 5, 2008 at 5:01 PM, Marcus Lindblom <macke at yar.nu> wrote:
>> How does gccxml work in compiler-work-alike mode?

[snip]

> I am not sure what you mean by this. GCC-XML doesn't tries to emulate
> other compilers. The configuration is used to find system and C++
> header files.

[snip]

> The bottom line - you will have to change the code. GCCXML defines
> __GCCXML__ define. (http://gccxml.org/HTML/Running.html ). So you can
> introduce the changes for this compiler only.

That was exactly what I needed to know.

Thanks a lot!

/Marcus

From macke at yar.nu  Sat Dec  6 10:52:50 2008
From: macke at yar.nu (Marcus Lindblom)
Date: Sat, 06 Dec 2008 10:52:50 +0100
Subject: [C++-sig] [py++] Problem: empty xml file created
In-Reply-To: <7465b6170812051109x7d24c823l8ddcf89bcd8ff0c@mail.gmail.com>
References: <4934F5DD.2060102@yar.nu>	<7465b6170812020108x3520eedeq37a802152802c156@mail.gmail.com>	<493534FA.1020703@yar.nu>	<7465b6170812020528y2834955bp3a09b8c0a18026cc@mail.gmail.com>	<49353FB5.20809@yar.nu>
	<4935423C.4030203@yar.nu>	<49394235.5030909@yar.nu>
	<7465b6170812051109x7d24c823l8ddcf89bcd8ff0c@mail.gmail.com>
Message-ID: <493A4B72.8030707@yar.nu>

Roman Yakovenko wrote:
> On Fri, Dec 5, 2008 at 5:01 PM, Marcus Lindblom <macke at yar.nu> wrote:
>> How does gccxml work in compiler-work-alike mode?

[snip]

> I am not sure what you mean by this. GCC-XML doesn't tries to emulate
> other compilers. The configuration is used to find system and C++
> header files.

[snip]

> The bottom line - you will have to change the code. GCCXML defines
> __GCCXML__ define. (http://gccxml.org/HTML/Running.html ). So you can
> introduce the changes for this compiler only.

That was exactly what I needed to know.

Thanks a lot!

/Marcus


From pyplusplus at assumetheposition.nl  Mon Dec  8 10:34:28 2008
From: pyplusplus at assumetheposition.nl (pyplusplus at assumetheposition.nl)
Date: Mon, 8 Dec 2008 10:34:28 +0100 (CET)
Subject: [C++-sig] Custom from-python converter (almost working)
In-Reply-To: <190802.95882.qm@web111403.mail.gq1.yahoo.com>
References: <34319.146.50.22.50.1228480763.squirrel@www.assumetheposition.nl>
	<190802.95882.qm@web111403.mail.gq1.yahoo.com>
Message-ID: <48434.146.50.22.50.1228728868.squirrel@www.assumetheposition.nl>

Hi,

>> void
>> p(vec3f *v)
>> {
>>     printf("%f, %f, %f\n", v->x, v->y, v->z);
>> }
>
> This is indeed not supported, since it would require "converters with
> write-back".

Ok, that makes sense.

> vec3f const* v
>
> should work, but
>
> ve3f* v
>
> doesn't because the missing const indicates that you want to modify the
> pointee in place.

Unfortunately I don't control the API. The actual use case I'm working on
(the vec3f was just a simplified example) is a class method that takes a
non-const pointer, but it does not modify the pointee. The pointee is
basically stored in the instance and is used as a data array. As such I
wanted a conversion from Python list of tuples to a C++ instance.

> For a Python tuple this can definitely not work since tuples are
> immutable.
> It could in theory work for a Python list, but Boost.Python doesn't
> support this.
> A few years ago we had extensive discussions about this, but it was too
> hard to
> implement.

I think I can imagine some obstacles :)

In SWIG it is possible to simply define an extra class method that takes a
PyObject* and perform any conversion you want in that method (although
this means having to add that extra method for all cases where the
non-const pointer is used). Would something similar work in Boost.Python,
e.g.

void
extra_method(C& self, PyObject *obj)
{

}

class_<C>("C")


>
> Ralf
>
> P.S.: I'm using this seasoned file for all "list/tuple/iter <-> small C++
> array" conversions:
>
> http://cctbx.svn.sourceforge.net/viewvc/cctbx/trunk/scitbx/boost_python/container_conversions.h?view=markup
>
> Probably, all you need is
>
>   tuple_mapping_fixed_capacity<vec3f>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>



From pyplusplus at assumetheposition.nl  Mon Dec  8 10:37:58 2008
From: pyplusplus at assumetheposition.nl (pyplusplus at assumetheposition.nl)
Date: Mon, 8 Dec 2008 10:37:58 +0100 (CET)
Subject: [C++-sig] Custom from-python converter (almost working)
In-Reply-To: <48434.146.50.22.50.1228728868.squirrel@www.assumetheposition.nl>
References: <34319.146.50.22.50.1228480763.squirrel@www.assumetheposition.nl>
	<190802.95882.qm@web111403.mail.gq1.yahoo.com>
	<48434.146.50.22.50.1228728868.squirrel@www.assumetheposition.nl>
Message-ID: <49804.146.50.22.50.1228729078.squirrel@www.assumetheposition.nl>

Dang, my mailer got in the way and sent this before I finished it ...

> Hi,
>
>>> void
>>> p(vec3f *v)
>>> {
>>>     printf("%f, %f, %f\n", v->x, v->y, v->z);
>>> }
>>
>> This is indeed not supported, since it would require "converters with
>> write-back".
>
> Ok, that makes sense.
>
>> vec3f const* v
>>
>> should work, but
>>
>> ve3f* v
>>
>> doesn't because the missing const indicates that you want to modify the
>> pointee in place.
>
> Unfortunately I don't control the API. The actual use case I'm working on
> (the vec3f was just a simplified example) is a class method that takes a
> non-const pointer, but it does not modify the pointee. The pointee is
> basically stored in the instance and is used as a data array. As such I
> wanted a conversion from Python list of tuples to a C++ instance.
>
>> For a Python tuple this can definitely not work since tuples are
>> immutable.
>> It could in theory work for a Python list, but Boost.Python doesn't
>> support this.
>> A few years ago we had extensive discussions about this, but it was too
>> hard to
>> implement.
>
> I think I can imagine some obstacles :)
>
In SWIG it is possible to simply define an extra class method that takes a
PyObject* and perform any conversion you want in that method (although
this means having to add that extra method for all cases where the
non-const pointer is used). Would something similar work in Boost.Python,
e.g.

class C
{
public:
    void set_data(data* d);
};

void
extra_method(C& self, PyObject *obj)
{
    // if obj is list of tuples, convert to
    // data* and
    // call self.set_data(d)
}

class_<C>("C")
   .def("set_data", &C::set_data)
   .def("set_data", &C::extra_method)
;

Or will this completely muck up the way Python objects are handled on the
interface?

>> Ralf
>>
>> P.S.: I'm using this seasoned file for all "list/tuple/iter <-> small
>> C++
>> array" conversions:
>>
>> http://cctbx.svn.sourceforge.net/viewvc/cctbx/trunk/scitbx/boost_python/container_conversions.h?view=markup
>>
>> Probably, all you need is
>>
>>   tuple_mapping_fixed_capacity<vec3f>

Thanks for the hint,

Paul


From pyplusplus at assumetheposition.nl  Mon Dec  8 15:23:50 2008
From: pyplusplus at assumetheposition.nl (pyplusplus at assumetheposition.nl)
Date: Mon, 8 Dec 2008 15:23:50 +0100 (CET)
Subject: [C++-sig] Getting address of wrapped instance?
Message-ID: <46613.146.50.22.50.1228746230.squirrel@www.assumetheposition.nl>

Is there are way to get the address of the C++ instance pointed to by a
given Boost.Python wrapper object? I don't need a real pointer, the
address alone suffices.

The use case is to deduce interrelations between C++ objects (think a DAG)
from Python. As different Python wrapper objects might reference the same
C++ instance I can't use id() or repr() of the Python object.

Thanks,
Paul


From renatox at gmail.com  Mon Dec  8 15:36:15 2008
From: renatox at gmail.com (Renato Araujo)
Date: Mon, 8 Dec 2008 11:36:15 -0300
Subject: [C++-sig] Getting address of wrapped instance?
In-Reply-To: <46613.146.50.22.50.1228746230.squirrel@www.assumetheposition.nl>
References: <46613.146.50.22.50.1228746230.squirrel@www.assumetheposition.nl>
Message-ID: <95291a80812080636s6cc9acddm34d800cfee5fe5b1@mail.gmail.com>

Hi Paul,

Maybe this can help you:
http://www.boost.org/doc/libs/1_36_0/libs/python/doc/v2/faq.html#xref

I use:
  python::object o(python::ptr(my_cpp_pointer));

then I got the python object.

BR

On Mon, Dec 8, 2008 at 11:23 AM,  <pyplusplus at assumetheposition.nl> wrote:
> Is there are way to get the address of the C++ instance pointed to by a
> given Boost.Python wrapper object? I don't need a real pointer, the
> address alone suffices.
>
> The use case is to deduce interrelations between C++ objects (think a DAG)
> from Python. As different Python wrapper objects might reference the same
> C++ instance I can't use id() or repr() of the Python object.
>
> Thanks,
> Paul
>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>



-- 
Renato Araujo Oliveira Filho

From pyplusplus at assumetheposition.nl  Mon Dec  8 16:04:41 2008
From: pyplusplus at assumetheposition.nl (Paul Melis)
Date: Mon, 8 Dec 2008 16:04:41 +0100 (CET)
Subject: [C++-sig] Getting address of wrapped instance?
In-Reply-To: <95291a80812080636s6cc9acddm34d800cfee5fe5b1@mail.gmail.com>
References: <46613.146.50.22.50.1228746230.squirrel@www.assumetheposition.nl>
	<95291a80812080636s6cc9acddm34d800cfee5fe5b1@mail.gmail.com>
Message-ID: <46765.146.50.22.50.1228748681.squirrel@www.assumetheposition.nl>

Hello Renato,

On Mon, December 8, 2008 3:36 pm, Renato Araujo wrote:
> Maybe this can help you:
> http://www.boost.org/doc/libs/1_36_0/libs/python/doc/v2/faq.html#xref
>
> I use:
>   python::object o(python::ptr(my_cpp_pointer));
>
> then I got the python object.

Unfortunately, I need the (address of) the wrapped C++ instance from a
Python wrapper object, not a Python wrapper object for a C++ instance.

Paul

>
> BR
>
> On Mon, Dec 8, 2008 at 11:23 AM,  <pyplusplus at assumetheposition.nl> wrote:
>> Is there are way to get the address of the C++ instance pointed to by a
>> given Boost.Python wrapper object? I don't need a real pointer, the
>> address alone suffices.
>>
>> The use case is to deduce interrelations between C++ objects (think a
>> DAG)
>> from Python. As different Python wrapper objects might reference the
>> same
>> C++ instance I can't use id() or repr() of the Python object.
>>
>> Thanks,
>> Paul
>>
>> _______________________________________________
>> Cplusplus-sig mailing list
>> Cplusplus-sig at python.org
>> http://mail.python.org/mailman/listinfo/cplusplus-sig
>>
>
>
>
> --
> Renato Araujo Oliveira Filho
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>



From seefeld at sympatico.ca  Mon Dec  8 16:12:06 2008
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Mon, 08 Dec 2008 10:12:06 -0500
Subject: [C++-sig] Getting address of wrapped instance?
In-Reply-To: <46613.146.50.22.50.1228746230.squirrel@www.assumetheposition.nl>
References: <46613.146.50.22.50.1228746230.squirrel@www.assumetheposition.nl>
Message-ID: <493D3946.3030801@sympatico.ca>

pyplusplus at assumetheposition.nl wrote:
> Is there are way to get the address of the C++ instance pointed to by a
> given Boost.Python wrapper object? I don't need a real pointer, the
> address alone suffices.
>   

You may try using

type const &ref = extract<type const &>(object);


HTH,
       Stefan

-- 

      ...ich hab' noch einen Koffer in Berlin...


From pyplusplus at assumetheposition.nl  Mon Dec  8 16:23:49 2008
From: pyplusplus at assumetheposition.nl (Paul Melis)
Date: Mon, 8 Dec 2008 16:23:49 +0100 (CET)
Subject: [C++-sig] Getting address of wrapped instance?
In-Reply-To: <493D3946.3030801@sympatico.ca>
References: <46613.146.50.22.50.1228746230.squirrel@www.assumetheposition.nl>
	<493D3946.3030801@sympatico.ca>
Message-ID: <46846.146.50.22.50.1228749829.squirrel@www.assumetheposition.nl>

On Mon, December 8, 2008 4:12 pm, Stefan Seefeld wrote:
> pyplusplus at assumetheposition.nl wrote:
>> Is there are way to get the address of the C++ instance pointed to by a
>> given Boost.Python wrapper object? I don't need a real pointer, the
>> address alone suffices.
>>
>
> You may try using
>
> type const &ref = extract<type const &>(object);

Hmmm, reading my own post I now see I wasn't completely clear.

I need this *in Python*. So for a given Python instance that wraps a C++
instance I need the address of the wrapped C++ instance.

Paul

>
>
> HTH,
>        Stefan
>
> --
>
>       ...ich hab' noch einen Koffer in Berlin...
>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>



From meine at informatik.uni-hamburg.de  Mon Dec  8 17:33:11 2008
From: meine at informatik.uni-hamburg.de (Hans Meine)
Date: Mon, 8 Dec 2008 17:33:11 +0100
Subject: [C++-sig] Getting address of wrapped instance?
In-Reply-To: <46846.146.50.22.50.1228749829.squirrel@www.assumetheposition.nl>
References: <46613.146.50.22.50.1228746230.squirrel@www.assumetheposition.nl>
	<493D3946.3030801@sympatico.ca>
	<46846.146.50.22.50.1228749829.squirrel@www.assumetheposition.nl>
Message-ID: <200812081733.12323.meine@informatik.uni-hamburg.de>

On Monday 08 December 2008 16:23:49 Paul Melis wrote:
> On Mon, December 8, 2008 4:12 pm, Stefan Seefeld wrote:
> > You may try using
> >
> > type const &ref = extract<type const &>(object);
>
> Hmmm, reading my own post I now see I wasn't completely clear.
>
> I need this *in Python*. So for a given Python instance that wraps a C++
> instance I need the address of the wrapped C++ instance.

How about adding an id() method (returning this) to the C++ class and 
exporting that?

HTH,
  Hans

From seefeld at sympatico.ca  Mon Dec  8 17:42:53 2008
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Mon, 08 Dec 2008 11:42:53 -0500
Subject: [C++-sig] Getting address of wrapped instance?
In-Reply-To: <200812081733.12323.meine@informatik.uni-hamburg.de>
References: <46613.146.50.22.50.1228746230.squirrel@www.assumetheposition.nl>	<493D3946.3030801@sympatico.ca>	<46846.146.50.22.50.1228749829.squirrel@www.assumetheposition.nl>
	<200812081733.12323.meine@informatik.uni-hamburg.de>
Message-ID: <493D4E8D.7000700@sympatico.ca>

Hans Meine wrote:
> How about adding an id() method (returning this) to the C++ class and
> exporting that?
>   

Well, this 'id' isn't really a property of the wrapped type / object, 
but the wrapper. So, I think there is little sense in adding such a 
tautological 'id' function to the class itself. A free-standing

type const *id(type const *t) { return t;}

function would be more clean, IMO.

my 2 cents,

       Stefan

-- 

      ...ich hab' noch einen Koffer in Berlin...


From gjcarneiro at gmail.com  Mon Dec  8 17:49:00 2008
From: gjcarneiro at gmail.com (Gustavo Carneiro)
Date: Mon, 8 Dec 2008 16:49:00 +0000
Subject: [C++-sig] ANN: PyBindGen 0.10 released
Message-ID: <a467ca4f0812080849s292ad71cle0cad75cfc05529a@mail.gmail.com>

PyBindGen is a Python module that is geared to generating C/C++ code that
binds a C/C++ library for Python. It does so without extensive use of either
C++ templates or C pre-processor macros. It has modular handling of C/C++
types, and can be easily extended with Python plugins. The generated code is
almost as clean as what a human programmer would write.

It can be downloaded from:

        http://code.google.com/p/pybindgen/

Bug reports should be filed here:

        https://bugs.launchpad.net/pybindgen


=== pybindgen 0.10 ===

    - New null_ok, default_value options for pointer-to-class parameters;

    - Thread safety fixes;

    - Map C++ operator() into Python's tp_call (__call__);

    - Initial support for std containers (except mapping containers);

    - Generate __copy__ methods for classes with copy constructor;

    - Add 'wrapper registry' optional feature (off by default): allows
      C++ instances returned by pointer to be consistently wrapped
      always by the same Python wrapper, so that the 'is' operator can
      be used for identity testing;

    - New C type expression parser, is_const not longer needed for
      parameter types, just put const in the type string as
      appropriate;

    - Sort the declarations returned by (py)gccxml.  If generating
      python script files from gccxml scanning, now the output file
      becomes more stable and less prone to move declarations up and
      down with different people scanning and the output file is under
      version control;

    - Map binary comparison operators (< <= == >= >) from C++ to Python;

    - Map some binary numeric operators (+ - * /) from C++ to Python;

    - Allow installation of pybindgen even if no C/C++ compiler is detected;

    - Fix compatibility with older Python versions (tested with 2.3, 2.4,
and 2.5)

-- 
Gustavo J. A. M. Carneiro
INESC Porto, Telecommunications and Multimedia Unit
"The universe is always one step beyond logic." -- Frank Herbert
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081208/ec018c9e/attachment-0001.htm>

From rwgk at yahoo.com  Mon Dec  8 18:11:09 2008
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Mon, 8 Dec 2008 09:11:09 -0800 (PST)
Subject: [C++-sig] Custom from-python converter (almost working)
References: <34319.146.50.22.50.1228480763.squirrel@www.assumetheposition.nl>
	<190802.95882.qm@web111403.mail.gq1.yahoo.com>
	<48434.146.50.22.50.1228728868.squirrel@www.assumetheposition.nl>
	<49804.146.50.22.50.1228729078.squirrel@www.assumetheposition.nl>
Message-ID: <2754.79036.qm@web111402.mail.gq1.yahoo.com>

> In SWIG it is possible to simply define an extra class method that takes a
> PyObject* and perform any conversion you want in that method (although
> this means having to add that extra method for all cases where the
> non-const pointer is used). Would something similar work in Boost.Python,
> e.g.
> 
> class C
> {
> public:
>     void set_data(data* d);
> };
> 
> void
> extra_method(C& self, PyObject *obj)
> {
>     // if obj is list of tuples, convert to
>     // data* and
>     // call self.set_data(d)
> }
> 
> class_<C>("C")
>   .def("set_data", &C::set_data)
>   .def("set_data", &C::extra_method)
> ;

Yes, this should work if you use boost::python::object:

  void
  extra_method(C& self, boost::python::object obj)

The conversion code in the body of this function will probably be
similar to what you see in the container_conversion.h file.

You could also use

  void
  extra_method(C& self, boost::python::list list_of_tuples)

which will save you one manual conversion step.

Ralf


From Matthew.Scouten at tradingtechnologies.com  Mon Dec  8 18:16:00 2008
From: Matthew.Scouten at tradingtechnologies.com (Matthew Scouten (TT))
Date: Mon, 8 Dec 2008 11:16:00 -0600
Subject: [C++-sig] Pickle an enum from c++
Message-ID: <32490DFF7774554A85D65D23A9F0F93809A51DA7@chiex01>

>From the lack of response I assume that no one has any clever ideas to
make an enum pickleable.  Thank you to anyone who put thought into this.
If I come up anything that works, I will let the group know.

 

From: Matthew Scouten (TT) 
Sent: Tuesday, December 02, 2008 2:24 PM
To: Development of Python/C++ integration
Subject: Pickle an enum from c++

 

I have an enum in my c++ library. In fact I have a crap load of them.
169 to be exact. 

