From bidihall at gmail.com  Sun Jan  3 06:12:03 2010
From: bidihall at gmail.com (Blair Hall)
Date: Sun, 3 Jan 2010 18:12:03 +1300
Subject: [C++-sig] boost.build or distutils?
Message-ID: <37e118f61001022112t20b1e389x1166817a19c7732c@mail.gmail.com>

I have recently built a Python module using Boost.Python that gives me
access to a library of C++ classes and functions. It works very nicely :-)
Now I would like to be able to distribute it to friends, etc.

I am confused about the best way to do this.

I am happy to use bjam to build the extension module. I am working on
Windows and have the MSVC9 Express compiler. I also have Python 2.5.4
installed.

Is there some way to use distutils to build a windows installer that will
simply place my binaries in the correct location on a target machine (which
we will assume is also Windows and has some version of Python 2.5.x
installed)? As far as I can see, distutils is used to build the extension
again and I have not had success with this so far (probably need to upgrade
to Python 2.6). It seems silly to be struggling with this problem when bjam
can do the the build without problems.

Can anyone offer advice on how I should proceed?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100103/5fbb5eb1/attachment.htm>

From sipickles at googlemail.com  Mon Jan  4 14:53:41 2010
From: sipickles at googlemail.com (Simon Pickles)
Date: Mon, 04 Jan 2010 13:53:41 +0000
Subject: [C++-sig] Boost Python Tuple - Find length of tuple?
Message-ID: <4B41F2E5.6060002@googlemail.com>

Hello,

I have a simple question. How do I find out in c++ how many elements are 
contained within a boost::python::tuple?

I tried this:

        using namespace boost::python;
        tuple a = make_tuple("hello", 42);
        object b = a.attr("length"); // exception
        unsigned int c = extract<unsigned int>(b);
        assert( c==2 );

AttributeError: 'tuple' object has no attribute 'length'

Am I missing something?

Thanks

Si

From brokenn at gmail.com  Mon Jan  4 15:02:13 2010
From: brokenn at gmail.com (Brian O'Kennedy)
Date: Mon, 4 Jan 2010 14:02:13 +0000
Subject: [C++-sig] Boost Python Tuple - Find length of tuple?
In-Reply-To: <4B41F2E5.6060002@googlemail.com>
References: <4B41F2E5.6060002@googlemail.com>
Message-ID: <cd543dea1001040602i467d3abt765f576cc210792d@mail.gmail.com>

You could try this instead: (untested!)

      using namespace boost::python;
      tuple a = make_tuple("hello", 42);
      std::cout << "Length = " << boost::python::len(a) << std::endl;  //
explicit namespace not needed

hth,
 Brian

2010/1/4 Simon Pickles <sipickles at googlemail.com>

> Hello,
>
> I have a simple question. How do I find out in c++ how many elements are
> contained within a boost::python::tuple?
>
> I tried this:
>
>       using namespace boost::python;
>       tuple a = make_tuple("hello", 42);
>       object b = a.attr("length"); // exception
>       unsigned int c = extract<unsigned int>(b);
>       assert( c==2 );
>
> AttributeError: 'tuple' object has no attribute 'length'
>
> Am I missing something?
>
> Thanks
>
> Si
> _______________________________________________
> 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/20100104/cf7f23b3/attachment.htm>

From seefeld at sympatico.ca  Mon Jan  4 15:08:39 2010
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Mon, 04 Jan 2010 09:08:39 -0500
Subject: [C++-sig] Boost Python Tuple - Find length of tuple?
In-Reply-To: <4B41F2E5.6060002@googlemail.com>
References: <4B41F2E5.6060002@googlemail.com>
Message-ID: <4B41F667.40900@sympatico.ca>

On 01/04/2010 08:53 AM, Simon Pickles wrote:
> Hello,
>
> I have a simple question. How do I find out in c++ how many elements 
> are contained within a boost::python::tuple?
>
> I tried this:
>
>        using namespace boost::python;
>        tuple a = make_tuple("hello", 42);
>        object b = a.attr("length"); // exception
>        unsigned int c = extract<unsigned int>(b);
>        assert( c==2 );
>
> AttributeError: 'tuple' object has no attribute 'length'
>
> Am I missing something?

The attribute is spelled "__len__", not "length", and the pythonic way 
to call it is "len(a)". Thus:

tuple a = ...;
unsigned int c = len(a);

should work fine.

     Stefan

-- 

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


From sipickles at googlemail.com  Mon Jan  4 15:24:50 2010
From: sipickles at googlemail.com (Simon Pickles)
Date: Mon, 04 Jan 2010 14:24:50 +0000
Subject: [C++-sig] Boost Python Tuple - Find length of tuple?
In-Reply-To: <cd543dea1001040602i467d3abt765f576cc210792d@mail.gmail.com>
References: <4B41F2E5.6060002@googlemail.com>
	<cd543dea1001040602i467d3abt765f576cc210792d@mail.gmail.com>
Message-ID: <4B41FA32.8070004@googlemail.com>

On 04/01/2010 14:02, Brian O'Kennedy wrote:
> You could try this instead: (untested!)
>
>       using namespace boost::python;
>       tuple a = make_tuple("hello", 42);
>       std::cout << "Length = " << boost::python::len(a) << std::endl; 
>  // explicit namespace not needed
>
>
Thanks, thats perfect.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100104/c7cc28d1/attachment.htm>

From blp330 at msn.com  Tue Jan  5 07:50:04 2010
From: blp330 at msn.com (blp330)
Date: Mon, 4 Jan 2010 22:50:04 -0800 (PST)
Subject: [C++-sig] [boost.python] How can I link against certain python
 version?
Message-ID: <26984580.post@talk.nabble.com>


Hi,

I try to upgrade my boost from 1.36 to 1.41, but it will link against
python26.lib.

My own python version is 2.5,

How can I tell boost to link with python25.lib?

Thanks.
-- 
View this message in context: http://old.nabble.com/-boost.python--How-can-I-link-against-certain-python-version--tp26984580p26984580.html
Sent from the Python - c++-sig mailing list archive at Nabble.com.


From blake at hailmail.net  Tue Jan  5 15:09:50 2010
From: blake at hailmail.net (Dane Springmeyer)
Date: Tue, 5 Jan 2010 06:09:50 -0800
Subject: [C++-sig] [boost.python] How can I link against certain python
	version?
In-Reply-To: <26984580.post@talk.nabble.com>
References: <26984580.post@talk.nabble.com>
Message-ID: <567C0133-C413-4713-AF48-CEFB655E62A0@hailmail.net>


On Jan 4, 2010, at 10:50 PM, blp330 wrote:

>
> Hi,
>
> I try to upgrade my boost from 1.36 to 1.41, but it will link against
> python26.lib.
>
> My own python version is 2.5,
>
> How can I tell boost to link with python25.lib?
>

What operating system are you on? Mac OS X?