 

I need them to be pickle-able.  The enum_ object does not have a
.def_pickle function, so I cannot use the pickle suite. I tried to just
up and pickle it but that does not work either.

 

 enum box {cardboard, cereal, toy};

 

BOOST_PYTHON_MODULE(busybox)

{

 

    enum_<box>("box")

        .value("cardboard",cardboard )

        .value("cereal",cereal )

        .value("toy", toy);

 

}

 

C:\Documents and Settings\mscouten>cd C:\tt-dev\manticore\

 

C:\tt-dev\manticore>call setpath.bat

Python 2.5.1 (r251:54863, Apr 18 2007, 08:51:08) [MSC v.1310 32 bit
(Intel)] on win32

Type "help", "copyright", "credits" or "license" for more information.

>>> import pickle, busybox

>>> 

>>> b = busybox.box.toy

>>> 

>>> bs = pickle.dumps(b)

Traceback (most recent call last):

  File "<stdin>", line 1, in <module>

  File "C:\Python25\lib\pickle.py", line 1366, in dumps

    Pickler(file, protocol).dump(obj)

  File "C:\Python25\lib\pickle.py", line 224, in dump

    self.save(obj)

  File "C:\Python25\lib\pickle.py", line 331, in save

    self.save_reduce(obj=obj, *rv)

  File "C:\Python25\lib\pickle.py", line 401, in save_reduce

    save(args)

  File "C:\Python25\lib\pickle.py", line 286, in save

    f(self, obj) # Call unbound method with explicit self

  File "C:\Python25\lib\pickle.py", line 562, in save_tuple

    save(element)

  File "C:\Python25\lib\pickle.py", line 286, in save

    f(self, obj) # Call unbound method with explicit self

  File "C:\Python25\lib\pickle.py", line 748, in save_global

    (obj, module, name))

pickle.PicklingError: Can't pickle <class '__main__.busybox.box'>: it's
not found as __main__.busybox.box

>>> 

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081208/d2ff7e4e/attachment.htm>

From roman.yakovenko at gmail.com  Mon Dec  8 19:34:45 2008
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Mon, 8 Dec 2008 20:34:45 +0200
Subject: [C++-sig] Getting address of wrapped instance?
In-Reply-To: <493D4E8D.7000700@sympatico.ca>
References: <46613.146.50.22.50.1228746230.squirrel@www.assumetheposition.nl>
	<493D3946.3030801@sympatico.ca>
	<46846.146.50.22.50.1228749829.squirrel@www.assumetheposition.nl>
	<200812081733.12323.meine@informatik.uni-hamburg.de>
	<493D4E8D.7000700@sympatico.ca>
Message-ID: <7465b6170812081034y4a67f08aobcfc69b41738d209@mail.gmail.com>

On Mon, Dec 8, 2008 at 6:42 PM, Stefan Seefeld <seefeld at sympatico.ca> wrote:
> Hans Meine wrote:
>>
>> How about adding an id() method (returning this) to the C++ class and
>> exporting that?
>>
>
> Well, this 'id' isn't really a property of the wrapped type / object, but
> the wrapper. So, I think there is little sense in adding such a tautological
> 'id' function to the class itself. A free-standing
>
> type const *id(type const *t) { return t;}
>
> function would be more clean, IMO.

I think he wants size_t as return type:

template< class type >
size_t id( const type* t ){ return size_t( t ); }

class_< XXX >(....)
    .def( "id", &id<XXX> );

Py++ has this and other functionality to allow better experience,
while exposing "C" libraries:

http://language-binding.net/pyplusplus/documentation/ctypes/ctypes_integration.html
http://language-binding.net/pyplusplus/documentation/ctypes/this_and_sizeof.html

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

From roman.yakovenko at gmail.com  Mon Dec  8 20:02:41 2008
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Mon, 8 Dec 2008 21:02:41 +0200
Subject: [C++-sig] Pickle an enum from c++
In-Reply-To: <32490DFF7774554A85D65D23A9F0F93809A51DA7@chiex01>
References: <32490DFF7774554A85D65D23A9F0F93809A51DA7@chiex01>
Message-ID: <7465b6170812081102m3236d3d5w5b7634abbe77a8b@mail.gmail.com>

2008/12/8 Matthew Scouten (TT) <Matthew.Scouten at tradingtechnologies.com>:
> From the lack of response I assume that no one has any clever ideas to make
> an enum pickleable.  Thank you to anyone who put thought into this.

:-).

I thought about work around: you can define your enums in Python. May
be youcan add this functionality from Python?

Also did you see the following discussion?
http://mail.python.org/pipermail/cplusplus-sig/2006-November/011372.html

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

From meine at informatik.uni-hamburg.de  Tue Dec  9 13:40:17 2008
From: meine at informatik.uni-hamburg.de (Hans Meine)
Date: Tue, 9 Dec 2008 13:40:17 +0100
Subject: [C++-sig] Pickle an enum from c++
In-Reply-To: <7465b6170812081102m3236d3d5w5b7634abbe77a8b@mail.gmail.com>
References: <32490DFF7774554A85D65D23A9F0F93809A51DA7@chiex01>
	<7465b6170812081102m3236d3d5w5b7634abbe77a8b@mail.gmail.com>
Message-ID: <200812091340.17902.meine@informatik.uni-hamburg.de>

On Monday 08 December 2008 20:02:41 Roman Yakovenko wrote:
> 2008/12/8 Matthew Scouten (TT) <Matthew.Scouten at tradingtechnologies.com>:
> > From the lack of response I assume that no one has any clever ideas to
> > make an enum pickleable.  Thank you to anyone who put thought into this.
> >
> :-).
>
> I thought about work around: you can define your enums in Python. May
> be youcan add this functionality from Python?
>
> Also did you see the following discussion?
> http://mail.python.org/pipermail/cplusplus-sig/2006-November/011372.html

Oh, this is continued here:
http://mail.python.org/pipermail/cplusplus-sig/2006-November/011376.html

Looks like there was a solution in 2006; Ralf even wanted to have a test case 
for check-in.  Ralf, did you receive stuff from Shashank back then?  (Or did 
you wait until today with committing the fix?)

Greetings,
  Hans

From rwgk at yahoo.com  Tue Dec  9 17:02:36 2008
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Tue, 9 Dec 2008 08:02:36 -0800 (PST)
Subject: [C++-sig] Pickle an enum from c++
References: <32490DFF7774554A85D65D23A9F0F93809A51DA7@chiex01>
	<7465b6170812081102m3236d3d5w5b7634abbe77a8b@mail.gmail.com>
	<200812091340.17902.meine@informatik.uni-hamburg.de>
Message-ID: <947038.27443.qm@web111408.mail.gq1.yahoo.com>

Indeed (I had forgotten this already...):

------------------------------------------------------------------------
r36256 | rwgk | 2006-12-03 12:43:48 -0800 (Sun, 03 Dec 2006) | 2 lines

fixes to support pickling of enums (by Shashank Bapat)

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


And there is this test (boost/libs/python/test/enum.py):

# pickling of enums only works with Python 2.3 or higher
exercise_pickling = '''
>>> import pickle
>>> p = pickle.dumps(color.green, pickle.HIGHEST_PROTOCOL)
>>> l = pickle.loads(p)
>>> identity(l)
enum_ext.color.green
'''

So this is supposed to work.



----- Original Message ----
From: Hans Meine <meine at informatik.uni-hamburg.de>
To: Development of Python/C++ integration <cplusplus-sig at python.org>
Cc: Ralf W. Grosse-Kunstleve <rwgk at yahoo.com>
Sent: Tuesday, December 9, 2008 4:40:17 AM
Subject: Re: [C++-sig] Pickle an enum from c++

On Monday 08 December 2008 20:02:41 Roman Yakovenko wrote:
> 2008/12/8 Matthew Scouten (TT) <Matthew.Scouten at tradingtechnologies.com>:
> > From the lack of response I assume that no one has any clever ideas to
> > make an enum pickleable.  Thank you to anyone who put thought into this.
> >
> :-).
>
> I thought about work around: you can define your enums in Python. May
> be youcan add this functionality from Python?
>
> Also did you see the following discussion?
> http://mail.python.org/pipermail/cplusplus-sig/2006-November/011372.html

Oh, this is continued here:
http://mail.python.org/pipermail/cplusplus-sig/2006-November/011376.html

Looks like there was a solution in 2006; Ralf even wanted to have a test case 
for check-in.  Ralf, did you receive stuff from Shashank back then?  (Or did 
you wait until today with committing the fix?)

Greetings,
  Hans


From pyplusplus at assumetheposition.nl  Tue Dec  9 20:10:37 2008
From: pyplusplus at assumetheposition.nl (Paul Melis)
Date: Tue, 09 Dec 2008 20:10:37 +0100
Subject: [C++-sig] Custom from-python converter (almost working)
In-Reply-To: <2754.79036.qm@web111402.mail.gq1.yahoo.com>
References: <34319.146.50.22.50.1228480763.squirrel@www.assumetheposition.nl>	<190802.95882.qm@web111403.mail.gq1.yahoo.com>	<48434.146.50.22.50.1228728868.squirrel@www.assumetheposition.nl>	<49804.146.50.22.50.1228729078.squirrel@www.assumetheposition.nl>
	<2754.79036.qm@web111402.mail.gq1.yahoo.com>
Message-ID: <493EC2AD.4040603@assumetheposition.nl>

Ralf W. Grosse-Kunstleve wrote:
>> In SWIG it is possible to simply define an extra class method that takes a
>> PyObject* and perform any conversion you want in that method (although
>> this means having to add that extra method for all cases where the
>> non-const pointer is used). Would something similar work in Boost.Python,
>> e.g.
>>
>> class C
>> {
>> public:
>>     void set_data(data* d);
>> };
>>
>> void
>> extra_method(C& self, PyObject *obj)
>> {
>>     // if obj is list of tuples, convert to
>>     // data* and
>>     // call self.set_data(d)
>> }
>>
>> class_<C>("C")
>>   .def("set_data", &C::set_data)
>>   .def("set_data", &C::extra_method)
>> ;
>>     
>
> Yes, this should work if you use boost::python::object:
>
>   void
>   extra_method(C& self, boost::python::object obj)
>
> The conversion code in the body of this function will probably be
> similar to what you see in the container_conversion.h file.
>
> You could also use
>
>   void
>   extra_method(C& self, boost::python::list list_of_tuples)
>
> which will save you one manual conversion step.
>   
Hey, that's really nice! I'll try that...

Thanks,
Paul


From pyplusplus at assumetheposition.nl  Tue Dec  9 20:11:20 2008
From: pyplusplus at assumetheposition.nl (Paul Melis)
Date: Tue, 09 Dec 2008 20:11:20 +0100
Subject: [C++-sig] Getting address of wrapped instance?
In-Reply-To: <7465b6170812081034y4a67f08aobcfc69b41738d209@mail.gmail.com>
References: <46613.146.50.22.50.1228746230.squirrel@www.assumetheposition.nl>	<493D3946.3030801@sympatico.ca>	<46846.146.50.22.50.1228749829.squirrel@www.assumetheposition.nl>	<200812081733.12323.meine@informatik.uni-hamburg.de>	<493D4E8D.7000700@sympatico.ca>
	<7465b6170812081034y4a67f08aobcfc69b41738d209@mail.gmail.com>
Message-ID: <493EC2D8.7070001@assumetheposition.nl>

Roman Yakovenko wrote:
> On Mon, Dec 8, 2008 at 6:42 PM, Stefan Seefeld <seefeld at sympatico.ca> wrote:
>   
>> Hans Meine wrote:
>>     
>>> How about adding an id() method (returning this) to the C++ class and
>>> exporting that?
>>>
>>>       
>> Well, this 'id' isn't really a property of the wrapped type / object, but
>> the wrapper. So, I think there is little sense in adding such a tautological
>> 'id' function to the class itself. A free-standing
>>
>> type const *id(type const *t) { return t;}
>>
>> function would be more clean, IMO.
>>     
>
> I think he wants size_t as return type:
>
> template< class type >
> size_t id( const type* t ){ return size_t( t ); }
>
> class_< XXX >(....)
>     .def( "id", &id<XXX> );
>
>   
This is indeed what I was looking for, thanks!

Paul
> Py++ has this and other functionality to allow better experience,
> while exposing "C" libraries:
>
> http://language-binding.net/pyplusplus/documentation/ctypes/ctypes_integration.html
> http://language-binding.net/pyplusplus/documentation/ctypes/this_and_sizeof.html
>
>   


From jeremie.delaitre at technogerma.fr  Wed Dec 10 17:08:55 2008
From: jeremie.delaitre at technogerma.fr (=?ISO-8859-1?Q?J=E9r=E9mie_Delaitre?=)
Date: Wed, 10 Dec 2008 17:08:55 +0100
Subject: [C++-sig] Dynamic resolution of members and methods
Message-ID: <493FE997.2050005@technogerma.fr>

Hello,

I have a C++ class that use a custom property system.
This property system allows me to add and remove properties at runtime.

I want to expose this class into Python.

The problem is: I want "a dynamic resolution" of the properties in Python.

For example:

# import my python extension
import myModule

# get an instance of my C++ class
t = myModule.newInstance()

# access the property 'myProperty'
print(t.myProperty)

When I access the property, I want python to call a "callback" of mine 
(written
in the C++ extension) which will check if the requested property exists, and
if it does, return the corresponding value, otherwise generate an exception.

By looking at the python C API, I did not find a way to achieve this. It 
seems
that declaring a new type requires to have static arrays for members and 
methods
(yes, my property system allows me to declare new methods too, so I want 
to expose
them with a similar system).

Any idea how I could resolve this (with Python3.0) ?

Regards,

J?r?mie Delaitre



From amohr at pixar.com  Wed Dec 10 18:55:48 2008
From: amohr at pixar.com (Alex Mohr)
Date: Wed, 10 Dec 2008 09:55:48 -0800
Subject: [C++-sig] Dynamic resolution of members and methods
In-Reply-To: <493FE997.2050005@technogerma.fr>
References: <493FE997.2050005@technogerma.fr>
Message-ID: <494002A4.5070101@pixar.com>

Override the python special methods __getattr__ or __getattribute__.

http://docs.python.org/reference/datamodel.html#attribute-access

Alex

J?r?mie Delaitre wrote:
> Hello,
> 
> I have a C++ class that use a custom property system.
> This property system allows me to add and remove properties at runtime.
> 
> I want to expose this class into Python.
> 
> The problem is: I want "a dynamic resolution" of the properties in Python.
> 
> For example:
> 
> # import my python extension
> import myModule
> 
> # get an instance of my C++ class
> t = myModule.newInstance()
> 
> # access the property 'myProperty'
> print(t.myProperty)
> 
> When I access the property, I want python to call a "callback" of mine 
> (written
> in the C++ extension) which will check if the requested property exists, 
> and
> if it does, return the corresponding value, otherwise generate an 
> exception.
> 
> By looking at the python C API, I did not find a way to achieve this. It 
> seems
> that declaring a new type requires to have static arrays for members and 
> methods
> (yes, my property system allows me to declare new methods too, so I want 
> to expose
> them with a similar system).
> 
> Any idea how I could resolve this (with Python3.0) ?
> 
> Regards,
> 
> J?r?mie Delaitre
> 
> 
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig


From luca.sbardella at gmail.com  Thu Dec 11 16:29:34 2008
From: luca.sbardella at gmail.com (Luca Sbardella)
Date: Thu, 11 Dec 2008 15:29:34 +0000
Subject: [C++-sig] conversion problem
Message-ID: <8315652a0812110729q7bb7fddfo8288e6db7a665ec9@mail.gmail.com>

My authomatic conversion function from std::pair does not work when one of
the pair type is a boost::python::object.

Template for conversion

template<class K, class T>
struct pair_to_tuple  {
typedef pair_to_tuple<K,T> converter;
typedef std::pair<K,T> ctype;

static PyObject* convert(ctype const& v)  {
using namespace boost::python;
return incref(make_tuple(v.first,v.second).ptr());
}
static void register_to_python()  {
boost::python::to_python_converter<ctype,converter>();
}
     };




using namespace boost::python;
pair_to_tuple<double,object>::register_to_python();           <---------
FAILED
pair_to_tuple<double,std::string>::register_to_python();     <--------- OK

The error I get is
*error C2668: 'boost::python::make_tuple' : ambiguous call to overloaded
function
could be 'boost::python::tuple boost::python::make_tuple<_Ty1,_Ty2>(const A0
&,const A1 &)'**
or       'boost::tuples::tuple<T0,T1,T2,T3,T4,T5,T6,T7,T8,T9>
boost::tuples::make_tuple<_Ty1,_Ty2>(const T0 &,const T1 &)' [found using
argument-dependent lookup]
*
What I'm I doing wrong?*
*
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081211/2c5bd61a/attachment.htm>

From paniq at paniq.org  Thu Dec 11 17:18:26 2008
From: paniq at paniq.org (Leonard Ritter)
Date: Thu, 11 Dec 2008 17:18:26 +0100
Subject: [C++-sig] preventing boost::python::throw_error_already_set
Message-ID: <32c88000812110818q21289c79i94a42ca8a0b6f09c@mail.gmail.com>

hi everyone,

i'm having a problem where a boost::python::throw_error_already_set is being
thrown and causes a segfault (linux). unfortunately the error happens from
within python, when reaching the end of an iteration through a vector array,
which was exported using the vector indexing suite.

is there any possibility to catch the previous exception?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081211/888bce5f/attachment.htm>

From seefeld at sympatico.ca  Thu Dec 11 17:32:49 2008
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Thu, 11 Dec 2008 11:32:49 -0500
Subject: [C++-sig] preventing boost::python::throw_error_already_set
In-Reply-To: <32c88000812110818q21289c79i94a42ca8a0b6f09c@mail.gmail.com>
References: <32c88000812110818q21289c79i94a42ca8a0b6f09c@mail.gmail.com>
Message-ID: <494140B1.1040200@sympatico.ca>

Leonard Ritter wrote:
> hi everyone,
>
> i'm having a problem where a boost::python::throw_error_already_set is 
> being thrown and causes a segfault (linux). unfortunately the error 
> happens from within python, when reaching the end of an iteration 
> through a vector array, which was exported using the vector indexing 
> suite.
>
> is there any possibility to catch the previous exception?

That's hard to tell without more context: Who is doing the iteration, 
for example ?

Regards,
       Stefan

-- 

      ...ich hab' noch einen Koffer in Berlin...


From sipickles at hotmail.com  Fri Dec 12 10:00:11 2008
From: sipickles at hotmail.com (Simon Pickles)
Date: Fri, 12 Dec 2008 09:00:11 +0000
Subject: [C++-sig] Need help with inheritance problem
Message-ID: <BAY114-DAV12963053F217117B19C267BEF90@phx.gbl>

Hi,

I am trying to expose part of the bullet dynamics library to my python code.

I simply need to expose a few accessors of a type btVector3. This 
inherits from a class which is itself derived. I seem to be struggling 
to get this right.

Here is my attempt:

BOOST_PYTHON_MODULE(HybridZone)
{
    using namespace boost::python;

    class_<btQuadWordStorage>("btQuadWordStorage")
        ;
       
    // I am trying to expose getX()
    class_<btQuadWord, bases<btQuadWordStorage> >("btQuadWord")
        .def("getX", &btQuadWord::getX, return_internal_reference<1, 
with_custodian_and_ward<1, 2> >());
        //.add_property("x", &btQuadWord::x) // doesn't compile either
        ;
       
    class_<btVector3, bases<btQuadWord> >("btVector3", init<float, 
float, float>())
        ;
}

You can see my structure is an attempt to mimic the inheritance diagram 
found here

http://www.continuousphysics.com/Bullet/BulletFull/classbtVector3.html

I am trying to expose btVector3::getX(), inherited from btQuadWord

Thanks for your time and advice.

Simon

ps: heres the error message, on gcc4.2.4, Ubuntu 8.04:

...updating 3 targets...
gcc.compile.c++ bin/gcc-4.2.4/debug/HybridZone.o
/home/simon/Source/boost_1_37_0/boost/python/object/make_instance.hpp: 
In static member function \u2018static PyObject* 
boost::python::objects::make_instance_impl<T, Holder, 
Derived>::execute(Arg&) [with Arg = float*, T = float, Holder = 
boost::python::objects::pointer_holder<float*, float>, Derived = 
boost::python::objects::make_ptr_instance<float, 
boost::python::objects::pointer_holder<float*, float> >]\u2019:
/home/simon/Source/boost_1_37_0/boost/python/to_python_indirect.hpp:106:   
instantiated from \u2018static PyObject* 
boost::python::detail::make_reference_holder::execute(T*) [with T = 
float]\u2019
/home/simon/Source/boost_1_37_0/boost/python/to_python_indirect.hpp:70:   
instantiated from \u2018PyObject* boost::python::to_python_indirect<T, 
MakeHolder>::execute(const U&, mpl_::false_) const [with U = float, T = 
const float&, MakeHolder = 
boost::python::detail::make_reference_holder]\u2019
/home/simon/Source/boost_1_37_0/boost/python/to_python_indirect.hpp:41:   
instantiated from \u2018PyObject* boost::python::to_python_indirect<T, 
MakeHolder>::operator()(const U&) const [with U = float, T = const 
float&, MakeHolder = boost::python::detail::make_reference_holder]\u2019
/home/simon/Source/boost_1_37_0/boost/python/detail/invoke.hpp:88:   
instantiated from \u2018PyObject* 
boost::python::detail::invoke(boost::python::detail::invoke_tag_<false, 
true>, const RC&, F&, TC&) [with RC = 
boost::python::to_python_indirect<const float&, 
boost::python::detail::make_reference_holder>, F = const btScalar& 
(btQuadWord::*)()const, TC = 
boost::python::arg_from_python<btQuadWord&>]\u2019
/home/simon/Source/boost_1_37_0/boost/python/detail/caller.hpp:223:   
instantiated from \u2018PyObject* 
boost::python::detail::caller_arity<1u>::impl<F, Policies, 
Sig>::operator()(PyObject*, PyObject*) [with F = const btScalar& 
(btQuadWord::*)()const, Policies = 
boost::python::return_internal_reference<1u, 
boost::python::with_custodian_and_ward<1u, 2u, 
boost::python::default_call_policies> >, Sig = boost::mpl::vector2<const 
btScalar&, btQuadWord&>]\u2019
/home/simon/Source/boost_1_37_0/boost/python/object/py_function.hpp:38:   
instantiated from \u2018PyObject* 
boost::python::objects::caller_py_function_impl<Caller>::operator()(PyObject*, 
PyObject*) [with Caller = boost::python::detail::caller<const btScalar& 
(btQuadWord::*)()const, boost::python::return_internal_reference<1u, 
boost::python::with_custodian_and_ward<1u, 2u, 
boost::python::default_call_policies> >, boost::mpl::vector2<const 
btScalar&, btQuadWord&> >]\u2019
HybridZone.cpp:88:   instantiated from here
/home/simon/Source/boost_1_37_0/boost/python/object/make_instance.hpp:24: 
error: invalid application of \u2018sizeof\u2019 to incomplete type 
\u2018boost::STATIC_ASSERTION_FAILURE<false>\u2019

    "g++"  -ftemplate-depth-128 -O0 -fno-inline -Wall -g -fPIC  
-DBOOST_DATE_TIME_DYN_LINK=1 -DDATE_TIME_INLINE  
-I"/home/simon/Source/boost_1_37_0" 
-I"/home/simon/Source/bullet-2.72/src" -I"/usr/local/include/python2.5" 
-c -o "bin/gcc-4.2.4/debug/HybridZone.o" "HybridZone.cpp"

...failed gcc.compile.c++ bin/gcc-4.2.4/debug/HybridZone.o...
...skipped <pbin/gcc-4.2.4/debug>HybridZone.so for lack of 
<pbin/gcc-4.2.4/debug>HybridZone.o...
...skipped <p.>HybridZone.so for lack of 
<pbin/gcc-4.2.4/debug>HybridZone.so...
...failed updating 1 target...
...skipped 2 targets...

-- 
Linux Counter: User# 424693 




From nico_ml at mgdesign.org  Fri Dec 12 10:48:35 2008
From: nico_ml at mgdesign.org (Nicolas Lelong)
Date: Fri, 12 Dec 2008 10:48:35 +0100
Subject: [C++-sig] Bug and patch for boost.python
	withenable_shared_from_this
References: <229148.259.qm@web31108.mail.mud.yahoo.com>
Message-ID: <006501c95c3e$cd570610$d100a8c0@nicopc>

Hi,

> Could you at least ask Peter Dimov if he thinks the
> dont_enable_shared_from_this patch could be generally useful?
> Maybe he's OK adding it, then it wouldn't be a big deal
> fixing Boost.Python, which would probably prevent a lot of
> confusion and lost time in the future.

I stumbled once again on this problem, so I took some time to bother Peter Dimov with the patch suggestion. Here, is a copy of our exchange :
  Hello Peter,

  I wanted to contact you about a problem that was raised, on february, on c++-sig mailing list about a problem happening when wrapping classes derived from 'enable_shared_from_this' using boost.python library.

  The problem is briefly described in this post - including a patch to boost/shared_ptr.hpp that would allow to fix the problem.
  http://mail.python.org/pipermail/cplusplus-sig/2008-February/012973.html

  The original poster was encouraged by Ralf W. Grosse-Kunstleve to contact you about this, but we never heard from him on the list (http://mail.python.org/pipermail/cplusplus-sig/2008-February/013003.html).

  To make it short, boost.python creates shared_ptr objects, holding the wrapped c++ objects, with a custom deleter managing the python object reference count. This leads to something like that :

  #include <boost/enable_shared_from_this.hpp>
  namespace {
  class A : public boost::enable_shared_from_this<A>
    {
    public:
      ~A() {};
    };
    void my_deleter(void*)
    {
    }
  };
  BOOST_AUTO_TEST_CASE( test_enable_shared_from_this )
  {
    boost::shared_ptr<A> a( new A );
    {
      boost::shared_ptr<A> ater = boost::shared_ptr<A>( a.get(), my_deleter );
      // OP patch proposal :: boost::shared_ptr<A> ater = boost::shared_ptr<A>( a.get(), my_deleter
      //                        , boost::dont_enable_shared_from_this() );
      BOOST_CHECK( a == ater );
    }
    boost::shared_ptr<A> abug = a->shared_from_this(); // this throws bad_weak_ptr
  }

  Did you hear about this problem ? Do you think this patch is viable and could be applied in boost trunk ? Do you happen to have another idea of workaround that could be applied in client code and/or in shared_ptr ?

  Thanks for your time,
  best regards,

  Nicolas Lelong
  mgdesign
Followed by Peter's answer that lead to Trac ticket #2584 :
  Yes, I am aware of this problem. It is possible today to emulate the 
  proposed functionality by using the aliasing constructor, but I think that a 
  more proper fix for the issue is to make enable_shared_from_this more 
  intelligent so that it doesn't reinitialize itself in such scenarios.

  I'll appreciate it if you file a Trac ticket for this and assign it to me. 
  I've a few pending enable_shared_from_this issues anyway, and I'll try to 
  find the time to address them before the next release ships.

  --
  Peter Dimov
  http://www.pdplayer.com 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081212/4c26663e/attachment-0001.htm>

From luca.sbardella at gmail.com  Fri Dec 12 12:54:20 2008
From: luca.sbardella at gmail.com (Luca Sbardella)
Date: Fri, 12 Dec 2008 11:54:20 +0000
Subject: [C++-sig] conversion problem
In-Reply-To: <8315652a0812110729q7bb7fddfo8288e6db7a665ec9@mail.gmail.com>
References: <8315652a0812110729q7bb7fddfo8288e6db7a665ec9@mail.gmail.com>
Message-ID: <8315652a0812120354y38388043o68e8a4dbd84d9311@mail.gmail.com>

I substituted the convert method in pair_to_tuple with the following

     static PyObject* convert(ctype const& v)  {
            using namespace boost::python;
            list values;
            values.append(v.first);
            values.append(v.second);
            return incref(tuple(values).ptr());
        }

And this works in all situations. However, I admit it is not a great
solution.

2008/12/11 Luca Sbardella <luca.sbardella at gmail.com>

> My authomatic conversion function from std::pair does not work when one of
> the pair type is a boost::python::object.
>
> Template for conversion
>
> template<class K, class T>
> struct pair_to_tuple  {
> typedef pair_to_tuple<K,T> converter;
> typedef std::pair<K,T> ctype;
>
> static PyObject* convert(ctype const& v)  {
> using namespace boost::python;
> return incref(make_tuple(v.first,v.second).ptr());
> }
> static void register_to_python()  {
> boost::python::to_python_converter<ctype,converter>();
> }
>      };
>
>
>
>
> using namespace boost::python;
> pair_to_tuple<double,object>::register_to_python();           <---------
> FAILED
> pair_to_tuple<double,std::string>::register_to_python();     <--------- OK
>
> The error I get is
> *error C2668: 'boost::python::make_tuple' : ambiguous call to overloaded
> function
> could be 'boost::python::tuple boost::python::make_tuple<_Ty1,_Ty2>(const
> A0 &,const A1 &)'**
> or       'boost::tuples::tuple<T0,T1,T2,T3,T4,T5,T6,T7,T8,T9>
> boost::tuples::make_tuple<_Ty1,_Ty2>(const T0 &,const T1 &)' [found using
> argument-dependent lookup]
> *
> What I'm I doing wrong?*
> *
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081212/2f8f281c/attachment.htm>

From jeremie.delaitre at technogerma.fr  Fri Dec 12 13:02:00 2008
From: jeremie.delaitre at technogerma.fr (=?ISO-8859-1?Q?J=E9r=E9mie_Delaitre?=)
Date: Fri, 12 Dec 2008 13:02:00 +0100
Subject: [C++-sig] Dynamic resolution of members and methods
In-Reply-To: <494002A4.5070101@pixar.com>
References: <493FE997.2050005@technogerma.fr> <494002A4.5070101@pixar.com>
Message-ID: <494252B8.4040608@technogerma.fr>


It works ! Thanks a lot.

J?r?mie

Alex Mohr a ?crit :
> Override the python special methods __getattr__ or __getattribute__.
>
> http://docs.python.org/reference/datamodel.html#attribute-access
>
> Alex
>
> J?r?mie Delaitre wrote:
>> Hello,
>>
>> I have a C++ class that use a custom property system.
>> This property system allows me to add and remove properties at runtime.
>>
>> I want to expose this class into Python.
>>
>> The problem is: I want "a dynamic resolution" of the properties in 
>> Python.
>>
>> For example:
>>
>> # import my python extension
>> import myModule
>>
>> # get an instance of my C++ class
>> t = myModule.newInstance()
>>
>> # access the property 'myProperty'
>> print(t.myProperty)
>>
>> When I access the property, I want python to call a "callback" of 
>> mine (written
>> in the C++ extension) which will check if the requested property 
>> exists, and
>> if it does, return the corresponding value, otherwise generate an 
>> exception.
>>
>> By looking at the python C API, I did not find a way to achieve this. 
>> It seems
>> that declaring a new type requires to have static arrays for members 
>> and methods
>> (yes, my property system allows me to declare new methods too, so I 
>> want to expose
>> them with a similar system).
>>
>> Any idea how I could resolve this (with Python3.0) ?
>>
>> Regards,
>>
>> J?r?mie Delaitre
>>
>>
>> _______________________________________________
>> Cplusplus-sig mailing list
>> Cplusplus-sig at python.org
>> http://mail.python.org/mailman/listinfo/cplusplus-sig
>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>



From meine at informatik.uni-hamburg.de  Fri Dec 12 13:23:04 2008
From: meine at informatik.uni-hamburg.de (Hans Meine)
Date: Fri, 12 Dec 2008 13:23:04 +0100
Subject: [C++-sig] conversion problem
In-Reply-To: <8315652a0812110729q7bb7fddfo8288e6db7a665ec9@mail.gmail.com>
References: <8315652a0812110729q7bb7fddfo8288e6db7a665ec9@mail.gmail.com>
Message-ID: <200812121323.05522.meine@informatik.uni-hamburg.de>

On Thursday 11 December 2008 16:29:34 Luca Sbardella wrote:
> The error I get is
> *error C2668: 'boost::python::make_tuple' : ambiguous call to overloaded
> function
> could be 'boost::python::tuple boost::python::make_tuple<_Ty1,_Ty2>(const
> A0 &,const A1 &)'**
> or       'boost::tuples::tuple<T0,T1,T2,T3,T4,T5,T6,T7,T8,T9>
> boost::tuples::make_tuple<_Ty1,_Ty2>(const T0 &,const T1 &)' [found using
> argument-dependent lookup]

Did you try explicitly writing boost::python::make_tuple instead of just 
make_tuple?  -> The above error msg. shows that the second function is from a 
completely different namespace (boost::tuples).

That should be more proper than your list-solution.

Greetings,
  Hans


From renatox at gmail.com  Fri Dec 12 13:48:59 2008
From: renatox at gmail.com (Renato Araujo)
Date: Fri, 12 Dec 2008 09:48:59 -0300
Subject: [C++-sig] take ownership in a call policy
Message-ID: <95291a80812120448u502fd079ja803c734b4b15040@mail.gmail.com>

Hi guys,

I'm creating a 2 new call_policy to my functions where I need take and
give back the ownership of my object. I would like reproduce this in
my precall or postcall function policy functions.

http://www.boost.org/doc/libs/1_36_0/libs/python/doc/v2/faq.html#ownership

How I can do this? How I can get a ref to  auto_ptr or shared_ptr from
my PyObject?



Thanks
-- 
Renato Araujo Oliveira Filho

From luca.sbardella at gmail.com  Fri Dec 12 15:11:04 2008
From: luca.sbardella at gmail.com (Luca Sbardella)
Date: Fri, 12 Dec 2008 14:11:04 +0000
Subject: [C++-sig] conversion problem
In-Reply-To: <200812121323.05522.meine@informatik.uni-hamburg.de>
References: <8315652a0812110729q7bb7fddfo8288e6db7a665ec9@mail.gmail.com>
	<200812121323.05522.meine@informatik.uni-hamburg.de>
Message-ID: <8315652a0812120611q6136a662i573f67239156e5ae@mail.gmail.com>

you are right, that was silly.
thanks a lot

2008/12/12 Hans Meine <meine at informatik.uni-hamburg.de>

> On Thursday 11 December 2008 16:29:34 Luca Sbardella wrote:
> > The error I get is
> > *error C2668: 'boost::python::make_tuple' : ambiguous call to overloaded
> > function
> > could be 'boost::python::tuple boost::python::make_tuple<_Ty1,_Ty2>(const
> > A0 &,const A1 &)'**
> > or       'boost::tuples::tuple<T0,T1,T2,T3,T4,T5,T6,T7,T8,T9>
> > boost::tuples::make_tuple<_Ty1,_Ty2>(const T0 &,const T1 &)' [found using
> > argument-dependent lookup]
>
> Did you try explicitly writing boost::python::make_tuple instead of just
> make_tuple?  -> The above error msg. shows that the second function is from
> a
> completely different namespace (boost::tuples).
>
> That should be more proper than your list-solution.
>
> Greetings,
>  Hans
>
> _______________________________________________
> 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/20081212/f53f1090/attachment.htm>

From nico_ml at mgdesign.org  Fri Dec 12 15:55:16 2008
From: nico_ml at mgdesign.org (Nicolas Lelong)
Date: Fri, 12 Dec 2008 15:55:16 +0100
Subject: [C++-sig] Bug and patch for
	boost.pythonwithenable_shared_from_this
References: <229148.259.qm@web31108.mail.mud.yahoo.com>
	<006501c95c3e$cd570610$d100a8c0@nicopc>
Message-ID: <17fe01c95c69$a505c4f0$d100a8c0@nicopc>

FWIW, it seems I can't reproduce the problem with current boost svn trunk code, the problem still existed in boost_1_37_0 though. I could not spot the changes made that fixed the problem.

I took a look at the shared_ptr aliasing constructor, and made a patch to my boost_1_37_0 that solves the problem without altering shared_ptr code. All boost.python tests still pass after that. It may be useful to someone (the test case from Chad is attached)

--- python_1_37\converter\shared_ptr_from_python.hpp    2008-12-12 15:49:04.156250000 +0100
+++ python\converter\shared_ptr_from_python.hpp 2008-12-12 14:24:24.781250000 +0100
@@ -45,10 +45,14 @@
         if (data->convertible == source)
             new (storage) shared_ptr<T>();
         else
+        {
+            boost::shared_ptr<void> hold_convertible_ref_count( (void*)0, shared_ptr_deleter(handle<>(borrowed(source))) );
+            // use aliasing constructor
             new (storage) shared_ptr<T>(
-                static_cast<T*>(data->convertible),
-                shared_ptr_deleter(handle<>(borrowed(source)))
+                hold_convertible_ref_count,
+                static_cast<T*>(data->convertible)
                 );
+        }

         data->convertible = storage;
     }

---

to include the test in python test pass :

Index: libs/python/test/Jamfile.v2
===================================================================
--- libs/python/test/Jamfile.v2 (revision 72)
+++ libs/python/test/Jamfile.v2 (working copy)
@@ -75,6 +75,7 @@
 [ bpl-test return_arg ]
 [ bpl-test staticmethod ]
 [ bpl-test shared_ptr ]
+[ bpl-test enable_shared_from_this ]
 [ bpl-test andreas_beyer ]
 [ bpl-test polymorphism ]
 [ bpl-test polymorphism2 ]
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081212/7adda3e8/attachment-0001.htm>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: enable_shared_from_this.cpp
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081212/7adda3e8/attachment-0001.txt>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: enable_shared_from_this.py
Type: application/octet-stream
Size: 600 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081212/7adda3e8/attachment-0001.obj>

From Matthew.Scouten at tradingtechnologies.com  Fri Dec 12 16:55:32 2008
From: Matthew.Scouten at tradingtechnologies.com (Matthew Scouten (TT))
Date: Fri, 12 Dec 2008 09:55:32 -0600
Subject: [C++-sig] Pickle an enum from c++
In-Reply-To: <947038.27443.qm@web111408.mail.gq1.yahoo.com>
References: <32490DFF7774554A85D65D23A9F0F93809A51DA7@chiex01><7465b6170812081102m3236d3d5w5b7634abbe77a8b@mail.gmail.com><200812091340.17902.meine@informatik.uni-hamburg.de>
	<947038.27443.qm@web111408.mail.gq1.yahoo.com>
Message-ID: <32490DFF7774554A85D65D23A9F0F93809AF0663@chiex01>

Is there something I am supposed to do to the enum_ to make it work? If
I try to 'dumps()' my enum I still get the exception I included in my
earlier email.  Funny thing, dumps-ing it with cPickle just gets the
exception, while dumps-ing it with regular pickle also produces a crash
from within the bowels of Python25.dll.

-----Original Message-----
From:
cplusplus-sig-bounces+matthew.scouten=tradingtechnologies.com at python.org
[mailto:cplusplus-sig-bounces+matthew.scouten=tradingtechnologies.com at py
thon.org] On Behalf Of Ralf W. Grosse-Kunstleve
Sent: Tuesday, December 09, 2008 10:03 AM
To: Development of Python/C++ integration
Subject: Re: [C++-sig] Pickle an enum from c++

Indeed (I had forgotten this already...):

------------------------------------------------------------------------
r36256 | rwgk | 2006-12-03 12:43:48 -0800 (Sun, 03 Dec 2006) | 2 lines

fixes to support pickling of enums (by Shashank Bapat)

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


And there is this test (boost/libs/python/test/enum.py):

# pickling of enums only works with Python 2.3 or higher
exercise_pickling = '''
>>> import pickle
>>> p = pickle.dumps(color.green, pickle.HIGHEST_PROTOCOL)
>>> l = pickle.loads(p)
>>> identity(l)
enum_ext.color.green
'''

So this is supposed to work.