Generally, I on Mac OSX I set up a user-config.jam file and then point  
to it using the bjam options:

bjam --with-python --ignore-site-config --user-config=user-config.jam  
toolset=darwin -d2

My 'user-config.jam' is based on the template that looks like this:

import option ;
import feature ;
if ! darwin in [ feature.values <toolset> ]
{
     using darwin ;
}
project : default-build <toolset>darwin ;
using python
      : %(num)s # version
      : %(system)s/Library/Frameworks/Python.framework/Versions/% 
(num)s/bin/python%(num)s # cmd-or-prefix
      : %(system)s/Library/Frameworks/Python.framework/Versions/% 
(num)s/include/python%(num)s # includes
      : %(system)s/Library/Frameworks/Python.framework/Versions/% 
(num)s/lib/python%(num)s/config # a lib actually symlink
      : <toolset>darwin # condition
      ;
libraries = --with-python ;

If it turns out you are on Mac OSX I can explain those variables in  
the '%()s' syntax a bit more - they are not for bjam but just where  
you might change info depending on which python version you want to  
link against.

Dane

From nico_ml at mgdesign.org  Tue Jan  5 18:26:27 2010
From: nico_ml at mgdesign.org (Nicolas Lelong)
Date: Tue, 05 Jan 2010 18:26:27 +0100
Subject: [C++-sig] boost python calling conventions support patch submission
 (__stdcall, __cdecl, __fastcall)
Message-ID: <4B437643.5000103@mgdesign.org>

An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100105/e1dd8014/attachment-0001.htm>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: boost_python_calling_conventions.patch
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100105/e1dd8014/attachment.asc>

From renatox at gmail.com  Tue Jan  5 18:51:12 2010
From: renatox at gmail.com (Renato Araujo)
Date: Tue, 5 Jan 2010 14:51:12 -0300
Subject: [C++-sig] Avoid Implicitly conversion for non direct conversion
Message-ID: <95291a81001050951u62329b80v7f9dd6d994d80426@mail.gmail.com>

Hi guys,

I'm trying use the "implicitly_convertible" to specify to boost.python
the possible conversions for every type, but I start get some errors
in my functions call because, the boost.python accept conversions in
more then one level.  Like in this code:

The boost.python allow me to call the function "value" passing "int"
as argument type. Because "int" is implicitly convertible to "X" and
"X" is implicitly convertible to "Y".
This make a mess in my functions because boost try converter the
values before try other functions and call the wrong functions.

I would like to know if is possible to avoid this level of conversion
and tell to boost.python only try the direct conversion, because this
make the functions call slower then normal calls, and some times cause
a lot of problems in my classes.



-------------------------------
#include <boost/python/class.hpp>
#include <boost/python/implicit.hpp>
#include <boost/python/module.hpp>
#include <boost/python/def.hpp>

using namespace boost::python;

struct X
{
    X(int x) : v(x) {}
    int v;
};

struct Y
{
    Y(const X& w) :  v(w.v) {}
    int v;
};

int value(Y const& x)
{
    return x.v;
}

BOOST_PYTHON_MODULE(implicit_ext)
{
    def("value", value);

    class_<X>("X",
        init<int>())
        ;

    class_<Y>("Y",
        init<const X&>())
        ;

    implicitly_convertible<X,Y>();
    implicitly_convertible<int,X>();
}
---------------------------------------




BR
Renato Araujo Oliveira Filho

From Terry.Rankine at csiro.au  Wed Jan  6 07:50:42 2010
From: Terry.Rankine at csiro.au (Terry.Rankine at csiro.au)
Date: Wed, 6 Jan 2010 14:50:42 +0800
Subject: [C++-sig] boost 1.41 and autolink (maybe the wrong list)
Message-ID: <4E88C091A57493498EEC9F2A21D5BF4321AECF0923@EXWA-MBX01.nexus.csiro.au>

I have compiled boost, and now want to use it in my python app but I am having trouble with autolink.hpp in my c++ files.

C:\source\esys13>cl /Fo.sconf_temp\conftest_4.obj /c .sconf_temp\conftest_4.cpp /TP /nologo
/IC:\python26\include "/IC: \source\boost_1_41_0"
conftest_4.cpp
C:\source\boost_1_41_0\boost/config/auto_link.hpp(297) : fatal error C1189: #error :  "Mixing a dll boost
 library with a static runtime is a really bad idea..."

I haven't done any linking at this point - I am just trying to compile.

Here is my test file.

.sconf_temp\conftest_4.cpp:
    #include "boost/python.hpp"

Yep - just the include and thats it.

VS2008, and python 2.6

Any ideas what I am doing wrong?

Terry Rankine
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100106/f992b7bc/attachment.htm>

From hans_meine at gmx.net  Wed Jan  6 08:59:40 2010
From: hans_meine at gmx.net (Hans Meine)
Date: Wed, 6 Jan 2010 08:59:40 +0100
Subject: [C++-sig] Avoid Implicitly conversion for non direct conversion
In-Reply-To: <95291a81001050951u62329b80v7f9dd6d994d80426@mail.gmail.com>
References: <95291a81001050951u62329b80v7f9dd6d994d80426@mail.gmail.com>
Message-ID: <201001060859.42133.hans_meine@gmx.net>

On Dienstag 05 Januar 2010, Renato Araujo wrote:
> I would like to know if is possible to avoid this level of conversion
> and tell to boost.python only try the direct conversion,

Not as far as I know.

> because this 
> make the functions call slower then normal calls,

But only if you pass objects that need to be converted, right?

> and some times cause 
> a lot of problems in my classes.

Details?

You might have problems with overloaded functions, in which case Troy's 
recently suggested BPL extension would most likely help.  (I am looking 
forward to its integration).

Ciao, /  /                                                    .o.
     /--/                                                     ..o
    /  / ANS                                                  ooo
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part.
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100106/d5fbd7c5/attachment.pgp>

From sipickles at googlemail.com  Wed Jan  6 12:50:10 2010
From: sipickles at googlemail.com (Simon Pickles)
Date: Wed, 06 Jan 2010 11:50:10 +0000
Subject: [C++-sig] boost 1.41 and autolink (maybe the wrong list)
In-Reply-To: <4E88C091A57493498EEC9F2A21D5BF4321AECF0923@EXWA-MBX01.nexus.csiro.au>
References: <4E88C091A57493498EEC9F2A21D5BF4321AECF0923@EXWA-MBX01.nexus.csiro.au>
Message-ID: <4B4478F2.5070508@googlemail.com>

Hello Terry,

Try this:

.sconf_temp\conftest_4.cpp:

     #define BOOST_PYTHON_DYN_LINK
     #include "boost/python.hpp"

This will make boost look for shared library in the autolink feature

Si

On 06/01/2010 06:50, Terry.Rankine at csiro.au wrote:
>
> I have compiled boost, and now want to use it in my python app but I 
> am having trouble with autolink.hpp in my c++ files.
>
> C:\source\esys13>cl /Fo.sconf_temp\conftest_4.obj /c 
> .sconf_temp\conftest_4.cpp /TP /nologo
>
> /IC:\python26\include "/IC: \source\boost_1_41_0"
>
> conftest_4.cpp
>
> C:\source\boost_1_41_0\boost/config/auto_link.hpp(297) : fatal error 
> C1189: #error :  "Mixing a dll boost
>
>  library with a static runtime is a really bad idea..."
>
> I haven't done any linking at this point -- I am just trying to compile.
>
> Here is my test file.
>
> .sconf_temp\conftest_4.cpp:
>
>     #include "boost/python.hpp"
>
> Yep -- just the include and thats it.
>
> VS2008, and python 2.6
>
> Any ideas what I am doing wrong?
>
> Terry Rankine
>
>
> _______________________________________________
> 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/20100106/4adf9e8e/attachment.htm>