----- Original Message ----
From: Hans Meine <meine at informatik.uni-hamburg.de>
To: Development of Python/C++ integration <cplusplus-sig at python.org>
Cc: Ralf W. Grosse-Kunstleve <rwgk at yahoo.com>
Sent: Tuesday, December 9, 2008 4:40:17 AM
Subject: Re: [C++-sig] Pickle an enum from c++

On Monday 08 December 2008 20:02:41 Roman Yakovenko wrote:
> 2008/12/8 Matthew Scouten (TT)
<Matthew.Scouten at tradingtechnologies.com>:
> > From the lack of response I assume that no one has any clever ideas
to
> > make an enum pickleable.  Thank you to anyone who put thought into
this.
> >
> :-).
>
> I thought about work around: you can define your enums in Python. May
> be youcan add this functionality from Python?
>
> Also did you see the following discussion?
>
http://mail.python.org/pipermail/cplusplus-sig/2006-November/011372.html

Oh, this is continued here:
http://mail.python.org/pipermail/cplusplus-sig/2006-November/011376.html

Looks like there was a solution in 2006; Ralf even wanted to have a test
case 
for check-in.  Ralf, did you receive stuff from Shashank back then?  (Or
did 
you wait until today with committing the fix?)

Greetings,
  Hans

_______________________________________________
Cplusplus-sig mailing list
Cplusplus-sig at python.org
http://mail.python.org/mailman/listinfo/cplusplus-sig

From rwgk at yahoo.com  Fri Dec 12 17:29:02 2008
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Fri, 12 Dec 2008 08:29:02 -0800 (PST)
Subject: [C++-sig] Pickle an enum from c++
References: <32490DFF7774554A85D65D23A9F0F93809A51DA7@chiex01><7465b6170812081102m3236d3d5w5b7634abbe77a8b@mail.gmail.com><200812091340.17902.meine@informatik.uni-hamburg.de>
	<947038.27443.qm@web111408.mail.gq1.yahoo.com>
	<32490DFF7774554A85D65D23A9F0F93809AF0663@chiex01>
Message-ID: <714948.24769.qm@web111404.mail.gq1.yahoo.com>

> Is there something I am supposed to do to the enum_ to make it work?

Did you look at thid test?

boost/libs/python/test/enum.cpp

Do you have the .export_values()? (Not sure if that's important; I never use enum.)

Ralf


From Matthew.Scouten at tradingtechnologies.com  Fri Dec 12 19:01:26 2008
From: Matthew.Scouten at tradingtechnologies.com (Matthew Scouten (TT))
Date: Fri, 12 Dec 2008 12:01:26 -0600
Subject: [C++-sig] Pickle an enum from c++
In-Reply-To: <714948.24769.qm@web111404.mail.gq1.yahoo.com>
References: <32490DFF7774554A85D65D23A9F0F93809A51DA7@chiex01><7465b6170812081102m3236d3d5w5b7634abbe77a8b@mail.gmail.com><200812091340.17902.meine@informatik.uni-hamburg.de><947038.27443.qm@web111408.mail.gq1.yahoo.com><32490DFF7774554A85D65D23A9F0F93809AF0663@chiex01>
	<714948.24769.qm@web111404.mail.gq1.yahoo.com>
Message-ID: <32490DFF7774554A85D65D23A9F0F93809AF0748@chiex01>

No, that didn't make a difference.

-----Original Message-----
From:
cplusplus-sig-bounces+matthew.scouten=tradingtechnologies.com at python.org
[mailto:cplusplus-sig-bounces+matthew.scouten=tradingtechnologies.com at py
thon.org] On Behalf Of Ralf W. Grosse-Kunstleve
Sent: Friday, December 12, 2008 10:29 AM
To: Development of Python/C++ integration
Subject: Re: [C++-sig] Pickle an enum from c++

> Is there something I am supposed to do to the enum_ to make it work?

Did you look at thid test?

boost/libs/python/test/enum.cpp

Do you have the .export_values()? (Not sure if that's important; I never
use enum.)

Ralf

_______________________________________________
Cplusplus-sig mailing list
Cplusplus-sig at python.org
http://mail.python.org/mailman/listinfo/cplusplus-sig

From rwgk at yahoo.com  Fri Dec 12 19:23:21 2008
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Fri, 12 Dec 2008 10:23:21 -0800 (PST)
Subject: [C++-sig] Pickle an enum from c++
References: <32490DFF7774554A85D65D23A9F0F93809A51DA7@chiex01><7465b6170812081102m3236d3d5w5b7634abbe77a8b@mail.gmail.com><200812091340.17902.meine@informatik.uni-hamburg.de><947038.27443.qm@web111408.mail.gq1.yahoo.com><32490DFF7774554A85D65D23A9F0F93809AF0663@chiex01>
	<714948.24769.qm@web111404.mail.gq1.yahoo.com>
	<32490DFF7774554A85D65D23A9F0F93809AF0748@chiex01>
Message-ID: <913058.62639.qm@web111402.mail.gq1.yahoo.com>

Could you try if the enum test works with your version of boost?
What version do you have?
I just tried it out (using the current boost svn trunk) and it works for me.
I think it should work with any boost version after Dec 2006.
Note that you have to use pickle.HIGHEST_PROTOCOL as in the test, otherwise
it crashes.




----- Original Message ----
From: Matthew Scouten (TT) <Matthew.Scouten at tradingtechnologies.com>
To: Development of Python/C++ integration <cplusplus-sig at python.org>
Sent: Friday, December 12, 2008 10:01:26 AM
Subject: Re: [C++-sig] Pickle an enum from c++

No, that didn't make a difference.

-----Original Message-----
From:
cplusplus-sig-bounces+matthew.scouten=tradingtechnologies.com at python.org
[mailto:cplusplus-sig-bounces+matthew.scouten=tradingtechnologies.com at py
thon.org] On Behalf Of Ralf W. Grosse-Kunstleve
Sent: Friday, December 12, 2008 10:29 AM
To: Development of Python/C++ integration
Subject: Re: [C++-sig] Pickle an enum from c++

> Is there something I am supposed to do to the enum_ to make it work?

Did you look at thid test?

boost/libs/python/test/enum.cpp

Do you have the .export_values()? (Not sure if that's important; I never
use enum.)

Ralf


From hugo.lima at openbossa.org  Fri Dec 12 20:39:13 2008
From: hugo.lima at openbossa.org (Hugo Lima)
Date: Fri, 12 Dec 2008 16:39:13 -0300
Subject: [C++-sig] take ownership in a call policy
In-Reply-To: <95291a80812120448u502fd079ja803c734b4b15040@mail.gmail.com>
References: <95291a80812120448u502fd079ja803c734b4b15040@mail.gmail.com>
Message-ID: <7ebf1b2a0812121139h3ba149e2rb16ee96c18a701f9@mail.gmail.com>

On Fri, Dec 12, 2008 at 9:48 AM, Renato Araujo <renatox at gmail.com> wrote:
> Hi guys,
>
> I'm creating a 2 new call_policy to my functions where I need take and
> give back the ownership of my object. I would like reproduce this in
> my precall or postcall function policy functions.
>
> http://www.boost.org/doc/libs/1_36_0/libs/python/doc/v2/faq.html#ownership
>
> How I can do this? How I can get a ref to  auto_ptr or shared_ptr from
> my PyObject?

Good question!

>
>
> Thanks
> --
> Renato Araujo Oliveira Filho
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>

From smnemeth at gmail.com  Fri Dec 12 22:13:03 2008
From: smnemeth at gmail.com (Rocketman@JSC)
Date: Fri, 12 Dec 2008 13:13:03 -0800 (PST)
Subject: [C++-sig]  boost::python::throw_error_already_set not in library
Message-ID: <20983772.post@talk.nabble.com>


Seems like this function only gets built into the debug version.  We built
the gcc library for 1.35 and it only appears in the
libboost_python_gcc32-mt-d-1_35 version.

Does anyone know why this does not get built into the non-debug version or
how to make it get built in? 

TIA,

Scott
-- 
View this message in context: http://www.nabble.com/boost%3A%3Apython%3A%3Athrow_error_already_set-not-in-library-tp20983772p20983772.html
Sent from the Python - c++-sig mailing list archive at Nabble.com.


From dave at boostpro.com  Fri Dec 12 23:45:30 2008
From: dave at boostpro.com (David Abrahams)
Date: Fri, 12 Dec 2008 17:45:30 -0500
Subject: [C++-sig] boost::python::throw_error_already_set not in library
In-Reply-To: <20983772.post@talk.nabble.com> (smnemeth@gmail.com's message of
	"Fri, 12 Dec 2008 13:13:03 -0800 (PST)")
References: <20983772.post@talk.nabble.com>
Message-ID: <87hc59au7p.fsf@mcbain.luannocracy.com>


on Fri Dec 12 2008, "Rocketman-AT-JSC" <smnemeth-AT-gmail.com> wrote:

> Seems like this function only gets built into the debug version.  

Are you certain?  It's hard to imagine that's actually the case.

> We built the gcc library for 1.35 and it only appears in the
> libboost_python_gcc32-mt-d-1_35 version.

How did you do the build?

> Does anyone know why this does not get built into the non-debug version or
> how to make it get built in? 

There's nothing different about that function from many of the other
functions that are compiled into the library.  You can see it for
yourself in the source.

-- 
Dave Abrahams
BoostPro Computing
http://www.boostpro.com

From smnemeth at gmail.com  Sat Dec 13 00:08:20 2008
From: smnemeth at gmail.com (Rocketman@JSC)
Date: Fri, 12 Dec 2008 15:08:20 -0800 (PST)
Subject: [C++-sig] boost::python::throw_error_already_set not in library
In-Reply-To: <87hc59au7p.fsf@mcbain.luannocracy.com>
References: <20983772.post@talk.nabble.com>
	<87hc59au7p.fsf@mcbain.luannocracy.com>
Message-ID: <20985178.post@talk.nabble.com>


Yes it's probably user error--but I have no clue why this is happening.  I
did not do the build and the guy doing the build is not an expert boost user
either.  We have two builds one with gcc and one with intel.  Notice the
non-debug versions don't have the symbol.  Here are the results from
objdump:

/ips/fd/analysis2/proto/oss/lib $ ls libboost_python-*
libboost_python-gcc32-mt-1_35.so           libboost_python-gcc32-mt-d.so 
libboost_python-il-mt-1_36.so.1.36.0
libboost_python-gcc32-mt-1_35.so.1.35.0    libboost_python-gcc32-mt.so   
libboost_python-il-mt.a
libboost_python-gcc32-mt-d-1_35.so         libboost_python-il-mt-1_36.a  
libboost_python-il-mt.so
libboost_python-gcc32-mt-d-1_35.so.1.35.0  libboost_python-il-mt-1_36.so

/ips/fd/analysis2/proto/oss/lib $ objdump -t libboost_python-il-mt*.so |
grep throw_error_already_set

/ips/fd/analysis2/proto/oss/lib $ objdump -t libboost_python-gcc32-mt.so |
grep throw_error_already_set

/ips/fd/analysis2/proto/oss/lib $ objdump -t libboost_python-gcc32-mt-d.so |
grep throw_error_already_set
0009cd14 g     F .text  0000007c             
_ZN5boost6python23throw_error_already_setEv

-- 
View this message in context: http://www.nabble.com/boost%3A%3Apython%3A%3Athrow_error_already_set-not-in-library-tp20983772p20985178.html
Sent from the Python - c++-sig mailing list archive at Nabble.com.


From dave at boostpro.com  Sat Dec 13 02:05:23 2008
From: dave at boostpro.com (David Abrahams)
Date: Fri, 12 Dec 2008 20:05:23 -0500
Subject: [C++-sig] boost::python::throw_error_already_set not in library
In-Reply-To: <20985178.post@talk.nabble.com> (smnemeth@gmail.com's message of
	"Fri, 12 Dec 2008 15:08:20 -0800 (PST)")
References: <20983772.post@talk.nabble.com>
	<87hc59au7p.fsf@mcbain.luannocracy.com>
	<20985178.post@talk.nabble.com>
Message-ID: <87oczganqk.fsf@mcbain.luannocracy.com>


on Fri Dec 12 2008, "Rocketman-AT-JSC" <smnemeth-AT-gmail.com> wrote:

> Yes it's probably user error--but I have no clue why this is happening.  I
> did not do the build and the guy doing the build is not an expert boost user
> either.  

Did he use bjam?

You can verify the build by following the Boost.Python tutorial to make
sure that build/test works.  The library you build and the one linked to
by the tutorial should be identical, apart from SONAME and associated
mangling.

> We have two builds one with gcc and one with intel.  Notice the
> non-debug versions don't have the symbol.  Here are the results from
> objdump:
>
> /ips/fd/analysis2/proto/oss/lib $ ls libboost_python-*
> libboost_python-gcc32-mt-1_35.so           libboost_python-gcc32-mt-d.so 
> libboost_python-il-mt-1_36.so.1.36.0
> libboost_python-gcc32-mt-1_35.so.1.35.0    libboost_python-gcc32-mt.so   
> libboost_python-il-mt.a
> libboost_python-gcc32-mt-d-1_35.so         libboost_python-il-mt-1_36.a  
> libboost_python-il-mt.so
> libboost_python-gcc32-mt-d-1_35.so.1.35.0  libboost_python-il-mt-1_36.so
>
> /ips/fd/analysis2/proto/oss/lib $ objdump -t libboost_python-il-mt*.so |
> grep throw_error_already_set
>
> /ips/fd/analysis2/proto/oss/lib $ objdump -t libboost_python-gcc32-mt.so |
> grep throw_error_already_set
>
> /ips/fd/analysis2/proto/oss/lib $ objdump -t libboost_python-gcc32-mt-d.so |
> grep throw_error_already_set
> 0009cd14 g     F .text  0000007c             
> _ZN5boost6python23throw_error_already_setEv

This doesn't make any sense to me.  I don't see the string
"throw_error_already_set" in the grep output.  is this really what came
up on your console?

Anyway, I think you need to use objdump -T/--dynamic-syms, not objdump
-t/--syms.

The non-debug ones don't show /any/ symbols if you use -t.

-- 
Dave Abrahams
BoostPro Computing
http://www.boostpro.com

From dave at boostpro.com  Sat Dec 13 02:08:51 2008
From: dave at boostpro.com (David Abrahams)
Date: Fri, 12 Dec 2008 20:08:51 -0500
Subject: [C++-sig] take ownership in a call policy
In-Reply-To: <95291a80812120448u502fd079ja803c734b4b15040@mail.gmail.com>
	(Renato Araujo's message of "Fri, 12 Dec 2008 09:48:59 -0300")
References: <95291a80812120448u502fd079ja803c734b4b15040@mail.gmail.com>
Message-ID: <87fxksanks.fsf@mcbain.luannocracy.com>


on Fri Dec 12 2008, "Renato Araujo" <renatox-AT-gmail.com> wrote:

> Hi guys,
>
> I'm creating a 2 new call_policy to my functions where I need take and
> give back the ownership of my object. I would like reproduce this in
> my precall or postcall function policy functions.
>
> http://www.boost.org/doc/libs/1_36_0/libs/python/doc/v2/faq.html#ownership
>
> How I can do this? How I can get a ref to  auto_ptr or shared_ptr from
> my PyObject?

I'm sorry, I don't understand the question.  Could you please show the
function signature you're trying to wrap and describe the ownership
semantics required?

Thanks,

-- 
Dave Abrahams
BoostPro Computing
http://www.boostpro.com

From renatox at gmail.com  Sat Dec 13 03:52:37 2008
From: renatox at gmail.com (Renato Araujo)
Date: Fri, 12 Dec 2008 21:52:37 -0500
Subject: [C++-sig] take ownership in a call policy
In-Reply-To: <87fxksanks.fsf@mcbain.luannocracy.com>
References: <95291a80812120448u502fd079ja803c734b4b15040@mail.gmail.com>
	<87fxksanks.fsf@mcbain.luannocracy.com>
Message-ID: <95291a80812121852s713d6d05u5bfb3782447001cd@mail.gmail.com>

Thanks David I will try explain more about my problems with ownership:

I already expose this problem before here the ideas not fit yet.


This is my problem.
attention: I can't change the c++ code;


struct object
{

    object(object *parent=0)
    {
       //if (parent): append myself in parent child list (parent take
ownership of object, parent will delete this object)
    }

    void set_parent(object *parent)
    {
       //the same case of constructor: if parent is null remove old parent
    }

    void add_child(object *child)
    {
       //add child in my children list get ownership
    }

    void remove_child(object *child)
    {
       //remove child from my list, give back the ownership to the child object
    }

    ~object()
    { // destroy all children; }

private:
     list<object*> children;
}

this is a simple example how c++ code works, we have allot of others
functions to get ownership of a object, then is impossible rewrite all
functions like in the boost python FAQ example:
(http://www.boost.org/doc/libs/1_36_0/libs/python/doc/v2/faq.html#ownership)

I thinking about create a call policy to do this for me. One for take
ownership and another to restore ownership.

I'm trying do something like that, I don't know if this is the best
idea, sugestions are welcome:

class<object_wrapper>("object", init< optional<object*>
>()[parent_child_policy_add<2,1>()])
	.def("set_parent", &object::set_parent,
parent_child_policy_add<2,1>()) //this case (index 2 = arg1) = parent
object, (index 1 = self) = child object
        .def("remove_child", &object::remove_child,
parent_child_policy_remove<1,2>() //this case self is the parent and
arg1 is the child


I know how implement a call policy and get access to PyObject from
child and parent. What I don't know is how inform to boost to not
delete c++ object during the python object destructor (in
parent_child_policy_add) because this will be done by the parent, or
inform to boost to delete the c++ object (in
parent_child_policy_remove) because the object no have parent.


On Fri, Dec 12, 2008 at 8:08 PM, David Abrahams <dave at boostpro.com> wrote:
>
> on Fri Dec 12 2008, "Renato Araujo" <renatox-AT-gmail.com> wrote:
>
>> Hi guys,
>>
>> I'm creating a 2 new call_policy to my functions where I need take and
>> give back the ownership of my object. I would like reproduce this in
>> my precall or postcall function policy functions.
>>
>> http://www.boost.org/doc/libs/1_36_0/libs/python/doc/v2/faq.html#ownership
>>
>> How I can do this? How I can get a ref to  auto_ptr or shared_ptr from
>> my PyObject?
>
> I'm sorry, I don't understand the question.  Could you please show the
> function signature you're trying to wrap and describe the ownership
> semantics required?
>
> Thanks,
>
> --
> Dave Abrahams
> BoostPro Computing
> http://www.boostpro.com
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>



-- 
Renato Araujo Oliveira Filho

From dave at boostpro.com  Sun Dec 14 20:47:26 2008
From: dave at boostpro.com (David Abrahams)
Date: Sun, 14 Dec 2008 19:47:26 -0000
Subject: [C++-sig] Passing C++ instance to embedded py
In-Reply-To: <20081120161605.1C4E.FILEOFFSET@gmail.com>
	(fileoffset@gmail.com's message of "Thu, 20 Nov 2008 16:29:49 +1100")
References: <20081120161605.1C4E.FILEOFFSET@gmail.com>
Message-ID: <87k5a2ilo9.fsf@mcbain.luannocracy.com>


on Thu Nov 20 2008, fileoffset <fileoffset-AT-gmail.com> wrote:

> To best explain my problem, here is some code:
>
> struct A
> {
> 	A()
> 	{
> 		mTest = 1;
>
> 		std::cout << "Test: " << mTest;
>
> 		Py_Initialize();
> 		object main_module = import("__main__");
> 		object main_namespace = main_module.attr("__dict__");
>
> 		main_namespace["A_Instance"] = this;
>
> 		object result = exec_file(str("test.py"), main_namespace,
> main_namespace);
>
> 		std::cout << "Test: " << mTest;
> 	}
>
> 	int mTest;
> }
>
> in main.cpp or similar:
>
> BOOST_PYTHON_MODULE(MyModule)
> {
>     class_<A>("A", init<>())
>         .def_readwrite("Test", &A::mTest)
>     ;
> }
>
> test.py:
>
> from MyModule import *
>
> A_Instance.Test = 5
>
>
> Essentially I want to be able to access and modify the properties of an instance of a
> C++ class
> at runtime.
>
> When I try code similar to above, python excepts, as I believe it
> attempts to copy-construct the A class instead of just pass the
> reference (this) so that i can modify it.

Correct.

>
> Is what I want possible? If not what would be a better way to do it?

  main_module.attr("A_Instance") = ptr(this);
  object d = main_module.attr("__dict__");
  object result = exec_file(str("test.py"), d, d);
  main_module.attr("A_Instance") = object(); // don't leave a dangling reference

(http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/callbacks.html#argument_handling)

-- 
Dave Abrahams
BoostPro Computing
http://www.boostpro.com

From paniq at paniq.org  Sun Dec 14 23:22:51 2008
From: paniq at paniq.org (Leonard Ritter)
Date: Sun, 14 Dec 2008 23:22:51 +0100
Subject: [C++-sig] preventing boost::python::throw_error_already_set
In-Reply-To: <494140B1.1040200@sympatico.ca>
References: <32c88000812110818q21289c79i94a42ca8a0b6f09c@mail.gmail.com>
	<494140B1.1040200@sympatico.ca>
Message-ID: <32c88000812141422l51e0ae0blbb074049904af71c@mail.gmail.com>

On Thu, Dec 11, 2008 at 5:32 PM, Stefan Seefeld <seefeld at sympatico.ca>wrote:

> Leonard Ritter wrote:
>
>> hi everyone,
>>
>> i'm having a problem where a boost::python::throw_error_already_set is
>> being thrown and causes a segfault (linux). unfortunately the error happens
>> from within python, when reaching the end of an iteration through a vector
>> array, which was exported using the vector indexing suite.
>>
>> is there any possibility to catch the previous exception?
>>
>
> That's hard to tell without more context: Who is doing the iteration, for
> example ?


python code.

i don't have the feeling that the issue is really related to your library.
some machines do not cause that exception. the problem is that this is very
hard to track for me. this is not meant to sound rough, but i would prefer
if you would not try to find the problem for me, but give me instructions
how i can deal with such problems myself.

this is not the only situation where i get throw_error_already_set
exceptions, and i have no idea how to track them down. is such stuff
unusual? did no one else get these?

thank you.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081214/c30e9657/attachment.htm>

From jeremie.delaitre at technogerma.fr  Mon Dec 15 13:58:29 2008
From: jeremie.delaitre at technogerma.fr (=?ISO-8859-1?Q?J=E9r=E9mie_Delaitre?=)
Date: Mon, 15 Dec 2008 13:58:29 +0100
Subject: [C++-sig] Dynamic resolution of members and methods
In-Reply-To: <494252B8.4040608@technogerma.fr>
References: <493FE997.2050005@technogerma.fr> <494002A4.5070101@pixar.com>
	<494252B8.4040608@technogerma.fr>
Message-ID: <49465475.5090205@technogerma.fr>


Hi,

When defining my new type in C++, I have set the "tp_getattro" field to 
point to a function (say "mytype_getattro").
This function determine dynamically if the requested attribute exists or 
not, and if it does, return its value.
So, it works well for attributes/properties.

But, "mytype_getattro" is also called when, in Python, I do a function 
call 'myobject.myfct(...)'. It allows me to determine
dynamically if the requested function "myfct" is available or not. But 
if it does, I need to return a callable object to allow
python to process the call.

I succeeded to make Python to call a function of mine using a 
PyMethodDef array and using the Py_FindMethod function
to get a callable PyObject*.
But it is not a good solution for at least two reasons :
1 - Py_FindMethod is not supported in Python 3.0
2 - I need to declare each possible function in the array (which is 
impossible has I need to determine them at runtime)

What I have tried to do is to declare only one "dipatch" function in the 
array which will call the right C++ function.
This way, in "mytype_getattro" I do a "return Py_FindMethod(myarray, 
self, "dispatch")". Thus, python call my
dispatching method. The problem is that I need to attach additionnal 
values to the function (like the name of the function
which must be called in reality). In Lua, I used the closure mechanism, 
but I did not success with Python closures.

Any hints ?
Is closure the right approach (If so, how I should use it ?)? Do I need 
to set another field when defining my type ?

Regards,

J?r?mie Delaitre


J?r?mie Delaitre a ?crit :
>
> It works ! Thanks a lot.
>
> J?r?mie
>
> Alex Mohr a ?crit :
>> Override the python special methods __getattr__ or __getattribute__.
>>
>> http://docs.python.org/reference/datamodel.html#attribute-access
>>
>> Alex
>>
>> J?r?mie Delaitre wrote:
>>> Hello,
>>>
>>> I have a C++ class that use a custom property system.
>>> This property system allows me to add and remove properties at runtime.
>>>
>>> I want to expose this class into Python.
>>>
>>> The problem is: I want "a dynamic resolution" of the properties in 
>>> Python.
>>>
>>> For example:
>>>
>>> # import my python extension
>>> import myModule
>>>
>>> # get an instance of my C++ class
>>> t = myModule.newInstance()
>>>
>>> # access the property 'myProperty'
>>> print(t.myProperty)
>>>
>>> When I access the property, I want python to call a "callback" of 
>>> mine (written
>>> in the C++ extension) which will check if the requested property 
>>> exists, and
>>> if it does, return the corresponding value, otherwise generate an 
>>> exception.
>>>
>>> By looking at the python C API, I did not find a way to achieve 
>>> this. It seems
>>> that declaring a new type requires to have static arrays for members 
>>> and methods
>>> (yes, my property system allows me to declare new methods too, so I 
>>> want to expose
>>> them with a similar system).
>>>
>>> Any idea how I could resolve this (with Python3.0) ?
>>>
>>> Regards,
>>>
>>> J?r?mie Delaitre
>>>
>>>
>>> _______________________________________________
>>> Cplusplus-sig mailing list
>>> Cplusplus-sig at python.org
>>> http://mail.python.org/mailman/listinfo/cplusplus-sig
>>
>> _______________________________________________
>> Cplusplus-sig mailing list
>> Cplusplus-sig at python.org
>> http://mail.python.org/mailman/listinfo/cplusplus-sig
>>
>
>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>



From renatox at gmail.com  Mon Dec 15 15:17:16 2008
From: renatox at gmail.com (Renato Araujo)
Date: Mon, 15 Dec 2008 11:17:16 -0300
Subject: [C++-sig] custom constructor problems
Message-ID: <95291a80812150617x7b727926keeccdf8b813ccecc@mail.gmail.com>

Hi again :D,

To solve my old problem, I'm trying create custom constructor which
return a shared_ptr<MyClass> like that code:

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
class my_class_wrapper : my_class, python::wrapper<my_class>
{
......
}

shared_ptr<my_class_wrapper> _create_object(parent *p)
{
    return shared_ptr<my_class_wrapper>(new my_class_wrapper(p),
_my_delete_function);
}

shared_ptr<my_class_wrapper> _create_object2()
{
    return shared_ptr<my_class_wrapper>(new my_class_wrapper(0),
_my_delete_function);
}


class_<my_class_wrapper>("MyClass", python::no_init)
    .def("__init__", python::make_constructor(_create_object))
    .def("__init__", python::make_constructor(_create_object2))
    .def("get_parent", &MyClass:getParent,
return_value_policy<reference_existing_object>());

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

but during my python test, the function get_parent not return the same
python object created in the object constructor. my test is:

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
parent = MyClass()
child = MyClass(parent)

if (child.get_parent() != parent)
   print "FAIL"
else
   print "OK"
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Always I got the "FAIL" message, but if I remove the custom
constructor the test pass. What I doing wrong all my classes inherit
from "python::wrapper<X>". Then I think all classes have implicit
conversion to "shared_ptr" correct?

BR
-- 
Renato Araujo Oliveira Filho

From paniq at paniq.org  Mon Dec 15 16:39:12 2008
From: paniq at paniq.org (Leonard Ritter)
Date: Mon, 15 Dec 2008 16:39:12 +0100
Subject: [C++-sig] preventing boost::python::throw_error_already_set
In-Reply-To: <32c88000812141422l51e0ae0blbb074049904af71c@mail.gmail.com>
References: <32c88000812110818q21289c79i94a42ca8a0b6f09c@mail.gmail.com>
	<494140B1.1040200@sympatico.ca>
	<32c88000812141422l51e0ae0blbb074049904af71c@mail.gmail.com>
Message-ID: <32c88000812150739i35b5ff43n72eb6f07c30d21f6@mail.gmail.com>

On Sun, Dec 14, 2008 at 11:22 PM, Leonard Ritter <paniq at paniq.org> wrote:

> That's hard to tell without more context: Who is doing the iteration, for
>> example ?
>
>
I'm sorry. I was a bit stressed out yesterday. I recognize that you only
want to help, and I should be more patient.

Here is the top part of a stack trace, as it usually occurs:

#0  0xb65593dc in __cxa_allocate_exception () from /usr/lib/libstdc++.so.6
#1  0xb6a85a9e in boost::python::throw_error_already_set ()
   from /usr/lib/libboost_python-gcc42-1_34_1-py25.so.1.34.1
#2  0xb6a88ab1 in boost::python::objects::stop_iteration_error ()
   from /usr/lib/libboost_python-gcc42-1_34_1-py25.so.1.34.1
#3  0xb6ca38c3 in
boost::python::objects::iterator_range<boost::python::return_internal_reference<1u,
boost::python::default_call_policies>,
__gnu_cxx::__normal_iterator<halebopp::wave_range*,
std::vector<halebopp::wave_range, std::allocator<halebopp::wave_range> > >
>::next::operator() (this=0x9931cec,
    self=@0x91fcfcc) at /usr/include/boost/python/object/iterator.hpp:64

This problem occurred first when iterating an array that was exported
through the indexing suite. I do not have anything else that uses
StopIteration, so I can't say if it's really a problem of the indexing
suite. My guess is: no. When I rewrote the offending code to be a simple
get-by-index operation, the next (above) part where an iteration would be
caused inside C++ code created a problem. So somehow I end up with an
unhandled python exception in boost that does not get forwarded - my guess.

Could this in some way relate to assertions (put by assert()) put in the C++
code? Something like: boost catches these assertions, converts them into
exceptions, then iterates a bit, hits a StopIteration and has the previously
unhandled exception still active? I am just wildly guessing here.

Please help!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081215/53e54fa7/attachment.htm>

From troy at resophonic.com  Mon Dec 15 17:09:03 2008
From: troy at resophonic.com (troy d. straszheim)
Date: Mon, 15 Dec 2008 11:09:03 -0500
Subject: [C++-sig] preventing boost::python::throw_error_already_set
In-Reply-To: <32c88000812150739i35b5ff43n72eb6f07c30d21f6@mail.gmail.com>
References: <32c88000812110818q21289c79i94a42ca8a0b6f09c@mail.gmail.com>	<494140B1.1040200@sympatico.ca>	<32c88000812141422l51e0ae0blbb074049904af71c@mail.gmail.com>
	<32c88000812150739i35b5ff43n72eb6f07c30d21f6@mail.gmail.com>
Message-ID: <4946811F.5010505@resophonic.com>

Hey Leonard,

Here's a short doc about running python scripts that load
boost::python bindings under gdb:

   http://software.icecube.wisc.edu/offline-software.trunk/gdb_python.html

In short (i'm guessing this is more or less what you'll need):

   gdb --args /usr/bin/python /path/to/my/python/script.py
   run
     # it runs, throws an exception, but the stacktrace isn't helpful
   catch throw # this only works after the first run
   run
   # jackpot, you're at the point of the throw

-t






Leonard Ritter wrote:
> On Sun, Dec 14, 2008 at 11:22 PM, Leonard Ritter <paniq at paniq.org> wrote:
> 
>> That's hard to tell without more context: Who is doing the iteration, for
>>> example ?
>>
> I'm sorry. I was a bit stressed out yesterday. I recognize that you only
> want to help, and I should be more patient.
> 
> Here is the top part of a stack trace, as it usually occurs:
> 
> #0  0xb65593dc in __cxa_allocate_exception () from /usr/lib/libstdc++.so.6
> #1  0xb6a85a9e in boost::python::throw_error_already_set ()
>    from /usr/lib/libboost_python-gcc42-1_34_1-py25.so.1.34.1
> #2  0xb6a88ab1 in boost::python::objects::stop_iteration_error ()
>    from /usr/lib/libboost_python-gcc42-1_34_1-py25.so.1.34.1
> #3  0xb6ca38c3 in
> boost::python::objects::iterator_range<boost::python::return_internal_reference<1u,
> boost::python::default_call_policies>,
> __gnu_cxx::__normal_iterator<halebopp::wave_range*,
> std::vector<halebopp::wave_range, std::allocator<halebopp::wave_range> > >
>> ::next::operator() (this=0x9931cec,
>     self=@0x91fcfcc) at /usr/include/boost/python/object/iterator.hpp:64
> 
> This problem occurred first when iterating an array that was exported
> through the indexing suite. I do not have anything else that uses
> StopIteration, so I can't say if it's really a problem of the indexing
> suite. My guess is: no. When I rewrote the offending code to be a simple
> get-by-index operation, the next (above) part where an iteration would be
> caused inside C++ code created a problem. So somehow I end up with an
> unhandled python exception in boost that does not get forwarded - my guess.
> 
> Could this in some way relate to assertions (put by assert()) put in the C++
> code? Something like: boost catches these assertions, converts them into
> exceptions, then iterates a bit, hits a StopIteration and has the previously
> unhandled exception still active? I am just wildly guessing here.
> 
> Please help!
> 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig


From seefeld at sympatico.ca  Mon Dec 15 17:27:28 2008
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Mon, 15 Dec 2008 11:27:28 -0500
Subject: [C++-sig] preventing boost::python::throw_error_already_set
In-Reply-To: <4946811F.5010505@resophonic.com>
References: <32c88000812110818q21289c79i94a42ca8a0b6f09c@mail.gmail.com>	<494140B1.1040200@sympatico.ca>	<32c88000812141422l51e0ae0blbb074049904af71c@mail.gmail.com>	<32c88000812150739i35b5ff43n72eb6f07c30d21f6@mail.gmail.com>
	<4946811F.5010505@resophonic.com>
Message-ID: <49468570.4010808@sympatico.ca>

troy d. straszheim wrote:
>
> In short (i'm guessing this is more or less what you'll need):
>
>   gdb --args /usr/bin/python /path/to/my/python/script.py
>   run
>     # it runs, throws an exception, but the stacktrace isn't helpful
>   catch throw # this only works after the first run
>   run
>   # jackpot, you're at the point of the throw

The throw point isn't very interesting in this case: In Python, 
iteration is terminated by means of a 'StopIteration' exception. The 
question is why this exception manages to escape uncaught.

FWIW,
       Stefan

-- 

      ...ich hab' noch einen Koffer in Berlin...


From paniq at paniq.org  Mon Dec 15 17:29:57 2008
From: paniq at paniq.org (Leonard Ritter)
Date: Mon, 15 Dec 2008 17:29:57 +0100
Subject: [C++-sig] preventing boost::python::throw_error_already_set
In-Reply-To: <49468570.4010808@sympatico.ca>
References: <32c88000812110818q21289c79i94a42ca8a0b6f09c@mail.gmail.com>
	<494140B1.1040200@sympatico.ca>
	<32c88000812141422l51e0ae0blbb074049904af71c@mail.gmail.com>
	<32c88000812150739i35b5ff43n72eb6f07c30d21f6@mail.gmail.com>
	<4946811F.5010505@resophonic.com> <49468570.4010808@sympatico.ca>
Message-ID: <32c88000812150829s76142dbj6c0317fda05fc222@mail.gmail.com>

On Mon, Dec 15, 2008 at 5:27 PM, Stefan Seefeld <seefeld at sympatico.ca>wrote:

> The throw point isn't very interesting in this case: In Python, iteration
> is terminated by means of a 'StopIteration' exception. The question is why
> this exception manages to escape uncaught.


i tried both 1.34.1 and 1.35.0 available in my distribution (ubuntu), same
problem. i am fetching through the boost.python code to see if i can find
something interesting.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081215/9c8aa823/attachment.htm>

From paniq at paniq.org  Mon Dec 15 17:33:28 2008
From: paniq at paniq.org (Leonard Ritter)
Date: Mon, 15 Dec 2008 17:33:28 +0100
Subject: [C++-sig] preventing boost::python::throw_error_already_set
In-Reply-To: <32c88000812150829s76142dbj6c0317fda05fc222@mail.gmail.com>
References: <32c88000812110818q21289c79i94a42ca8a0b6f09c@mail.gmail.com>
	<494140B1.1040200@sympatico.ca>
	<32c88000812141422l51e0ae0blbb074049904af71c@mail.gmail.com>
	<32c88000812150739i35b5ff43n72eb6f07c30d21f6@mail.gmail.com>
	<4946811F.5010505@resophonic.com> <49468570.4010808@sympatico.ca>
	<32c88000812150829s76142dbj6c0317fda05fc222@mail.gmail.com>
Message-ID: <32c88000812150833v7bd2c8f4ve9dc483186a41346@mail.gmail.com>

On Mon, Dec 15, 2008 at 5:29 PM, Leonard Ritter <paniq at paniq.org> wrote:

> On Mon, Dec 15, 2008 at 5:27 PM, Stefan Seefeld <seefeld at sympatico.ca>wrote:
>
>> The throw point isn't very interesting in this case: In Python, iteration
>> is terminated by means of a 'StopIteration' exception. The question is why
>> this exception manages to escape uncaught.
>
>
> i tried both 1.34.1 and 1.35.0 available in my distribution (ubuntu), same
> problem. i am fetching through the boost.python code to see if i can find
> something interesting.
>

i should add that this problem does not happen on other machines with the
SAME distribution installed.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081215/ec492788/attachment.htm>

From paniq at paniq.org  Mon Dec 15 17:38:09 2008
From: paniq at paniq.org (Leonard Ritter)
Date: Mon, 15 Dec 2008 17:38:09 +0100
Subject: [C++-sig] preventing boost::python::throw_error_already_set
In-Reply-To: <32c88000812150833v7bd2c8f4ve9dc483186a41346@mail.gmail.com>
References: <32c88000812110818q21289c79i94a42ca8a0b6f09c@mail.gmail.com>
	<494140B1.1040200@sympatico.ca>
	<32c88000812141422l51e0ae0blbb074049904af71c@mail.gmail.com>
	<32c88000812150739i35b5ff43n72eb6f07c30d21f6@mail.gmail.com>
	<4946811F.5010505@resophonic.com> <49468570.4010808@sympatico.ca>
	<32c88000812150829s76142dbj6c0317fda05fc222@mail.gmail.com>
	<32c88000812150833v7bd2c8f4ve9dc483186a41346@mail.gmail.com>
Message-ID: <32c88000812150838j678e598fh1badf15351a6b8bd@mail.gmail.com>

On Mon, Dec 15, 2008 at 5:33 PM, Leonard Ritter <paniq at paniq.org> wrote:

> On Mon, Dec 15, 2008 at 5:29 PM, Leonard Ritter <paniq at paniq.org> wrote:
>
>> On Mon, Dec 15, 2008 at 5:27 PM, Stefan Seefeld <seefeld at sympatico.ca>wrote:
>>
>>> The throw point isn't very interesting in this case: In Python, iteration
>>> is terminated by means of a 'StopIteration' exception. The question is why
>>> this exception manages to escape uncaught.
>>
>>
>> i tried both 1.34.1 and 1.35.0 available in my distribution (ubuntu), same
>> problem. i am fetching through the boost.python code to see if i can find
>> something interesting.
>>
>
> i should add that this problem does not happen on other machines with the
> SAME distribution installed.
>

...and the error generated is a segfault, no standard exception!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081215/2ed221a0/attachment.htm>

From paniq at paniq.org  Mon Dec 15 17:48:30 2008
From: paniq at paniq.org (Leonard Ritter)
Date: Mon, 15 Dec 2008 17:48:30 +0100
Subject: [C++-sig] preventing boost::python::throw_error_already_set
In-Reply-To: <32c88000812150838j678e598fh1badf15351a6b8bd@mail.gmail.com>
References: <32c88000812110818q21289c79i94a42ca8a0b6f09c@mail.gmail.com>
	<494140B1.1040200@sympatico.ca>
	<32c88000812141422l51e0ae0blbb074049904af71c@mail.gmail.com>
	<32c88000812150739i35b5ff43n72eb6f07c30d21f6@mail.gmail.com>
	<4946811F.5010505@resophonic.com> <49468570.4010808@sympatico.ca>
	<32c88000812150829s76142dbj6c0317fda05fc222@mail.gmail.com>
	<32c88000812150833v7bd2c8f4ve9dc483186a41346@mail.gmail.com>
	<32c88000812150838j678e598fh1badf15351a6b8bd@mail.gmail.com>
Message-ID: <32c88000812150848g541b3fa1vb36445b7b1b26f42@mail.gmail.com>

On Mon, Dec 15, 2008 at 5:38 PM, Leonard Ritter <paniq at paniq.org> wrote:

> i tried both 1.34.1 and 1.35.0 available in my distribution (ubuntu), same
>>> problem. i am fetching through the boost.python code to see if i can find
>>> something interesting.
>>>
>>
>> i should add that this problem does not happen on other machines with the
>> SAME distribution installed.
>>
>
> ...and the error generated is a segfault, no standard exception!
>

searching for __cxa_allocate_exception segfault i got quite a few hits in
different projects, of which this wiki entry devotes a whole page to the
issue:

http://wiki.fifengine.de/Segfault_in_cxa_allocate_exception

The funny thing is: I never got the problem on a machine with an nVidia
driver. The two machines who have the reported error run a virtual video
card driver and I'm currently asking for the other one, but I bet it's also
pure mesa-based or ATI.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081215/c297bb6b/attachment.htm>

From paniq at paniq.org  Mon Dec 15 17:53:35 2008
From: paniq at paniq.org (Leonard Ritter)
Date: Mon, 15 Dec 2008 17:53:35 +0100
Subject: [C++-sig] preventing boost::python::throw_error_already_set
In-Reply-To: <32c88000812150848g541b3fa1vb36445b7b1b26f42@mail.gmail.com>
References: <32c88000812110818q21289c79i94a42ca8a0b6f09c@mail.gmail.com>
	<494140B1.1040200@sympatico.ca>
	<32c88000812141422l51e0ae0blbb074049904af71c@mail.gmail.com>
	<32c88000812150739i35b5ff43n72eb6f07c30d21f6@mail.gmail.com>
	<4946811F.5010505@resophonic.com> <49468570.4010808@sympatico.ca>
	<32c88000812150829s76142dbj6c0317fda05fc222@mail.gmail.com>
	<32c88000812150833v7bd2c8f4ve9dc483186a41346@mail.gmail.com>
	<32c88000812150838j678e598fh1badf15351a6b8bd@mail.gmail.com>
	<32c88000812150848g541b3fa1vb36445b7b1b26f42@mail.gmail.com>
Message-ID: <32c88000812150853t53b1157clad7f915fc38be01f@mail.gmail.com>

solved!

On Mon, Dec 15, 2008 at 5:48 PM, Leonard Ritter <paniq at paniq.org> wrote:

> searching for __cxa_allocate_exception segfault i got quite a few hits in
> different projects, of which this wiki entry devotes a whole page to the
> issue:
>
> http://wiki.fifengine.de/Segfault_in_cxa_allocate_exception
>
> The funny thing is: I never got the problem on a machine with an nVidia
> driver. The two machines who have the reported error run a virtual video
> card driver and I'm currently asking for the other one, but I bet it's also
> pure mesa-based or ATI.
>
>
I solved the problem by linking to libstdc++ before linking to anything else
as described in the above Wiki-page. The exception does no longer occur.

Gentlemen! I rest my case!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081215/4331a882/attachment.htm>

From rwgk at yahoo.com  Mon Dec 15 17:58:07 2008
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Mon, 15 Dec 2008 08:58:07 -0800 (PST)
Subject: [C++-sig] custom constructor problems
References: <95291a80812150617x7b727926keeccdf8b813ccecc@mail.gmail.com>
Message-ID: <631913.99242.qm@web111402.mail.gq1.yahoo.com>

I think this is related:

http://mail.python.org/pipermail/cplusplus-sig/2006-March/010092.html
http://mail.python.org/pipermail/cplusplus-sig/2006-March/010095.html

I gave up on the obj1 == obj2 comparison idea and added a
..memory_id() method, which returns the memory location (cast to Python int).




----- Original Message ----
From: Renato Araujo <renatox at gmail.com>
To: Development of Python/C++ integration <cplusplus-sig at python.org>
Sent: Monday, December 15, 2008 6:17:16 AM
Subject: [C++-sig] custom constructor problems

Hi again :D,

To solve my old problem, I'm trying create custom constructor which
return a shared_ptr<MyClass> like that code:

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
class my_class_wrapper : my_class, python::wrapper<my_class>
{
.......
}

shared_ptr<my_class_wrapper> _create_object(parent *p)
{
    return shared_ptr<my_class_wrapper>(new my_class_wrapper(p),
_my_delete_function);
}

shared_ptr<my_class_wrapper> _create_object2()
{
    return shared_ptr<my_class_wrapper>(new my_class_wrapper(0),
_my_delete_function);
}


class_<my_class_wrapper>("MyClass", python::no_init)
    .def("__init__", python::make_constructor(_create_object))
    .def("__init__", python::make_constructor(_create_object2))
    .def("get_parent", &MyClass:getParent,
return_value_policy<reference_existing_object>());

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

but during my python test, the function get_parent not return the same
python object created in the object constructor. my test is:

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
parent = MyClass()
child = MyClass(parent)

if (child.get_parent() != parent)
   print "FAIL"
else
   print "OK"
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Always I got the "FAIL" message, but if I remove the custom
constructor the test pass. What I doing wrong all my classes inherit
from "python::wrapper<X>". Then I think all classes have implicit
conversion to "shared_ptr" correct?

BR
-- 
Renato Araujo Oliveira Filho
_______________________________________________
Cplusplus-sig mailing list
Cplusplus-sig at python.org
http://mail.python.org/mailman/listinfo/cplusplus-sig



From seefeld at sympatico.ca  Mon Dec 15 17:58:22 2008
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Mon, 15 Dec 2008 11:58:22 -0500
Subject: [C++-sig] preventing boost::python::throw_error_already_set
In-Reply-To: <32c88000812150848g541b3fa1vb36445b7b1b26f42@mail.gmail.com>
References: <32c88000812110818q21289c79i94a42ca8a0b6f09c@mail.gmail.com>	<494140B1.1040200@sympatico.ca>	<32c88000812141422l51e0ae0blbb074049904af71c@mail.gmail.com>	<32c88000812150739i35b5ff43n72eb6f07c30d21f6@mail.gmail.com>	<4946811F.5010505@resophonic.com>
	<49468570.4010808@sympatico.ca>	<32c88000812150829s76142dbj6c0317fda05fc222@mail.gmail.com>	<32c88000812150833v7bd2c8f4ve9dc483186a41346@mail.gmail.com>	<32c88000812150838j678e598fh1badf15351a6b8bd@mail.gmail.com>
	<32c88000812150848g541b3fa1vb36445b7b1b26f42@mail.gmail.com>
Message-ID: <49468CAE.4050908@sympatico.ca>

Leonard Ritter wrote:
>
> searching for __cxa_allocate_exception segfault i got quite a few hits 
> in different projects, of which this wiki entry devotes a whole page 
> to the issue:
>
> http://wiki.fifengine.de/Segfault_in_cxa_allocate_exception
>
> The funny thing is: I never got the problem on a machine with an 
> nVidia driver. The two machines who have the reported error run a 
> virtual video card driver and I'm currently asking for the other one, 
> but I bet it's also pure mesa-based or ATI.

The first thing to do in order to confirm this (or in fact in any case 
to better understand the issue) is to write a minimal complete test case 
exhibiting the error. If you can indeed confirm that *any* exception 
will cause this segfault you have already eliminated boost (and python), 
which should make a further drill-down much easier.

Good luck,

       Stefan


-- 

      ...ich hab' noch einen Koffer in Berlin...


From smnemeth at gmail.com  Mon Dec 15 21:22:27 2008
From: smnemeth at gmail.com (Rocketman@JSC)
Date: Mon, 15 Dec 2008 12:22:27 -0800 (PST)
Subject: [C++-sig] boost::python::throw_error_already_set not in library
In-Reply-To: <87oczganqk.fsf@mcbain.luannocracy.com>
References: <20983772.post@talk.nabble.com>
	<87hc59au7p.fsf@mcbain.luannocracy.com>
	<20985178.post@talk.nabble.com>
	<87oczganqk.fsf@mcbain.luannocracy.com>
Message-ID: <21021185.post@talk.nabble.com>


Problem Solved: The symbols are in the shared object, and as David mentioned,
one must use the -T to see them in a debug library.

The problem was with the intel linker (via eclipse).  Apparently the linker
will not tell you whether a shared library is actually included using
-l<your library> to add to a shared object.  One could put in anything and
it will just build the .so without mentioning it could not find <your
library>.

Thanks for your help.

Scott
-- 
View this message in context: http://www.nabble.com/boost%3A%3Apython%3A%3Athrow_error_already_set-not-in-library-tp20983772p21021185.html
Sent from the Python - c++-sig mailing list archive at Nabble.com.


From jeremie.delaitre at technogerma.fr  Wed Dec 17 10:07:14 2008
From: jeremie.delaitre at technogerma.fr (=?ISO-8859-1?Q?J=E9r=E9mie_Delaitre?=)
Date: Wed, 17 Dec 2008 10:07:14 +0100
Subject: [C++-sig] Dynamic resolution of members and methods
In-Reply-To: <49465475.5090205@technogerma.fr>
References: <493FE997.2050005@technogerma.fr>
	<494002A4.5070101@pixar.com>	<494252B8.4040608@technogerma.fr>
	<49465475.5090205@technogerma.fr>
Message-ID: <4948C142.8030508@technogerma.fr>

Ok, I have find a way to resolve all of my problem.
I simply declare a new custom type which is callable.
This way, when the user type in python 'myObj.myFunc(...)', 
mytype_getattro is called
with 'myObj' as self and 'myFunc' as arg. Then, I return a new instance 
of the callable type
which is fill with all the information I need to identify the function. 
Then python simply call the
functor.

J?r?mie Delaitre


J?r?mie Delaitre a ?crit :
>
> Hi,
>
> When defining my new type in C++, I have set the "tp_getattro" field 
> to point to a function (say "mytype_getattro").
> This function determine dynamically if the requested attribute exists 
> or not, and if it does, return its value.
> So, it works well for attributes/properties.
>
> But, "mytype_getattro" is also called when, in Python, I do a function 
> call 'myobject.myfct(...)'. It allows me to determine
> dynamically if the requested function "myfct" is available or not. But 
> if it does, I need to return a callable object to allow
> python to process the call.
>
> I succeeded to make Python to call a function of mine using a 
> PyMethodDef array and using the Py_FindMethod function
> to get a callable PyObject*.
> But it is not a good solution for at least two reasons :
> 1 - Py_FindMethod is not supported in Python 3.0
> 2 - I need to declare each possible function in the array (which is 
> impossible has I need to determine them at runtime)
>
> What I have tried to do is to declare only one "dipatch" function in 
> the array which will call the right C++ function.
> This way, in "mytype_getattro" I do a "return Py_FindMethod(myarray, 
> self, "dispatch")". Thus, python call my
> dispatching method. The problem is that I need to attach additionnal 
> values to the function (like the name of the function
> which must be called in reality). In Lua, I used the closure 
> mechanism, but I did not success with Python closures.
>
> Any hints ?
> Is closure the right approach (If so, how I should use it ?)? Do I 
> need to set another field when defining my type ?
>
> Regards,
>
> J?r?mie Delaitre
>
>
> J?r?mie Delaitre a ?crit :
>>
>> It works ! Thanks a lot.
>>
>> J?r?mie
>>
>> Alex Mohr a ?crit :
>>> Override the python special methods __getattr__ or __getattribute__.
>>>
>>> http://docs.python.org/reference/datamodel.html#attribute-access
>>>
>>> Alex
>>>
>>> J?r?mie Delaitre wrote:
>>>> Hello,
>>>>
>>>> I have a C++ class that use a custom property system.
>>>> This property system allows me to add and remove properties at 
>>>> runtime.
>>>>
>>>> I want to expose this class into Python.
>>>>
>>>> The problem is: I want "a dynamic resolution" of the properties in 
>>>> Python.
>>>>
>>>> For example:
>>>>
>>>> # import my python extension
>>>> import myModule
>>>>
>>>> # get an instance of my C++ class
>>>> t = myModule.newInstance()
>>>>
>>>> # access the property 'myProperty'
>>>> print(t.myProperty)
>>>>
>>>> When I access the property, I want python to call a "callback" of 
>>>> mine (written
>>>> in the C++ extension) which will check if the requested property 
>>>> exists, and
>>>> if it does, return the corresponding value, otherwise generate an 
>>>> exception.
>>>>
>>>> By looking at the python C API, I did not find a way to achieve 
>>>> this. It seems
>>>> that declaring a new type requires to have static arrays for 
>>>> members and methods
>>>> (yes, my property system allows me to declare new methods too, so I 
>>>> want to expose
>>>> them with a similar system).
>>>>
>>>> Any idea how I could resolve this (with Python3.0) ?
>>>>
>>>> Regards,
>>>>
>>>> J?r?mie Delaitre
>>>>
>>>>
>>>> _______________________________________________
>>>> Cplusplus-sig mailing list
>>>> Cplusplus-sig at python.org
>>>> http://mail.python.org/mailman/listinfo/cplusplus-sig
>>>
>>> _______________________________________________
>>> Cplusplus-sig mailing list
>>> Cplusplus-sig at python.org
>>> http://mail.python.org/mailman/listinfo/cplusplus-sig
>>>
>>
>>
>> _______________________________________________
>> Cplusplus-sig mailing list
>> Cplusplus-sig at python.org
>> http://mail.python.org/mailman/listinfo/cplusplus-sig
>>
>
>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>



From seefeld at sympatico.ca  Wed Dec 17 12:35:16 2008
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Wed, 17 Dec 2008 06:35:16 -0500
Subject: [C++-sig] Dynamic resolution of members and methods
In-Reply-To: <4948C142.8030508@technogerma.fr>
References: <493FE997.2050005@technogerma.fr>	<494002A4.5070101@pixar.com>	<494252B8.4040608@technogerma.fr>	<49465475.5090205@technogerma.fr>
	<4948C142.8030508@technogerma.fr>
Message-ID: <4948E3F4.7020204@sympatico.ca>

J?r?mie Delaitre wrote:
> Ok, I have find a way to resolve all of my problem.
> I simply declare a new custom type which is callable.
> This way, when the user type in python 'myObj.myFunc(...)', 
> mytype_getattro is called
> with 'myObj' as self and 'myFunc' as arg. Then, I return a new 
> instance of the callable type
> which is fill with all the information I need to identify the 
> function. Then python simply call the
> functor.

 From this and your previous mail it doesn't sound as if you used any 
boost.python for this. Any reason for that ?

Regards,
       Stefan

-- 

      ...ich hab' noch einen Koffer in Berlin...


From jeremie.delaitre at technogerma.fr  Wed Dec 17 13:21:26 2008
From: jeremie.delaitre at technogerma.fr (=?ISO-8859-1?Q?J=E9r=E9mie_Delaitre?=)
Date: Wed, 17 Dec 2008 13:21:26 +0100
Subject: [C++-sig] Dynamic resolution of members and methods
In-Reply-To: <4948E3F4.7020204@sympatico.ca>
References: <493FE997.2050005@technogerma.fr>	<494002A4.5070101@pixar.com>	<494252B8.4040608@technogerma.fr>	<49465475.5090205@technogerma.fr>	<4948C142.8030508@technogerma.fr>
	<4948E3F4.7020204@sympatico.ca>
Message-ID: <4948EEC6.2050109@technogerma.fr>

Stefan Seefeld a ?crit :
> J?r?mie Delaitre wrote:
>> Ok, I have find a way to resolve all of my problem.
>> I simply declare a new custom type which is callable.
>> This way, when the user type in python 'myObj.myFunc(...)', 
>> mytype_getattro is called
>> with 'myObj' as self and 'myFunc' as arg. Then, I return a new 
>> instance of the callable type
>> which is fill with all the information I need to identify the 
>> function. Then python simply call the
>> functor.
>
> From this and your previous mail it doesn't sound as if you used any 
> boost.python for this. Any reason for that ?
>
> Regards,
>       Stefan
>
We are not using boost.python because we already have a system which 
allow us to bind members, methods, ... into a metaclass.
We just want to publish these metaclasses into Python.

As we already declare our bindings, we do not want to declare another 
one for Python. Moreover, we want to support other script engine.
Thus, we do not want to declare bindings for each but rather give to the 
script engines an access to our custom system.

Regards,

J?r?mie Delaitre



From hugo.lima at openbossa.org  Wed Dec 17 17:35:56 2008
From: hugo.lima at openbossa.org (Hugo Lima)
Date: Wed, 17 Dec 2008 13:35:56 -0300
Subject: [C++-sig] Enums with duplicated values.
Message-ID: <7ebf1b2a0812170835m1816634g8d8fbd8a82cb9849@mail.gmail.com>

Hi;

I need wrap enums with duplicated values, but boost::python do not
support it. At the boost website you can find the quote:

"Scott Snyder provided a patch; Dave was dissatisfied for some reason,
but maybe it should just be applied if no further action occurs
http://aspn.activestate.com/ASPN/Mail/Message/1824616."

Why this patch was not been accepted since 2003?

What can we do to get this patch accepted, so we do not have to
maintain a boost-copy on our repositories?

From yunlin07 at gmail.com  Thu Dec 18 01:59:43 2008
From: yunlin07 at gmail.com (lin yun)
Date: Wed, 17 Dec 2008 16:59:43 -0800
Subject: [C++-sig] Wrap a c++ function that returns boost::shared_ptr<> type?
Message-ID: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>

Hi, folks:

I am trying to wrap a c++ function that returns a boost::shared_ptr<some
class> type using boost.python, is that possible?

BTW, I used the call policy return_range included in Py++ for returning a
pointer to a vector. It worked !  My question is that which side is
responsible for releasing the memory of the vector, python side or C++ side?

Thank you for your help very much!

Yun
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081217/7025433f/attachment.htm>

From roman.yakovenko at gmail.com  Thu Dec 18 07:41:43 2008
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Thu, 18 Dec 2008 08:41:43 +0200
Subject: [C++-sig] Wrap a c++ function that returns boost::shared_ptr<>
	type?
In-Reply-To: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>
References: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>
Message-ID: <7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>

2008/12/18 lin yun <yunlin07 at gmail.com>:
> Hi, folks:
>
> I am trying to wrap a c++ function that returns a boost::shared_ptr<some
> class> type using boost.python, is that possible?

Yes. I believe Py++ deals with boost::shared_ptr without invoking the
user. Just use default call policy.