From renatox at gmail.com  Wed Jan  6 14:10:02 2010
From: renatox at gmail.com (Renato Araujo)
Date: Wed, 6 Jan 2010 10:10:02 -0300
Subject: [C++-sig] Avoid Implicitly conversion for non direct conversion
In-Reply-To: <201001060859.42133.hans_meine@gmx.net>
References: <95291a81001050951u62329b80v7f9dd6d994d80426@mail.gmail.com> 
	<201001060859.42133.hans_meine@gmx.net>
Message-ID: <95291a81001060510v1ed7c6bdy741e6334eabe8c59@mail.gmail.com>

Hi thanks for you help,

On Wed, Jan 6, 2010 at 4:59 AM, Hans Meine <hans_meine at gmx.net> wrote:
> On Dienstag 05 Januar 2010, Renato Araujo wrote:
>> I would like to know if is possible to avoid this level of conversion
>> and tell to boost.python only try the direct conversion,
>
> Not as far as I know.
>
>> because this
>> make the functions call slower then normal calls,
>
> But only if you pass objects that need to be converted, right?
Yes but this is the user level I would like to avoid this.


>
>> and some times cause
>> a lot of problems in my classes.
>
> Details?


I'm working on PySide (python bindings for Qt), and I have a lot of
classes and a lot of implicitly conversions.

Example:

I have a class QVariant (Docs:
http://doc.trolltech.com/4.6/qvariant.html) with many constructos, adn
each type has your implicitly conversion rule.
One example is QCorlor (Docs:
http://doc.trolltech.com/4.6/qcolor.html) this is implicitly
conversible from "int".

Then when I create a QVariant(1) (passing int as argument)
boost.python call the QVariant(QColor) constructor. I think this
happening because boost.python try the implicitly conversion before
verify the other basics signatures. This is one of the case I have a
lot of others because I have more then 100 classes and more then 200
implicitly conversions rules, it's almost impossible to me,  manager
this if the conversion can happen in every class level.

>
> You might have problems with overloaded functions, in which case Troy's
> recently suggested BPL extension would most likely help. ?(I am looking
> forward to its integration).

Where I can get more info about this I searched in ML but I did not
find any info related.

>
> Ciao, / ?/ ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?.o.
> ? ? /--/ ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ..o
> ? ?/ ?/ ANS ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?ooo


Thanks

From troy at resophonic.com  Wed Jan  6 16:09:54 2010
From: troy at resophonic.com (troy d. straszheim)
Date: Wed, 06 Jan 2010 10:09:54 -0500
Subject: [C++-sig] Avoid Implicitly conversion for non direct conversion
In-Reply-To: <95291a81001050951u62329b80v7f9dd6d994d80426@mail.gmail.com>
References: <95291a81001050951u62329b80v7f9dd6d994d80426@mail.gmail.com>
Message-ID: <4B44A7C2.2060306@resophonic.com>

Renato Araujo wrote:
> Hi guys,
> 
> I'm trying use the "implicitly_convertible" to specify to boost.python
> the possible conversions for every type, but I start get some errors
> in my functions call because, the boost.python accept conversions in
> more then one level.  Like in this code:
> 
> The boost.python allow me to call the function "value" passing "int"
> as argument type. Because "int" is implicitly convertible to "X" and
> "X" is implicitly convertible to "Y".
> This make a mess in my functions because boost try converter the
> values before try other functions and call the wrong functions.
> 
> I would like to know if is possible to avoid this level of conversion
> and tell to boost.python only try the direct conversion, because this
> make the functions call slower then normal calls, and some times cause
> a lot of problems in my classes.
> 
> 
> 
> -------------------------------
> #include <boost/python/class.hpp>
> #include <boost/python/implicit.hpp>
> #include <boost/python/module.hpp>
> #include <boost/python/def.hpp>
> 
> using namespace boost::python;
> 
> struct X
> {
>     X(int x) : v(x) {}
>     int v;
> };
> 
> struct Y
> {
>     Y(const X& w) :  v(w.v) {}
>     int v;
> };
> 
> int value(Y const& x)
> {
>     return x.v;
> }
> 
> BOOST_PYTHON_MODULE(implicit_ext)
> {
>     def("value", value);
> 
>     class_<X>("X",
>         init<int>())
>         ;
> 
>     class_<Y>("Y",
>         init<const X&>())
>         ;
> 
>     implicitly_convertible<X,Y>();
>     implicitly_convertible<int,X>();
> }
> ---------------------------------------
> 

Hi,

I'm in a rather elliptical orbit at the moment but should be coming back 
around soon.  I'll have a look at this:  if we're going to attempt to 
fix overloading, we might as well try to make boost.python obey the c++ 
rule of 'no more than 1 user-defined conversion in a row'.

Renato, the overloading stuff mentioned by Hans is in a thread titled 
"Implementation of proper overload resolution" on this list.

-t





From troy at resophonic.com  Wed Jan  6 16:10:54 2010
From: troy at resophonic.com (troy d. straszheim)
Date: Wed, 06 Jan 2010 10:10:54 -0500
Subject: [C++-sig] function types as user-defined converters
In-Reply-To: <200912282143.58716.lists_ravi@lavabit.com>
References: <55154.65.220.96.226.1257193994.squirrel@lavabit.com>	<4AF47E91.9070503@resophonic.com>
	<200912282143.58716.lists_ravi@lavabit.com>
Message-ID: <4B44A7FE.10408@resophonic.com>

Ravi wrote:
> On Friday 06 November 2009 14:52:49 troy d. straszheim wrote:
>> Currently, the converter type still leaks out to python in the signature:
>>
>>      class S(Boost.Python.instance)
>>       |  add(...)
>>       |      add( (S)arg1, (object)arg2) -> None :
>>       |
>>       |          C++ signature :
>>       |              void add(S*,to_optional<int> 
>> (*)(boost::python::api::object))
>>
>>
>> converters won't nest:
>>
>>    def("foo", as<returnvalue(arg1, conv1(conv2(arg2)))>(&somefn)); // nah
>>
>> I dont have a mechanism for hooking a user defined converter into the
>> return type.  I'm going to have to think about that one a bit more.
> 
> Did you have any ideas on this, Troy? This is the last bit I think I'd need in 
> order to replace my code with yours.
> 

Hi Ravi,

I haven't looked, I got pulled away.  I hope to be back to this soon though.

-t


From hans_meine at gmx.net  Wed Jan  6 16:17:45 2010
From: hans_meine at gmx.net (Hans Meine)
Date: Wed, 6 Jan 2010 16:17:45 +0100
Subject: [C++-sig] Avoid Implicitly conversion for non direct conversion
In-Reply-To: <95291a81001060510v1ed7c6bdy741e6334eabe8c59@mail.gmail.com>
References: <95291a81001050951u62329b80v7f9dd6d994d80426@mail.gmail.com>
	<201001060859.42133.hans_meine@gmx.net>
	<95291a81001060510v1ed7c6bdy741e6334eabe8c59@mail.gmail.com>
Message-ID: <201001061617.46520.hans_meine@gmx.net>

Am Mittwoch, 06. Januar 2010 14:10:02 schrieb Renato Araujo:
> On Wed, Jan 6, 2010 at 4:59 AM, Hans Meine <hans_meine at gmx.net> wrote:
> > But only if you pass objects that need to be converted, right?
>
> Yes but this is the user level I would like to avoid this.

Ah, you fear users saying "PySide is slow" because they're unknowingly using 
implicit conversions?  How slow is this really?  IMO BPL method calls are not 
blazingly fast anyway, since there is a lot of stuff happening at runtime, so 
one should not try to call millions of methods in the first place.

I am not convinced there's a problem, so maybe you could try to ignore it for 
now.

> I have a class QVariant (Docs:
> http://doc.trolltech.com/4.6/qvariant.html) with many constructos, adn
> each type has your implicitly conversion rule.
> One example is QCorlor (Docs:
> http://doc.trolltech.com/4.6/qcolor.html) this is implicitly
> conversible from "int".
>
> Then when I create a QVariant(1) (passing int as argument)
> boost.python call the QVariant(QColor) constructor. I think this
> happening because boost.python try the implicitly conversion before
> verify the other basics signatures.

Yes, but here the implicit conversion is not the problem IMO, but the fact 
that BPL tries the overloaded functions one-by-one and uses the first 
matching, instead of the "best" one.

> This is one of the case I have a
> lot of others because I have more then 100 classes and more then 200
> implicitly conversions rules, it's almost impossible to me,  manager
> this if the conversion can happen in every class level.

I see, this is exactly the problem that I had in mind when I wrote about the 
overloading problem.  (Without overloading, implicit conversion should not bug 
you, right?)

> > You might have problems with overloaded functions, in which case Troy's
> > recently suggested BPL extension would most likely help.  (I am looking
> > forward to its integration).
>
> Where I can get more info about this I searched in ML but I did not
> find any info related.

Cf. recent thread "Implementation of proper overload resolution".

HTH,
  Hans

From renatox at gmail.com  Wed Jan  6 17:42:59 2010
From: renatox at gmail.com (Renato Araujo)
Date: Wed, 6 Jan 2010 13:42:59 -0300
Subject: [C++-sig] Implementation of proper overload resolution
In-Reply-To: <874onjd67y.fsf@resophonic.com>
References: <87aaxha8ih.fsf@resophonic.com>
	<0EFC2868-BA76-4E73-8DF2-075C2D73ABD9@hailmail.net> 
	<9D77FF39-BA7A-476F-9CBF-48009A1D7919@hailmail.net>
	<874onjd67y.fsf@resophonic.com>
Message-ID: <95291a81001060842i68c59b00we79150aca3620301@mail.gmail.com>

This is great,  I will try this, but I'm afraid about the size,
because my main problem with boost.python is the library size compared
to others c++ python bindings, my current library has 16 MB compared
to 5 MB from SIP, and in my bindings I have a lot of overload
functions and this solution can produce some new bytes. :(

Renato Araujo Oliveira Filho


On Tue, Dec 22, 2009 at 12:42 AM, Troy D. Straszheim
<troy at resophonic.com> wrote:
> Dane Springmeyer <blake at hailmail.net> writes:
>
>> Troy,
>>
>> Incidentally, do you know the proper way ?(or is there a proper way?)
>> to support None type keyword arguments?
>>
>> Cheers,
>>
>> Dane
>>
>
> Could you elaborate? ?What're you trying to do?
>
> -t
>
>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>

From nico_ml at mgdesign.org  Wed Jan  6 18:09:18 2010
From: nico_ml at mgdesign.org (Nicolas Lelong)
Date: Wed, 06 Jan 2010 18:09:18 +0100
Subject: [C++-sig] boost 1.41 and autolink (maybe the wrong list)
In-Reply-To: <4E88C091A57493498EEC9F2A21D5BF4321AECF0923@EXWA-MBX01.nexus.csiro.au>
References: <4E88C091A57493498EEC9F2A21D5BF4321AECF0923@EXWA-MBX01.nexus.csiro.au>
Message-ID: <4B44C3BE.1090606@mgdesign.org>

An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100106/badc72e8/attachment.htm>

From tim at tizmoi.net  Wed Jan  6 23:33:33 2010
From: tim at tizmoi.net (Tim Couper)
Date: Wed, 06 Jan 2010 22:33:33 +0000
Subject: [C++-sig] Boost-python wrapping a vector of vectors?
Message-ID: <4B450FBD.4040104@tizmoi.net>


I'm trying to boost-python a vector-of-vectors, like

class A
{
public
    A(const std::vector<std::vector double >>& my_array);
};

and would intuitively write the wrapper:

BOOST_PYTHON_MODULE(foo)
{
using namespace boost::python

class_<A>("A")
     .def(init(std::vector<std::vector<double> >())
     ;

but get the error "a call to a constructor cannot appear in a constant 
expression"

Can anyone supply any guidance as to how to wrap this? Thanks

Tim

Dr Tim Couper

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100106/3ee301ee/attachment.htm>

From troy at resophonic.com  Thu Jan  7 00:16:26 2010
From: troy at resophonic.com (troy d. straszheim)
Date: Wed, 06 Jan 2010 18:16:26 -0500
Subject: [C++-sig] Boost-python wrapping a vector of vectors?
In-Reply-To: <4B450FBD.4040104@tizmoi.net>
References: <4B450FBD.4040104@tizmoi.net>
Message-ID: <4B4519CA.4040502@resophonic.com>

Tim Couper wrote:
> 
> I'm trying to boost-python a vector-of-vectors, like
> 
> class A
> {
> public
>    A(const std::vector<std::vector double >>& my_array);
> };
> 
> and would intuitively write the wrapper:
> 
> BOOST_PYTHON_MODULE(foo)
> {
> using namespace boost::python
> 
> class_<A>("A")
>     .def(init(std::vector<std::vector<double> >())
>     ;
> 
> but get the error "a call to a constructor cannot appear in a constant 
> expression"

Syntax error, init takes a type argument:  init<T>()

-t


From tim at tizmoi.net  Thu Jan  7 10:21:50 2010
From: tim at tizmoi.net (Tim Couper)
Date: Thu, 07 Jan 2010 09:21:50 +0000
Subject: [C++-sig] Boost-python wrapping a vector of vectors?
In-Reply-To: <4B4519CA.4040502@resophonic.com>
References: <4B450FBD.4040104@tizmoi.net> <4B4519CA.4040502@resophonic.com>
Message-ID: <4B45A7AE.3030805@tizmoi.net>

Troy

Thanks for the prompt reply .. I'm a python expert and a C++ novice ... 
Apologies for not getting this ...

I've seen I can define a python mapping using the vector_indexing suite

class_< std::vector< std::vector<double> > >("vector_double2")
      .def(vector_indexing_suite<std::vector< std::vector<double> > >

but I can't see how to get that information into the init argument in 
the C++

class_<A>("A")
     .def(init<vector_double2>( ))

as vector_double2 isn't known when the c++ compiles, so this clearly 
isn't the "type" argument to which you refer. Clearly there's something 
I'm missing. How do I define a type? Should I be using the named 
constructors/factories described in 
http://wiki.python.org/moin/boost.python/HowTo? If so how?