> BTW, I used the call policy return_range included in Py++ for returning a
> pointer to a vector. It worked !  My question is that which side is
> responsible for releasing the memory of the vector, python side or C++ side?

Can you provide small example of what you are trying to do?

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

From yunlin07 at gmail.com  Thu Dec 18 07:58:49 2008
From: yunlin07 at gmail.com (lin yun)
Date: Wed, 17 Dec 2008 22:58:49 -0800
Subject: [C++-sig] Wrap a c++ function that returns boost::shared_ptr<>
	type?
In-Reply-To: <7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>
References: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>
	<7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>
Message-ID: <e3c115370812172258m50f22693w248b887d57227f64@mail.gmail.com>

> Yes. I believe Py++ deals with boost::shared_ptr without invoking the
> user. Just use default call policy.
That's good news! How about vector<boost::shared_ptr<unsigned short *>> ?

> Can you provide small example of what you are trying to do?

I have a C++ function "imgreader" which reads an image and return a
"unsigned short *" pointer. I have another piece of Python code that calls
imgreader and gets the pointer to the image data. Where should I release the
memory of the image data when it is not needed any more? in python side or
C++ side?

Thanks a lot !

Yun






On Wed, Dec 17, 2008 at 10:41 PM, Roman Yakovenko <roman.yakovenko at gmail.com
> wrote:

> 2008/12/18 lin yun <yunlin07 at gmail.com>:
> > Hi, folks:
> >
> > I am trying to wrap a c++ function that returns a boost::shared_ptr<some
> > class> type using boost.python, is that possible?
>
> Yes. I believe Py++ deals with boost::shared_ptr without invoking the
> user. Just use default call policy.
>
> > BTW, I used the call policy return_range included in Py++ for returning a
> > pointer to a vector. It worked !  My question is that which side is
> > responsible for releasing the memory of the vector, python side or C++
> side?
>
> Can you provide small example of what you are trying to do?
>
> --
> Roman Yakovenko
> C++ Python language binding
> http://www.language-binding.net/
> _______________________________________________
> 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/20081217/af7ad77b/attachment-0001.htm>

From roman.yakovenko at gmail.com  Thu Dec 18 08:32:51 2008
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Thu, 18 Dec 2008 09:32:51 +0200
Subject: [C++-sig] Wrap a c++ function that returns boost::shared_ptr<>
	type?
In-Reply-To: <e3c115370812172258m50f22693w248b887d57227f64@mail.gmail.com>
References: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>
	<7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>
	<e3c115370812172258m50f22693w248b887d57227f64@mail.gmail.com>
Message-ID: <7465b6170812172332r768f921aj6f93f8e892fc7b50@mail.gmail.com>

2008/12/18 lin yun <yunlin07 at gmail.com>:
>> Yes. I believe Py++ deals with boost::shared_ptr without invoking the
>> user. Just use default call policy.
> That's good news! How about vector<boost::shared_ptr<unsigned short *>> ?

No :-(,  unsigned short is immutable type. You will have too write
wrapper. Take a look on Py++ - ctypes functionality:

http://language-binding.net/pyplusplus/documentation/ctypes/ctypes_integration.html

>> Can you provide small example of what you are trying to do?
>
> I have a C++ function "imgreader" which reads an image and return a
> "unsigned short *" pointer. I have another piece of Python code that calls
> imgreader and gets the pointer to the image data. Where should I release the
> memory of the image data when it is not needed any more? in python side or
> C++ side?

return_range call policy doesn't manage the memory. It assumes that
the buffer is owned by C++ class.

I guess you should use "as_tuple" call policy, with
memory_managers::delete_ memory manager.
This way you don't need to force user to manage the memory.

May be you will have to create new call policy - hybrid of
return_range and as_tuple:
* efficiency of return_range
* memory management ability of as_tuple

HTH

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

From yunlin07 at gmail.com  Thu Dec 18 08:44:11 2008
From: yunlin07 at gmail.com (lin yun)
Date: Wed, 17 Dec 2008 23:44:11 -0800
Subject: [C++-sig] Wrap a c++ function that returns boost::shared_ptr<>
	type?
In-Reply-To: <7465b6170812172332r768f921aj6f93f8e892fc7b50@mail.gmail.com>
References: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>
	<7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>
	<e3c115370812172258m50f22693w248b887d57227f64@mail.gmail.com>
	<7465b6170812172332r768f921aj6f93f8e892fc7b50@mail.gmail.com>
Message-ID: <e3c115370812172344w1847b40fib95c1c373b8ab7a7@mail.gmail.com>

Actually I used return_range with element type of unsigned short. Python
interprets the element as int.

You mean it is not difficult to wrap a function that returns
vector<boost::shared_ptr<char>> then?

OK. Then the C++ class needs to take care of the memory when using
return_range. I am still a newbie in boost python. I am learning and making
progress. I hope I could write new call policies in the future like you did
for py++.:)

Thanks.



On Wed, Dec 17, 2008 at 11:32 PM, Roman Yakovenko <roman.yakovenko at gmail.com
> wrote:

> 2008/12/18 lin yun <yunlin07 at gmail.com>:
> >> Yes. I believe Py++ deals with boost::shared_ptr without invoking the
> >> user. Just use default call policy.
> > That's good news! How about vector<boost::shared_ptr<unsigned short *>> ?
>
> No :-(,  unsigned short is immutable type. You will have too write
> wrapper. Take a look on Py++ - ctypes functionality:
>
>
> http://language-binding.net/pyplusplus/documentation/ctypes/ctypes_integration.html
>
> >> Can you provide small example of what you are trying to do?
> >
> > I have a C++ function "imgreader" which reads an image and return a
> > "unsigned short *" pointer. I have another piece of Python code that
> calls
> > imgreader and gets the pointer to the image data. Where should I release
> the
> > memory of the image data when it is not needed any more? in python side
> or
> > C++ side?
>
> return_range call policy doesn't manage the memory. It assumes that
> the buffer is owned by C++ class.
>
> I guess you should use "as_tuple" call policy, with
> memory_managers::delete_ memory manager.
> This way you don't need to force user to manage the memory.
>
> May be you will have to create new call policy - hybrid of
> return_range and as_tuple:
> * efficiency of return_range
> * memory management ability of as_tuple
>
> HTH
>
> --
>  Roman Yakovenko
> C++ Python language binding
> http://www.language-binding.net/
> _______________________________________________
> 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/20081217/36e9db01/attachment.htm>

From roman.yakovenko at gmail.com  Thu Dec 18 09:37:33 2008
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Thu, 18 Dec 2008 10:37:33 +0200
Subject: [C++-sig] Wrap a c++ function that returns boost::shared_ptr<>
	type?
In-Reply-To: <e3c115370812172344w1847b40fib95c1c373b8ab7a7@mail.gmail.com>
References: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>
	<7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>
	<e3c115370812172258m50f22693w248b887d57227f64@mail.gmail.com>
	<7465b6170812172332r768f921aj6f93f8e892fc7b50@mail.gmail.com>
	<e3c115370812172344w1847b40fib95c1c373b8ab7a7@mail.gmail.com>
Message-ID: <7465b6170812180037vc897c10wb343532dab43f3bc@mail.gmail.com>

2008/12/18 lin yun <yunlin07 at gmail.com>:
> Actually I used return_range with element type of unsigned short. Python
> interprets the element as int.
>
> You mean it is not difficult to wrap a function that returns
> vector<boost::shared_ptr<char>> then?

Yes. Take a look on example:
http://language-binding.net/pyplusplus/documentation/functions/transformation/output_static_array.html

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

From ndbecker2 at gmail.com  Thu Dec 18 13:19:35 2008
From: ndbecker2 at gmail.com (Neal Becker)
Date: Thu, 18 Dec 2008 07:19:35 -0500
Subject: [C++-sig] Wrap a c++ function that returns boost::shared_ptr<>
	type?
References: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>
	<7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>
	<e3c115370812172258m50f22693w248b887d57227f64@mail.gmail.com>
	<7465b6170812172332r768f921aj6f93f8e892fc7b50@mail.gmail.com>
	<e3c115370812172344w1847b40fib95c1c373b8ab7a7@mail.gmail.com>
	<7465b6170812180037vc897c10wb343532dab43f3bc@mail.gmail.com>
Message-ID: <gidf4n$rjk$1@ger.gmane.org>

Roman Yakovenko wrote:

> 2008/12/18 lin yun <yunlin07 at gmail.com>:
>> Actually I used return_range with element type of unsigned short. Python
>> interprets the element as int.
>>
>> You mean it is not difficult to wrap a function that returns
>> vector<boost::shared_ptr<char>> then?
> 
> Yes. Take a look on example:
> http://language-binding.net/pyplusplus/documentation/functions/transformation/output_static_array.html
> 

This link seems to be broken.  I keep getting 'server closed connection'


From ndbecker2 at gmail.com  Thu Dec 18 13:20:10 2008
From: ndbecker2 at gmail.com (Neal Becker)
Date: Thu, 18 Dec 2008 07:20:10 -0500
Subject: [C++-sig] Wrap a c++ function that returns boost::shared_ptr<>
	type?
References: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>
	<7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>
	<e3c115370812172258m50f22693w248b887d57227f64@mail.gmail.com>
	<7465b6170812172332r768f921aj6f93f8e892fc7b50@mail.gmail.com>
	<e3c115370812172344w1847b40fib95c1c373b8ab7a7@mail.gmail.com>
	<7465b6170812180037vc897c10wb343532dab43f3bc@mail.gmail.com>
Message-ID: <gidf5q$rjk$2@ger.gmane.org>

I'm curious about 'return_range'.  What does it do?  Is there any doc?


From yunlin07 at gmail.com  Thu Dec 18 18:08:41 2008
From: yunlin07 at gmail.com (lin yun)
Date: Thu, 18 Dec 2008 09:08:41 -0800
Subject: [C++-sig] Wrap a c++ function that returns boost::shared_ptr<>
	type?
In-Reply-To: <gidf4n$rjk$1@ger.gmane.org>
References: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>
	<7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>
	<e3c115370812172258m50f22693w248b887d57227f64@mail.gmail.com>
	<7465b6170812172332r768f921aj6f93f8e892fc7b50@mail.gmail.com>
	<e3c115370812172344w1847b40fib95c1c373b8ab7a7@mail.gmail.com>
	<7465b6170812180037vc897c10wb343532dab43f3bc@mail.gmail.com>
	<gidf4n$rjk$1@ger.gmane.org>
Message-ID: <e3c115370812180908x799c917fha88ec39b583235d7@mail.gmail.com>

The link is alive to me... I can open it. Thanks Roman.

On Thu, Dec 18, 2008 at 4:19 AM, Neal Becker <ndbecker2 at gmail.com> wrote:

> Roman Yakovenko wrote:
>
> > 2008/12/18 lin yun <yunlin07 at gmail.com>:
> >> Actually I used return_range with element type of unsigned short. Python
> >> interprets the element as int.
> >>
> >> You mean it is not difficult to wrap a function that returns
> >> vector<boost::shared_ptr<char>> then?
> >
> > Yes. Take a look on example:
> >
> http://language-binding.net/pyplusplus/documentation/functions/transformation/output_static_array.html
> >
>
> This link seems to be broken.  I keep getting 'server closed connection'
>
> _______________________________________________
> 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/20081218/1c559a79/attachment.htm>

From yunlin07 at gmail.com  Thu Dec 18 18:10:10 2008
From: yunlin07 at gmail.com (lin yun)
Date: Thu, 18 Dec 2008 09:10:10 -0800
Subject: [C++-sig] Wrap a c++ function that returns boost::shared_ptr<>
	type?
In-Reply-To: <gidf5q$rjk$2@ger.gmane.org>
References: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>
	<7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>
	<e3c115370812172258m50f22693w248b887d57227f64@mail.gmail.com>
	<7465b6170812172332r768f921aj6f93f8e892fc7b50@mail.gmail.com>
	<e3c115370812172344w1847b40fib95c1c373b8ab7a7@mail.gmail.com>
	<7465b6170812180037vc897c10wb343532dab43f3bc@mail.gmail.com>
	<gidf5q$rjk$2@ger.gmane.org>
Message-ID: <e3c115370812180910n2763480dqe42cb6751cd0efc5@mail.gmail.com>

Here it is:
http://www.language-binding.net/pyplusplus/documentation/functions/call_policies/return_range.html

return_range is a call policy that is used to wrap a C++ function which
returns a pointer to an array.

On Thu, Dec 18, 2008 at 4:20 AM, Neal Becker <ndbecker2 at gmail.com> wrote:

> I'm curious about 'return_range'.  What does it do?  Is there any doc?
>
> _______________________________________________
> 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/20081218/7ebeac61/attachment.htm>

From dave at boostpro.com  Thu Dec 18 19:17:32 2008
From: dave at boostpro.com (David Abrahams)
Date: Thu, 18 Dec 2008 09:17:32 -0900
Subject: [C++-sig] Wrap a c++ function that returns boost::shared_ptr<>
	type?
In-Reply-To: <7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>
	(Roman Yakovenko's message of "Thu, 18 Dec 2008 08:41:43 +0200")
References: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>
	<7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>
Message-ID: <87r645s5z7.fsf@mcbain.luannocracy.com>


on Wed Dec 17 2008, "Roman Yakovenko" <roman.yakovenko-AT-gmail.com> wrote:

> 2008/12/18 lin yun <yunlin07 at gmail.com>:
>> Hi, folks:
>>
>> I am trying to wrap a c++ function that returns a boost::shared_ptr<some
>> class> type using boost.python, is that possible?
>
> Yes. I believe Py++ deals with boost::shared_ptr without invoking the
> user. Just use default call policy.

Boost.Python itself also deals with shared_ptr without involving the
user.

-- 
Dave Abrahams
BoostPro Computing
http://www.boostpro.com

From dave at boostpro.com  Thu Dec 18 19:18:16 2008
From: dave at boostpro.com (David Abrahams)
Date: Thu, 18 Dec 2008 09:18:16 -0900
Subject: [C++-sig] Wrap a c++ function that returns boost::shared_ptr<>
	type?
In-Reply-To: <e3c115370812172258m50f22693w248b887d57227f64@mail.gmail.com>
	(lin yun's message of "Wed, 17 Dec 2008 22:58:49 -0800")
References: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>
	<7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>
	<e3c115370812172258m50f22693w248b887d57227f64@mail.gmail.com>
Message-ID: <87myets5xz.fsf@mcbain.luannocracy.com>


on Wed Dec 17 2008, "lin yun" <yunlin07-AT-gmail.com> wrote:

>> Yes. I believe Py++ deals with boost::shared_ptr without invoking the
>> user. Just use default call policy.
> That's good news! How about vector<boost::shared_ptr<unsigned short *>> ?

If you use the facilities of
http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/iterator.html,
everything *should* "just work."

-- 
Dave Abrahams
BoostPro Computing
http://www.boostpro.com

From roman.yakovenko at gmail.com  Thu Dec 18 20:20:13 2008
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Thu, 18 Dec 2008 21:20:13 +0200
Subject: [C++-sig] Wrap a c++ function that returns boost::shared_ptr<>
	type?
In-Reply-To: <87myets5xz.fsf@mcbain.luannocracy.com>
References: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>
	<7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>
	<e3c115370812172258m50f22693w248b887d57227f64@mail.gmail.com>
	<87myets5xz.fsf@mcbain.luannocracy.com>
Message-ID: <7465b6170812181120l7b21aaa0t84f35fd68a05b684@mail.gmail.com>

On Thu, Dec 18, 2008 at 8:18 PM, David Abrahams <dave at boostpro.com> wrote:
>
> on Wed Dec 17 2008, "lin yun" <yunlin07-AT-gmail.com> wrote:
>
>>> Yes. I believe Py++ deals with boost::shared_ptr without invoking the
>>> user. Just use default call policy.
>> That's good news! How about vector<boost::shared_ptr<unsigned short *>> ?
>
> If you use the facilities of
> http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/iterator.html,
> everything *should* "just work."

Do you mean to "redefine"  class_< std::vector< ... > > "get"
function, so it will return the iterators range?

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

From dave at boostpro.com  Fri Dec 19 03:13:05 2008
From: dave at boostpro.com (David Abrahams)
Date: Thu, 18 Dec 2008 17:13:05 -0900
Subject: [C++-sig] Enums with duplicated values.
In-Reply-To: <7ebf1b2a0812170835m1816634g8d8fbd8a82cb9849@mail.gmail.com>
	(Hugo Lima's message of "Wed, 17 Dec 2008 13:35:56 -0300")
References: <7ebf1b2a0812170835m1816634g8d8fbd8a82cb9849@mail.gmail.com>
Message-ID: <87ljuclxou.fsf@mcbain.luannocracy.com>


on Wed Dec 17 2008, "Hugo Lima" <hugo.lima-AT-openbossa.org> wrote:

> Hi;
>
> I need wrap enums with duplicated values, but boost::python do not
> support it. At the boost website you can find the quote:
>
> "Scott Snyder provided a patch; Dave was dissatisfied for some reason,

I think I didn't really like the asymmetry of the "othervalues" field; I
would probably prefer to store all values in a dict of lists.

> but maybe it should just be applied if no further action occurs
> http://aspn.activestate.com/ASPN/Mail/Message/1824616."
>
> Why this patch was not been accepted since 2003?

Well, I suppose it's a lot better than no patch at all, so it should be
applied.

> What can we do to get this patch accepted, so we do not have to
> maintain a boost-copy on our repositories?

You can run the Boost.Python test suite with the patch in place against
trunk, and assure me that all the tests pass.  If you do that, I'll be
inclined to apply it and move it to the release branch.

Thanks for the bump,

-- 
Dave Abrahams
BoostPro Computing
http://www.boostpro.com

From yunlin07 at gmail.com  Fri Dec 19 04:28:44 2008
From: yunlin07 at gmail.com (lin yun)
Date: Thu, 18 Dec 2008 19:28:44 -0800
Subject: [C++-sig] bjam: undefined Symbol Error
Message-ID: <e3c115370812181928n64cdf513hd2fa62231f3964d2@mail.gmail.com>

Hi, folks:

I am having headache with bjam...  I am getting undefined symbol error.

Part of the Jamroot file is shown as below.  multithread_lib is defined as a
system library which is used in generating my own library img_lib.
Bag_wrapper.cpp wraps a C++ class which uses img_lib.

I am getting importError: libimg_lib.so has undefined symbol.  I have tried
for a whole day. Still have not found a way out...

Could you please help?

Thanks a lot,

Yun


#system library
MTPATH =
/root/boost_1_36_0/bin.v2/libs/thread/build/gcc-4.1.2/release/threading-multi
;
lib multithread_lib : : <name>boost_thread-gcc41-mt-1_36 <search>$(MTPATH)
<link>shared ;

#Generate library
lib img_lib : read_write.cpp label_slice.cpp Region3d.cpp
collapse_mappings.cpp ObjectRegions.cpp Bag.cpp Volume.cpp
label_utilities.cpp multithread_lib : <include>$(IPPINCLUDE) ;

# Declare a Python extension called volReader
python-extension Bag_ext : Bag_wrapper.cpp img_lib ;
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081218/e07dd037/attachment.htm>

From yunlin07 at gmail.com  Fri Dec 19 04:30:19 2008
From: yunlin07 at gmail.com (lin yun)
Date: Thu, 18 Dec 2008 19:30:19 -0800
Subject: [C++-sig] Wrap a c++ function that returns boost::shared_ptr<>
	type?
In-Reply-To: <87myets5xz.fsf@mcbain.luannocracy.com>
References: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>
	<7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>
	<e3c115370812172258m50f22693w248b887d57227f64@mail.gmail.com>
	<87myets5xz.fsf@mcbain.luannocracy.com>
Message-ID: <e3c115370812181930p69d221aetcbc84f78a7b36012@mail.gmail.com>

Would you please elaborate your idea a little more?