Thanks again

Tim






On 06/01/2010 23:16, troy d. straszheim wrote:
> Tim Couper wrote:
>>
>> I'm trying to boost-python a vector-of-vectors, like
>>
>> class A
>> {
>> public
>>    A(const std::vector<std::vector double >>& my_array);
>> };
>>
>> and would intuitively write the wrapper:
>>
>> BOOST_PYTHON_MODULE(foo)
>> {
>> using namespace boost::python
>>
>> class_<A>("A")
>>     .def(init(std::vector<std::vector<double> >())
>>     ;
>>
>> but get the error "a call to a constructor cannot appear in a 
>> constant expression"
>
> Syntax error, init takes a type argument:  init<T>()
>
> -t
>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>
>
>
> No virus found in this incoming message.
> Checked by AVG - www.avg.com
> Version: 9.0.725 / Virus Database: 270.14.127/2603 - Release Date: 01/06/10 07:35:00
>
>    
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100107/56b2bd54/attachment.htm>

From tim at tizmoi.net  Thu Jan  7 11:21:11 2010
From: tim at tizmoi.net (Tim Couper)
Date: Thu, 07 Jan 2010 10:21:11 +0000
Subject: [C++-sig] Boost-python wrapping a vector of vectors?
In-Reply-To: <4B45A7AE.3030805@tizmoi.net>
References: <4B450FBD.4040104@tizmoi.net> <4B4519CA.4040502@resophonic.com>
	<4B45A7AE.3030805@tizmoi.net>
Message-ID: <4B45B597.3050909@tizmoi.net>

Ah I understand .. so will reply to my own posting for others new to pb ...

C++ constructors with one argument behave differently I understand, so 
taking a 2 argument case:

class A
{
public
    A(int i, const std::vector<std::vector double >>& my_array);
};


In the BOOST_PYTHON_MODULE(foo)  ..

define a type using typedef:

     typedef std::vector<std::vector<double> > double_trouble;

and add the python wrappers we're going to need - one for wrapping a 
vector of doubles, and one for the vector of vectors ..

     class_<std::vector<std::vector<double> > > ("v2_double")
         .def(vector_indexing_suite<std::vector<std::vector<double> > >());

     class_<std::vector<double> > ("v_double")
         .def(vector_indexing_suite<std::vector<double> >());

now define the constructor like Troy said ..

     class_<duh::A>("A")
         .def(init<int, double_trouble >())
         ;


bjam this lot and we then can write out nosetest tests:

def test_v_double():
     v = foo.v_double()
     v[:] = [1.4,2.5]
     assert_equals(len(v),2)
     assert_almost_equal(v[0],1.4)


def test_v2_double():
     v2 = foo.v2_double()
     v1a = foo.v_double()
     v2a = foo.v_double()
     v1a[:] = [1.4, 2.5, 5.6]
     v2a[:] = [3.4, 2.8, 7.8, 12.5]

     v2[:] = [v1a,v2a]
     assert_equals(len(v2),2)
     assert_almost_equal(v2[0][0],1.4)

def test_A():
     v2 = foo.v2_double()
     v1a = foo.v_double()
     v2a = foo.v_double()
     v1a[:] = [1.4, 2.5, 5.6]
     v2a[:] = [3.4, 2.8, 7.8, 12.5]

     v2[:] = [v1a,v2a]
     res = foo.A(1, v2)
     assert isinstance(res, foo.A)

and there we have it ..

I'll be back ... :-)

Tim

On 07/01/2010 09:21, Tim Couper wrote:
> Troy
>
> Thanks for the prompt reply .. I'm a python expert and a C++ novice 
> ... Apologies for not getting this ...
>
> I've seen I can define a python mapping using the vector_indexing suite
>
> class_< std::vector< std::vector<double> > >("vector_double2")
>      .def(vector_indexing_suite<std::vector< std::vector<double> > >
>
> but I can't see how to get that information into the init argument in 
> the C++
>
> class_<A>("A")
>     .def(init<vector_double2>( ))
>
> as vector_double2 isn't known when the c++ compiles, so this clearly 
> isn't the "type" argument to which you refer. Clearly there's 
> something I'm missing. How do I define a type? Should I be using the 
> named constructors/factories described in 
> http://wiki.python.org/moin/boost.python/HowTo? If so how?
>
> Thanks again
>
> Tim
>
>
>
>
>
>
> On 06/01/2010 23:16, troy d. straszheim wrote:
>> Tim Couper wrote:
>>>
>>> I'm trying to boost-python a vector-of-vectors, like
>>>
>>> class A
>>> {
>>> public
>>>    A(const std::vector<std::vector double >>& my_array);
>>> };
>>>
>>> and would intuitively write the wrapper:
>>>
>>> BOOST_PYTHON_MODULE(foo)
>>> {
>>> using namespace boost::python
>>>
>>> class_<A>("A")
>>>     .def(init(std::vector<std::vector<double> >())
>>>     ;
>>>
>>> but get the error "a call to a constructor cannot appear in a 
>>> constant expression"
>>
>> Syntax error, init takes a type argument:  init<T>()
>>
>> -t
>>
>> _______________________________________________
>> Cplusplus-sig mailing list
>> Cplusplus-sig at python.org
>> http://mail.python.org/mailman/listinfo/cplusplus-sig
>>
>>
>>
>> No virus found in this incoming message.
>> Checked by AVG -www.avg.com
>> Version: 9.0.725 / Virus Database: 270.14.127/2603 - Release Date: 01/06/10 07:35:00
>>
>>    
>
>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>
>
>
> No virus found in this incoming message.
> Checked by AVG - www.avg.com
> Version: 9.0.725 / Virus Database: 270.14.128/2604 - Release Date: 01/06/10 19:35:00
>
>    
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100107/d132b969/attachment-0001.htm>

From blp330 at msn.com  Thu Jan  7 13:13:19 2010
From: blp330 at msn.com (blp330)
Date: Thu, 7 Jan 2010 04:13:19 -0800 (PST)
Subject: [C++-sig] [boost.python] How can I link against certain python
 version?
In-Reply-To: <567C0133-C413-4713-AF48-CEFB655E62A0@hailmail.net>
References: <26984580.post@talk.nabble.com>
	<567C0133-C413-4713-AF48-CEFB655E62A0@hailmail.net>
Message-ID: <27059056.post@talk.nabble.com>


Thank you. But my operating system is Windows. :p

I am sorry I didn't describe my problem clearly.

I am using boostpro's compiled binary. I defined BOOST_PYTHON_STATIC_LIB.

Visual Studio will try to link python25.lib automatically in 1.36.
Now, I don't know how to make it still link against python25.lib after
upgrade to 1.41.

Thanks.
-- 
View this message in context: http://old.nabble.com/-boost.python--How-can-I-link-against-certain-python-version--tp26984580p27059056.html
Sent from the Python - c++-sig mailing list archive at Nabble.com.