On Thu, Dec 18, 2008 at 10:18 AM, David Abrahams <dave at boostpro.com> wrote:

>
> on Wed Dec 17 2008, "lin yun" <yunlin07-AT-gmail.com> wrote:
>
> >> Yes. I believe Py++ deals with boost::shared_ptr without invoking the
> >> user. Just use default call policy.
> > That's good news! How about vector<boost::shared_ptr<unsigned short *>> ?
>
> If you use the facilities of
> http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/iterator.html,
> everything *should* "just work."
>
> --
> Dave Abrahams
> BoostPro Computing
> http://www.boostpro.com
> _______________________________________________
> 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/20081218/bf0d0f3e/attachment.htm>

From yunlin07 at gmail.com  Fri Dec 19 04:42:15 2008
From: yunlin07 at gmail.com (lin yun)
Date: Thu, 18 Dec 2008 19:42:15 -0800
Subject: [C++-sig] bjam: undefined Symbol Error
In-Reply-To: <e3c115370812181928n64cdf513hd2fa62231f3964d2@mail.gmail.com>
References: <e3c115370812181928n64cdf513hd2fa62231f3964d2@mail.gmail.com>
Message-ID: <e3c115370812181942g5b658bbcpe4431a5c0a470d83@mail.gmail.com>

To me, the "undefined symbols" are from the multithread_lib which img_lib
does not get connected to. But why?

On Thu, Dec 18, 2008 at 7:28 PM, lin yun <yunlin07 at gmail.com> wrote:

> Hi, folks:
>
> I am having headache with bjam...  I am getting undefined symbol error.
>
> Part of the Jamroot file is shown as below.  multithread_lib is defined as
> a system library which is used in generating my own library img_lib.
> Bag_wrapper.cpp wraps a C++ class which uses img_lib.
>
> I am getting importError: libimg_lib.so has undefined symbol.  I have tried
> for a whole day. Still have not found a way out...
>
> Could you please help?
>
> Thanks a lot,
>
> Yun
>
>
> #system library
> MTPATH =
> /root/boost_1_36_0/bin.v2/libs/thread/build/gcc-4.1.2/release/threading-multi
> ;
> lib multithread_lib : : <name>boost_thread-gcc41-mt-1_36 <search>$(MTPATH)
> <link>shared ;
>
> #Generate library
> lib img_lib : read_write.cpp label_slice.cpp Region3d.cpp
> collapse_mappings.cpp ObjectRegions.cpp Bag.cpp Volume.cpp
> label_utilities.cpp multithread_lib : <include>$(IPPINCLUDE) ;
>
> # Declare a Python extension called volReader
> python-extension Bag_ext : Bag_wrapper.cpp img_lib ;
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20081218/21db9215/attachment.htm>

From dave at boostpro.com  Fri Dec 19 09:16:44 2008
From: dave at boostpro.com (David Abrahams)
Date: Thu, 18 Dec 2008 23:16:44 -0900
Subject: [C++-sig] Wrap a c++ function that returns boost::shared_ptr<>
	type?
In-Reply-To: <e3c115370812181930p69d221aetcbc84f78a7b36012@mail.gmail.com>
	(lin yun's message of "Thu, 18 Dec 2008 19:30:19 -0800")
References: <e3c115370812171659r5fa3acb7h70c120c0a91db117@mail.gmail.com>
	<7465b6170812172241y3ea8fabi75fdce2efac9c7ef@mail.gmail.com>
	<e3c115370812172258m50f22693w248b887d57227f64@mail.gmail.com>
	<87myets5xz.fsf@mcbain.luannocracy.com>
	<e3c115370812181930p69d221aetcbc84f78a7b36012@mail.gmail.com>
Message-ID: <877i5wk2ab.fsf@mcbain.luannocracy.com>


on Thu Dec 18 2008, "lin yun" <yunlin07-AT-gmail.com> wrote:

> Would you please elaborate your idea a little more?
>
>
> On Thu, Dec 18, 2008 at 10:18 AM, David Abrahams <dave at boostpro.com> wrote:
>
>>
>> on Wed Dec 17 2008, "lin yun" <yunlin07-AT-gmail.com> wrote:
>>
>> >> Yes. I believe Py++ deals with boost::shared_ptr without invoking the
>> >> user. Just use default call policy.
>> > That's good news! How about vector<boost::shared_ptr<unsigned short *>> ?
>>
>> If you use the facilities of
>> http://www.boost.org/doc/libs/1_37_0/libs/python/doc/v2/iterator.html,> everything *should* "just work."

If all you need to do is iterate over the objects pointed to by the
elements of a container of shared_ptrs, then expose the iterators using
one of the functions in that header.

-- 
Dave Abrahams
BoostPro Computing
http://www.boostpro.com

From python at foxcub.org  Sat Dec 20 21:01:36 2008
From: python at foxcub.org (Dmitriy Morozov)
Date: Sat, 20 Dec 2008 12:01:36 -0800
Subject: [C++-sig] pyste and compressed_pair
Message-ID: <2fd72fb20812201201n11be49bah76aeb106cb28d2dd@mail.gmail.com>

Hi,

If I have the simplest header file that includes compressed_pair.hpp,
and I run pyste on it, pyste crashes complaining about is_abstract.hpp
included from compressed_pair.hpp. An example and error messages are
below. I was wondering if this was a known problem, and if there are
any workarounds. I'm trying this right now with Boost 1.36.0, but I
believe the same problem occurs with 1.37.0 (I tried it earlier).

Thanks in advance for your help.
Dmitriy

test.h:
#include <boost/compressed_pair.hpp>

struct A
{
    int a;
};

test.pyste:
Class("A", "test.h")

When running "pyste.py test.pyste" I get:

In file included from /usr/include/boost/type_traits/is_convertible.hpp:25,
                 from /usr/include/boost/type_traits/is_empty.hpp:12,
                 from /usr/include/boost/detail/compressed_pair.hpp:26,
                 from /usr/include/boost/compressed_pair.hpp:21,
                 from test.h:1:
/usr/include/boost/type_traits/is_abstract.hpp:72: error: expected
primary-expression before ')' token
/usr/include/boost/type_traits/is_abstract.hpp:72: error: a function
call cannot appear in a constant-expression
In file included from /usr/include/boost/type_traits/is_empty.hpp:19,
                 from /usr/include/boost/detail/compressed_pair.hpp:26,
                 from /usr/include/boost/compressed_pair.hpp:21,
                 from test.h:1:
/usr/include/boost/type_traits/is_class.hpp:122: error: expected
primary-expression before ')' token
/usr/include/boost/type_traits/is_class.hpp:122: error: a function
call cannot appear in a constant-expression
Traceback (most recent call last):
  File "/usr/bin/pyste.py", line 8, in <module>
    pyste.main()
  File "/usr/lib/python2.5/site-packages/Pyste/pyste.py", line 418, in main
    status = Begin()
  File "/usr/lib/python2.5/site-packages/Pyste/pyste.py", line 257, in Begin
    return GenerateCode(parser, module, out, interfaces, multiple)
  File "/usr/lib/python2.5/site-packages/Pyste/pyste.py", line 376, in
GenerateCode
    declarations, parsed_header = parser.Parse(header, interface, tail)
  File "/usr/lib/python2.5/site-packages/Pyste/CppParser.py", line 165, in Parse
    declarations = self.ParseWithGCCXML(header, tail)
  File "/usr/lib/python2.5/site-packages/Pyste/CppParser.py", line
126, in ParseWithGCCXML
    raise CppParserError, 'Error executing gccxml'
Pyste.CppParser.CppParserError: Error executing gccxml

From timprepscius at gmail.com  Sat Dec 20 21:09:33 2008
From: timprepscius at gmail.com (Tim Prepscius)
Date: Sat, 20 Dec 2008 15:09:33 -0500
Subject: [C++-sig]  custom constructor with embedded back reference
Message-ID: <6fcb94430812201209u4f59b54esfaeb881ce3806ebb@mail.gmail.com>

I just wanted to thank
http://mail.python.org/pipermail/cplusplus-sig/2007-July/012293.html

I've extended what he wrote..  I needed late binding for c++ to python objects.
Hope this is useful for someone else.  Maybe there was some other way
I was supposed to do this, but I couldn't find it.

Um.. And yes this code was just cobbled together.. BasePtr is a
horrible name I know. :-)

#define BOOST_PYTHON_STATIC_LIB
#include <boost/python.hpp>

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

#include <iostream>
#include <map>
namespace python = boost::python;

class Base;

//------------------------------------------------------

#define IMPLEMENT_PYTHON_ACCESSOR_H \
public: \
	static void setClass (boost::python::object _clazz) { clazz = _clazz; } \
	static boost::python::object getClass() { return clazz; } \
	virtual void reflectBind (boost::python::object object) \
	{ \
		::bind(this, object); \
	} \
private: \
	static boost::python::object clazz; \

#define IMPLEMENT_PYTHON_ACCESSOR_CPP(x) \
	boost::python::object x::clazz;

//------------------------------------------------------

template<typename T>
void bind (T *self, boost::python::object object)
{
	// Insert new S object pointer into python object
	typedef boost::python::objects::pointer_holder<T *, T> holder;
	typedef boost::python::objects::instance<holder> instance_t;

	void* memory =
		holder::allocate(
			object.ptr(),
			offsetof(instance_t, storage),
			sizeof(holder)
		);

	try {
		(new (memory) holder(self))->install(object.ptr());
	}
	catch(...) {
		holder::deallocate(object.ptr(), memory);
		throw;
	}
}

//------------------------------------------------------

class BasePtr
{
public:
	BasePtr() { std::cout << this; };
	virtual ~BasePtr () { };

	virtual void reflectBind (boost::python::object pyobject) { }

	void bind (boost::python::object pyobject)
	{
		reflectBind (pyobject);
		this->object = pyobject;
	}

	python::object object;

	python::object getObject () const
	{
		return object;
	}

	void doBind (python::object o)
	{
		bind( o);
	}

	static python::object nowhere (python::object);
} ;

//----------------------------------------------------

class Base : public BasePtr
{
	IMPLEMENT_PYTHON_ACCESSOR_H

public:

	virtual ~Base() { };

	Base () { }

	virtual std::string hello()
	{
	  std::cout << this;
	  return "Base says hello";
	}

	virtual Base *myself ()
	{
		return this;
	}
};

IMPLEMENT_PYTHON_ACCESSOR_CPP (Base);

class Inherited : public Base
{
	IMPLEMENT_PYTHON_ACCESSOR_H

public:
	Inherited () { }
	virtual ~Inherited () { }

	std::string hello()
	{
	  std::cout << this;
	  return "Inherited says hello";
	}
} ;

IMPLEMENT_PYTHON_ACCESSOR_CPP (Inherited);

/*
Not needed after all

template<typename T>
struct BaseToReference
{
    static PyObject* convert(const T &x)
    {
		if (&x == 0)
			return python::incref(python::object().ptr());

		return python::incref(x.getObject().ptr());
    }
};
*/

python::object BasePtr::nowhere (python::object pyobject)
{
	return python::object();
}

//---------------------------------------------------

BOOST_PYTHON_MODULE(embedded_hello)
{
	Base::setClass(
	python::class_<Base, boost::noncopyable> ("Base", python::no_init)
		.def("__init__", &BasePtr::nowhere)
		.def ("hello", &Base::hello,
python::return_value_policy<python::return_by_value>())
		.def ("myself", &Base::myself,
python::return_value_policy<python::reference_existing_object>())
		);

	Inherited::setClass (
		python::class_<Inherited, python::bases<Base>, boost::noncopyable>
("Inherited", python::no_init)
		.def("__init__", &BasePtr::nowhere)
		.def ("hello", &Inherited::hello,
python::return_value_policy<python::return_by_value>())
		);

	// not needed after all
	// python::to_python_converter<Base, BaseToReference<Base> >();
}

//---------------------------------------------------

int main(int argc, char **argv)
{
	// Initialize the interpreter
	Py_Initialize();

	bool error_expected = false;

	if (PyImport_AppendInittab("embedded_hello", initembedded_hello) == -1)
	throw std::runtime_error("Failed to add embedded_hello to the interpreter's "
				 "builtin modules");

	PyObject *z2 = PyImport_ImportModule ("embedded_hello");

	python::object module (python::handle<>(PyImport_ImportModule
("embedded_hello")));

	python::object dict = module.attr("__dict__");

	// Retrieve the main module
	python::object main = python::import("__main__");

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

	python::object result1 = python::exec(
		"from embedded_hello import *        \n"
		"class PythonDerived(Base):          \n"
		"    def __init__(self):              \n"
		"        Base.__init__(self);         \n"
		"\n"
		"    def hello(self):                \n"
		"        print 'hello from python' \n"
		"        print Base.hello(self) \n"
		"        return 'Hello from Python!' \n"
		"\n"
		"    def test(self):                \n"
		"        print 'test1 says ' \n"
		"        print self.myself().hello()\n"
		,

		global, global);

	Base b;
	b.doBind(global["PythonDerived"]());

	python::object o(b.object); ;
	python::object r = o.attr("hello")();
	o.attr("test")();

	python::object result2 = python::exec(
		"from embedded_hello import *        \n"
		"class PythonDerived2(Inherited):          \n"
		"    def __init__(self):              \n"
		"        Inherited.__init__(self);         \n"
		"\n"
		"    def hello(self):                \n"
		"        print 'hello from python' \n"
		"        print Inherited.hello(self) \n"
		"        return 'Hello from Python!' \n"
		"\n"
		"    def test(self):                \n"
		"        print 'test2 says ' \n"
		"        print self.myself().hello()\n"
		,

		global, global);

	Inherited c;
	c.doBind(global["PythonDerived2"]());

	python::object oc(c.object); ;
	python::object rc = oc.attr("hello")();
	oc.attr("test")();

	std::cout << "success!" << std::endl;

}

From timprepscius at gmail.com  Mon Dec 22 04:48:11 2008
From: timprepscius at gmail.com (Tim Prepscius)
Date: Sun, 21 Dec 2008 22:48:11 -0500
Subject: [C++-sig] custom constructor with embedded back reference
Message-ID: <6fcb94430812211948xf30c055p42fca104a4e228e9@mail.gmail.com>

I just wanted to add one more clarification, because actually there is
a bug in the code I posted..

For this late binding to work properly..
You really want to have something like this  (sorry, ripping from
non-test code):

template<typename T>
struct PAccessorConverter
{
    static PyObject* convert(const T &x)
    {
		if (x==0)
// i believe this is sending back None, not sure have to verify
			return boost::python::incref(boost::python::object().ptr());

		return boost::python::incref(x->getPtr().ptr());
    }
};


and have that instantiated something like

#define IMPLEMENT_PYTHON_CONVERTER(x) \
	boost::python::to_python_converter<x *, PAccessorConverter<x *> >();


IMPLEMENT_PYTHON_CONVERTER(myclass)


and then when you're doing a class.def, don't return the
"existing_reference" because that will somehow cause boost python not
to look up the converter..

instead do a return by value... something like:

			.def("getParentUser", &PCamera::getParentUser,
return_value_policy<return_by_value>())

this return_value_policy<return_by_value> queues boost python to
search the converters..
which will eventually hit the converter that uses the embedded python object..

(which allows you to have a 1-to-1 relationship between python and c++)

-tim

From rwgk at yahoo.com  Tue Dec 23 08:58:51 2008
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Mon, 22 Dec 2008 23:58:51 -0800 (PST)
Subject: [C++-sig] Bug and patch for
	boost.pythonwithenable_shared_from_this
References: <229148.259.qm@web31108.mail.mud.yahoo.com>
	<006501c95c3e$cd570610$d100a8c0@nicopc>
	<17fe01c95c69$a505c4f0$d100a8c0@nicopc>
Message-ID: <201094.63149.qm@web111410.mail.gq1.yahoo.com>

Thanks for the patches! They are now in the boost trunk:

------------------------------------------------------------------------
r50368 | rwgk | 2008-12-22 23:55:33 -0800 (Mon, 22 Dec 2008) | 4 lines

Boost.Python enable_shared_from_this patches by Nicolas Lelong and Chad Austin:
  http://mail.python.org/pipermail/cplusplus-sig/2008-December/014103.html
  http://mail.python.org/pipermail/cplusplus-sig/2008-February/013003.html

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

From dave at boostpro.com  Tue Dec 23 10:10:33 2008
From: dave at boostpro.com (David Abrahams)
Date: Tue, 23 Dec 2008 00:10:33 -0900
Subject: [C++-sig] Bug and patch for
	boost.pythonwithenable_shared_from_this
In-Reply-To: <201094.63149.qm@web111410.mail.gq1.yahoo.com> (Ralf
	W. Grosse-Kunstleve's message of "Mon, 22 Dec 2008 23:58:51 -0800
	(PST)")
References: <229148.259.qm@web31108.mail.mud.yahoo.com>
	<006501c95c3e$cd570610$d100a8c0@nicopc>
	<17fe01c95c69$a505c4f0$d100a8c0@nicopc>
	<201094.63149.qm@web111410.mail.gq1.yahoo.com>
Message-ID: <87hc4vdzp2.fsf@mcbain.luannocracy.com>


on Mon Dec 22 2008, "Ralf W. Grosse-Kunstleve" <rwgk-AT-yahoo.com> wrote:

> Thanks for the patches! They are now in the boost trunk:
>
> ------------------------------------------------------------------------
> r50368 | rwgk | 2008-12-22 23:55:33 -0800 (Mon, 22 Dec 2008) | 4 lines
>
> Boost.Python enable_shared_from_this patches by Nicolas Lelong and Chad Austin:
>   http://mail.python.org/pipermail/cplusplus-sig/2008-December/014103.html
>   http://mail.python.org/pipermail/cplusplus-sig/2008-February/013003.html

FYI, Ralf, trunk changes don't get merged automatically to release, so
we now need a mechanism to make sure changes we want released are
actually released.

-- 
Dave Abrahams
BoostPro Computing
http://www.boostpro.com

From rwgk at yahoo.com  Tue Dec 23 22:27:30 2008
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Tue, 23 Dec 2008 13:27:30 -0800 (PST)
Subject: [C++-sig] Bug and patch for
	boost.pythonwithenable_shared_from_this
References: <229148.259.qm@web31108.mail.mud.yahoo.com>
	<006501c95c3e$cd570610$d100a8c0@nicopc>
	<17fe01c95c69$a505c4f0$d100a8c0@nicopc>
	<201094.63149.qm@web111410.mail.gq1.yahoo.com>
	<87hc4vdzp2.fsf@mcbain.luannocracy.com>
Message-ID: <341728.23668.qm@web111406.mail.gq1.yahoo.com>

Oh, that's new to me. Thanks!
Is there a web page describing the release procedure? (A quick search didn't get me to the right place.)




----- Original Message ----
From: David Abrahams <dave at boostpro.com>
To: Development of Python/C++ integration <cplusplus-sig at python.org>
Sent: Tuesday, December 23, 2008 1:10:33 AM
Subject: Re: [C++-sig] Bug and patch for boost.pythonwithenable_shared_from_this


on Mon Dec 22 2008, "Ralf W. Grosse-Kunstleve" <rwgk-AT-yahoo.com> wrote:

> Thanks for the patches! They are now in the boost trunk:
>
> ------------------------------------------------------------------------
> r50368 | rwgk | 2008-12-22 23:55:33 -0800 (Mon, 22 Dec 2008) | 4 lines
>
> Boost.Python enable_shared_from_this patches by Nicolas Lelong and Chad Austin:
>  http://mail.python.org/pipermail/cplusplus-sig/2008-December/014103.html
>  http://mail.python.org/pipermail/cplusplus-sig/2008-February/013003.html

FYI, Ralf, trunk changes don't get merged automatically to release, so
we now need a mechanism to make sure changes we want released are
actually released.

-- 
Dave Abrahams
BoostPro Computing
http://www.boostpro.com
_______________________________________________
Cplusplus-sig mailing list
Cplusplus-sig at python.org
http://mail.python.org/mailman/listinfo/cplusplus-sig