From dwwkelly at gmail.com  Sat Jan  9 19:52:48 2010
From: dwwkelly at gmail.com (devin kelly)
Date: Sat, 9 Jan 2010 13:52:48 -0500
Subject: [C++-sig] Exposing C++ data with Boost.Python
Message-ID: <c57e4bcc1001091052q176c8b05xedadebbb8776514@mail.gmail.com>

Hello,

I'm trying to expose some data that I develop in C++ to python.  Basically,
the reverse of this sample code:

#include <iostream>
#include <python2.6/Python.h>
#include <boost/python.hpp>
#include <boost/python/exec.hpp>

int main(){

        Py_Initialize();
        object main_module = import("__main__");
        object main_namespace = main_module.attr("__dict__");
        ignored = exec("result = 5 ** 2", main_namespace);
        int five_squared = extract<int>(main_namespace["result"]);
        std::cout << five_squared << std::endl;


        return 0;
}


So this code starts the python interpreter, squares 5 (in python) and then
extracts the result to an int called five_squared.  This works fine for me,
it's basically an example straight out of the boost.python webpage.

What I'd really like to do though is have an int that I initialize in C++
and then square in python.  So this would require me to pass (or expose)
that data to python.  I've been trying this for a while and have had no luck
whatsoever.  The best I can think of is code like this:

int main(){
        Py_Initialize();
        object main_module = import("__main__");
        object main_namespace = main_module.attr("__dict__");
        main_namespace["num2square"] = 6;
        ignored = exec("result = num2square ** 2", main_namespace);
        int five_squared = extract<int>(main_namespace["result"]);
        std::cout << five_squared <<
std::endl;
        return 0;
}

This doesn't work.  Python throws an error.  What am I doing wrong??

Thanks for any help,
Devin


-- 
http://users.wpi.edu/~dkelly/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100109/1e9a4118/attachment.htm>

From seefeld at sympatico.ca  Sat Jan  9 20:55:08 2010
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Sat, 09 Jan 2010 14:55:08 -0500
Subject: [C++-sig] Exposing C++ data with Boost.Python
In-Reply-To: <c57e4bcc1001091052q176c8b05xedadebbb8776514@mail.gmail.com>
References: <c57e4bcc1001091052q176c8b05xedadebbb8776514@mail.gmail.com>
Message-ID: <4B48DF1C.9090900@sympatico.ca>

On 01/09/2010 01:52 PM, devin kelly wrote:
> Hello,
>
> I'm trying to expose some data that I develop in C++ to python.  
> Basically, the reverse of this sample code:
>
> #include <iostream>
> #include <python2.6/Python.h>
> #include <boost/python.hpp>
> #include <boost/python/exec.hpp>
>
> int main(){
>
>         Py_Initialize();
>         object main_module = import("__main__");
>         object main_namespace = main_module.attr("__dict__");
>         ignored = exec("result = 5 ** 2", main_namespace);
>         int five_squared = extract<int>(main_namespace["result"]);
>         std::cout << five_squared << std::endl;
>
>         return 0;
> }
>
>
> So this code starts the python interpreter, squares 5 (in python) and 
> then extracts the result to an int called five_squared.  This works 
> fine for me, it's basically an example straight out of the 
> boost.python webpage.
>
> What I'd really like to do though is have an int that I initialize in 
> C++ and then square in python.  So this would require me to pass (or 
> expose) that data to python.  I've been trying this for a while and 
> have had no luck whatsoever.  The best I can think of is code like this:
>
> int main(){
>         Py_Initialize();
>         object main_module = import("__main__");
>         object main_namespace = main_module.attr("__dict__");
>         main_namespace["num2square"] = 6;
>         ignored = exec("result = num2square ** 2", main_namespace);
>         int five_squared = extract<int>(main_namespace["result"]);
>         std::cout << five_squared << std::endl;
>         return 0;
> }
>
> This doesn't work.  Python throws an error.  What am I doing wrong??

It might help indicating what error Python actually throws.

Also, if all you want is to evaluate an expression, I'd suggest you use 
"eval()", not "exec()".

         Stefan

-- 

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


From lists_ravi at lavabit.com  Sun Jan 10 17:18:25 2010
From: lists_ravi at lavabit.com (Ravi)
Date: Sun, 10 Jan 2010 11:18:25 -0500
Subject: [C++-sig] boost python calling conventions support patch
	submission (__stdcall, __cdecl, __fastcall)
In-Reply-To: <4B437643.5000103@mgdesign.org>
References: <4B437643.5000103@mgdesign.org>
Message-ID: <201001101118.25701.lists_ravi@lavabit.com>

On Tuesday 05 January 2010 12:26:27 Nicolas Lelong wrote:
> Is this mailing list the best route to submit this patch, what more work
>  should be done to get it accepted into trunk ?

Please file a trac ticket at svn.boost.org so that this does not get lost.

Ravi


From dwwkelly at gmail.com  Sun Jan 10 18:36:49 2010
From: dwwkelly at gmail.com (devin kelly)
Date: Sun, 10 Jan 2010 12:36:49 -0500
Subject: [C++-sig] Exposing C++ data with Boost.Python
In-Reply-To: <4B48DF1C.9090900@sympatico.ca>
References: <c57e4bcc1001091052q176c8b05xedadebbb8776514@mail.gmail.com> 
	<4B48DF1C.9090900@sympatico.ca>
Message-ID: <c57e4bcc1001100936k57e2a494r5f82f5c3a0e0d1c@mail.gmail.com>

Well what I really to do is turn an STL vector into and python list.  I want
to do this in a separate script, so I don't think eval is right for me.  I
just started with an int because that's easier.  I've gotten that part too,
my code looks like this:

int main(){
        int
five_squared=0;

        int a =3;
        try {
                Py_Initialize();
                object main_module = import("__main__");
                object main_namespace = main_module.attr("__dict__");
                main_namespace["var"]=a;
                object ignored = exec("result = 5 ** var", main_namespace);
                five_squared = extract<int>(main_namespace["result"]);
        } catch( error_already_set ) {
                PyErr_Print();
        }
        std::cout << five_squared << std::endl;
        return 0;
}

This code works just like you expect it to, it prints 5^3 or 125.

However I try to change my code to this

int main(){
        int
five_squared=0;

        std::vector<int> v(2);
        v[0]=2;
        v[1]=3;
        try {
                Py_Initialize();
                object main_module = import("__main__");
                object main_namespace = main_module.attr("__dict__");
                main_namespace["var"]=v;
                object ignored = exec("result = 5 ** var[1]",
main_namespace);
                five_squared = extract<int>(main_namespace["result"]);
        } catch( error_already_set ) {
                PyErr_Print();
        }
        std::cout << five_squared << std::endl;
        return 0;
}

I want this to print 5^2 or 25.  But instead I get this error:

TypeError: No to_python (by-value) converter found for C++ type:
std::vector<int, std::allocator<int> >

Which makes sense, python doesn't know how to handle stdd::vector.  So I
have to make something to do this conversion.  This is where I'm stuck now,
I think I have to do something like this

BOOST_PYTHON_MODULE(vector_indexing_suite_ext){
        boost::python::class_<std::vector<double> >("PyVec")
        .def(boost::python::vector_indexing_suite<std::vector<double> >());
}

Is this all I need?  Do I need to define __getitem__ ?  If this is all I
need how do I use it in main()?

Also does anyone know of any good resources for embedding in boost.python?

Thanks!

On Sat, Jan 9, 2010 at 2:55 PM, Stefan Seefeld <seefeld at sympatico.ca> wrote:

> On 01/09/2010 01:52 PM, devin kelly wrote:
>
>> Hello,
>>
>> I'm trying to expose some data that I develop in C++ to python.
>>  Basically, the reverse of this sample code:
>>
>> #include <iostream>
>> #include <python2.6/Python.h>
>> #include <boost/python.hpp>
>> #include <boost/python/exec.hpp>
>>
>> int main(){
>>
>>        Py_Initialize();
>>        object main_module = import("__main__");
>>        object main_namespace = main_module.attr("__dict__");
>>        ignored = exec("result = 5 ** 2", main_namespace);
>>        int five_squared = extract<int>(main_namespace["result"]);
>>        std::cout << five_squared << std::endl;
>>
>>        return 0;
>> }
>>
>>
>> So this code starts the python interpreter, squares 5 (in python) and then
>> extracts the result to an int called five_squared.  This works fine for me,
>> it's basically an example straight out of the boost.python webpage.
>>
>> What I'd really like to do though is have an int that I initialize in C++
>> and then square in python.  So this would require me to pass (or expose)
>> that data to python.  I've been trying this for a while and have had no luck
>> whatsoever.  The best I can think of is code like this:
>>
>> int main(){
>>        Py_Initialize();
>>        object main_module = import("__main__");
>>        object main_namespace = main_module.attr("__dict__");
>>        main_namespace["num2square"] = 6;
>>        ignored = exec("result = num2square ** 2", main_namespace);
>>        int five_squared = extract<int>(main_namespace["result"]);
>>        std::cout << five_squared << std::endl;
>>        return 0;
>> }
>>
>> This doesn't work.  Python throws an error.  What am I doing wrong??
>>
>
> It might help indicating what error Python actually throws.
>
> Also, if all you want is to evaluate an expression, I'd suggest you use
> "eval()", not "exec()".
>
>        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
>



-- 
http://users.wpi.edu/~dkelly/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100110/0a02e343/attachment.htm>

From seefeld at sympatico.ca  Sun Jan 10 18:53:25 2010
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Sun, 10 Jan 2010 12:53:25 -0500
Subject: [C++-sig] Exposing C++ data with Boost.Python
In-Reply-To: <c57e4bcc1001100936k57e2a494r5f82f5c3a0e0d1c@mail.gmail.com>
References: <c57e4bcc1001091052q176c8b05xedadebbb8776514@mail.gmail.com>
	<4B48DF1C.9090900@sympatico.ca>
	<c57e4bcc1001100936k57e2a494r5f82f5c3a0e0d1c@mail.gmail.com>
Message-ID: <4B4A1415.6020604@sympatico.ca>

On 01/10/2010 12:36 PM, devin kelly wrote:
>  So I have to make something to do this conversion.  This is where I'm 
> stuck now, I think I have to do something like this
>
> BOOST_PYTHON_MODULE(vector_indexing_suite_ext){
>         boost::python::class_<std::vector<double> >("PyVec")
>         .def(boost::python::vector_indexing_suite<std::vector<double> 
> >());
> }
>
> Is this all I need?  Do I need to define __getitem__ ?  If this is all 
> I need how do I use it in main()?

You presumably use it just like any other conversion you have defined. 
Have you looked at
http://www.boost.org/doc/libs/1_41_0/libs/python/doc/v2/indexing.html, 
and tried it out ? Reading the docs is typically the best way to get 
started...

>
> Also does anyone know of any good resources for embedding in boost.python?

This list is a good place to ask questions, for anything not covered by 
the (online) docs. (We may even improve the docs if it turns out 
something important isn't covered.)

     Stefan

-- 

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


From blp330 at msn.com  Mon Jan 11 05:05:05 2010
From: blp330 at msn.com (blp330)
Date: Sun, 10 Jan 2010 20:05:05 -0800 (PST)
Subject: [C++-sig]  [boost.python] How can I wrap operator==?
Message-ID: <27105703.post@talk.nabble.com>


Hi,

I want to wrap operator== for my object, Py++ generates that:

        Document_exposer.def( bp::self != bp::self);
        Document_exposer.def( bp::self == bp::self );

But I want to do something more, so I try to wrap it with:

        Document_exposer.def( 
            bp::self != bp::self
            , &Document__NotEqualWrap );
        Document_exposer.def( 
            bp::self == bp::self
            , &Document__EqualWrap );

It can't be compiled...

How can I do?

Thanks.
-- 
View this message in context: http://old.nabble.com/-boost.python--How-can-I-wrap-operator%3D%3D--tp27105703p27105703.html
Sent from the Python - c++-sig mailing list archive at Nabble.com.


From sipickles at googlemail.com  Tue Jan 12 09:16:54 2010
From: sipickles at googlemail.com (Simon Pickles)
Date: Tue, 12 Jan 2010 08:16:54 +0000
Subject: [C++-sig] Returning a simple char *
Message-ID: <4B4C2FF6.7080307@googlemail.com>

Hello,

I thought I was trying to do something simple.

While wrapping a 3rd party library, I have a c++ class with two member 
function returning pointers like this:

class videoInput
{
public:
    static char * getDeviceName(int deviceID);
    ....
};

and a module like this:

#include "boost//python.hpp"
#include "videoInput.h"

using namespace boost::python;
BOOST_PYTHON_MODULE(pyVideoInput)
{
    class_<videoInput>("videoInput")
        .def("getDeviceName",
            &videoInput::getDeviceName,
            return_value_policy<manage_new_object>())
        .staticmethod("getDeviceName")
        ;
}

Following the guide here 
(http://wiki.python.org/moin/boost.python/PointersAndSmartPointers) I 
thought this was the right way but I get errors:

pyVideoInput.cpp
C:\boost_1_41_0\boost/python/object/make_instance.hpp(24) : error C2027: 
use of undefined type 'boost::STATIC_ASSERTION_FAILURE<x>
'
        with
        [
            x=false
        ]
        C:\boost_1_41_0\boost/python/to_python_indirect.hpp(95) : see 
reference to function template instantiation 'PyObject *boos
t::python::objects::make_instance_impl<T,Holder,Derived>::execute<smart_pointer>(Arg 
&)' being compiled
        with
        [
            T=char,
            Holder=holder_t,
            
Derived=boost::python::objects::make_ptr_instance<char,holder_t>,
            Arg=smart_pointer
        ]

..................

        pyVideoInput.cpp(16) : see reference to function template 
instantiation 'boost::python::class_<W> &boost::python::class_<W
 >::def<char*(__cdecl 
*)(int),boost::python::return_value_policy<ResultConverterGenerator>>(const 
char *,A1,const A2 &)' being comp
iled
        with
        [
            W=videoInput,
            ResultConverterGenerator=boost::python::manage_new_object,
            A1=char *(__cdecl *)(int),
            
A2=boost::python::return_value_policy<boost::python::manage_new_object>
        ]


//////////////////////////

It seems to be attempting to use a smart_ptr?

Can anyone help explain this to me? Many thanks

Simon


From charlessolar at gmail.com  Tue Jan 12 16:08:47 2010
From: charlessolar at gmail.com (Charles Solar)
Date: Tue, 12 Jan 2010 09:08:47 -0600
Subject: [C++-sig] Member function bp::optional? Or workaround?
Message-ID: <708326ee1001120708h62fbaad8hb43db3c5ec37b5cf@mail.gmail.com>

I have a few default parameters in a couple of my member functions, and
these functions do not conform to the format required
forBOOST_PYTHON_MEMBER_FUNCTION_OVERLOAD..
I was wondering how to go about creating those thin wrappers for these
functions.

For example, my class would be something like

struct Foo
{
int bar( const Y* = NULL );
int bar( const X&, const Y* = NULL );
};
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100112/f48e92a1/attachment.htm>

From hans_meine at gmx.net  Tue Jan 12 20:21:33 2010
From: hans_meine at gmx.net (Hans Meine)
Date: Tue, 12 Jan 2010 20:21:33 +0100
Subject: [C++-sig] Member function bp::optional? Or workaround?
In-Reply-To: <708326ee1001120708h62fbaad8hb43db3c5ec37b5cf@mail.gmail.com>
References: <708326ee1001120708h62fbaad8hb43db3c5ec37b5cf@mail.gmail.com>
Message-ID: <201001122021.39062.hans_meine@gmx.net>

On Dienstag 12 Januar 2010, Charles Solar wrote:
> I have a few default parameters in a couple of my member functions, and
> these functions do not conform to the format required
> forBOOST_PYTHON_MEMBER_FUNCTION_OVERLOAD..

I never use BOOST_PYTHON_MEMBER_FUNCTION_OVERLOAD anyway.  You may 
freely .def'ine multiple overloads yourself.  (Or maybe you want to try 
pyplusplus, if you have many such functions.)

-- 
Ciao, /  /                                                    .o.
     /--/                                                     ..o
    /  / ANS                                                  ooo
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part.
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100112/062d2592/attachment.pgp>

From charlessolar at gmail.com  Tue Jan 12 20:46:10 2010
From: charlessolar at gmail.com (Charles Solar)
Date: Tue, 12 Jan 2010 13:46:10 -0600
Subject: [C++-sig] Member function bp::optional? Or workaround?
In-Reply-To: <201001122021.39062.hans_meine@gmx.net>
References: <708326ee1001120708h62fbaad8hb43db3c5ec37b5cf@mail.gmail.com> 
	<201001122021.39062.hans_meine@gmx.net>
Message-ID: <708326ee1001121146n11a748cdj8bbd32e30cfafa61@mail.gmail.com>

Well I want to define the overloads myself anyway, I just do not know how to
properly setup the small wrapper that will work.  In the doc it tells you
how to make flat function wrappers, but nothing on member function
wrappers.  I am unsure how I am supposed to handle the this pointer, for
example.
So idealy what I would like to know is what the macro does so I can manually
write out the wrappers myself.

And I have tried using Py++, it does not handle the default argument, it
just puts the call in there and python still requires the param to be there.


On Tue, Jan 12, 2010 at 1:21 PM, Hans Meine <hans_meine at gmx.net> wrote:

> On Dienstag 12 Januar 2010, Charles Solar wrote:
> > I have a few default parameters in a couple of my member functions, and
> > these functions do not conform to the format required
> > forBOOST_PYTHON_MEMBER_FUNCTION_OVERLOAD..
>
> I never use BOOST_PYTHON_MEMBER_FUNCTION_OVERLOAD anyway.  You may
> freely .def'ine multiple overloads yourself.  (Or maybe you want to try
> pyplusplus, if you have many such functions.)
>
> --
> Ciao, /  /                                                    .o.
>     /--/                                                     ..o
>    /  / ANS                                                  ooo
>
> _______________________________________________
> 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/20100112/6376fe46/attachment.htm>

From roman.yakovenko at gmail.com  Tue Jan 12 20:55:16 2010
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 12 Jan 2010 21:55:16 +0200
Subject: [C++-sig] [boost.python] How can I wrap operator==?
In-Reply-To: <27105703.post@talk.nabble.com>
References: <27105703.post@talk.nabble.com>
Message-ID: <7465b6171001121155s25319ffbt724aac143530af2f@mail.gmail.com>

On Mon, Jan 11, 2010 at 6:05 AM, blp330 <blp330 at msn.com> wrote:
>
> Hi,
>
> I want to wrap operator== for my object, Py++ generates that:
>
> ? ? ? ?Document_exposer.def( bp::self != bp::self);
> ? ? ? ?Document_exposer.def( bp::self == bp::self );
>
> But I want to do something more, so I try to wrap it with:
>
> ? ? ? ?Document_exposer.def(
> ? ? ? ? ? ?bp::self != bp::self
> ? ? ? ? ? ?, &Document__NotEqualWrap );
> ? ? ? ?Document_exposer.def(
> ? ? ? ? ? ?bp::self == bp::self
> ? ? ? ? ? ?, &Document__EqualWrap );
>
> It can't be compiled...
>
> How can I do?

It is not clear what you are asking for.

If the "Document" class has special function for comparison, than may
be you can define few global operators ( == and != ).

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

From troy at resophonic.com  Tue Jan 12 21:01:53 2010
From: troy at resophonic.com (troy d. straszheim)
Date: Tue, 12 Jan 2010 15:01:53 -0500
Subject: [C++-sig] Member function bp::optional? Or workaround?
In-Reply-To: <708326ee1001121146n11a748cdj8bbd32e30cfafa61@mail.gmail.com>
References: <708326ee1001120708h62fbaad8hb43db3c5ec37b5cf@mail.gmail.com>
	<201001122021.39062.hans_meine@gmx.net>
	<708326ee1001121146n11a748cdj8bbd32e30cfafa61@mail.gmail.com>
Message-ID: <4B4CD531.3060802@resophonic.com>

Charles Solar wrote:
> Well I want to define the overloads myself anyway, I just do not know 
> how to properly setup the small wrapper that will work.  In the doc it 
> tells you how to make flat function wrappers, but nothing on member 
> function wrappers.  I am unsure how I am supposed to handle the this 
> pointer, for example.
> So idealy what I would like to know is what the macro does so I can 
> manually write out the wrappers myself.
> 
> And I have tried using Py++, it does not handle the default argument, it 
> just puts the call in there and python still requires the param to be there.
> 

Personally I hate that overload-wrapping macro thing.  Here's one way to 
do it:

struct T
{
   void bar(int i) { cout << "bar(" << i << ")\n"; }
   void bar() { cout << "bar\n"; }
};

//
// use 'self' for 'this' for the sake of pythonicism
//
void bar_int (T* self, int i) { self->bar(i); }
void bar_void(T* self)        { self->bar();  }

BOOST_PYTHON_MODULE(foop)
{
   class_<T>("T")
     .def("bar", bar_int)
     .def("bar", bar_void)
     ;
}

-t


From charlessolar at gmail.com  Tue Jan 12 21:06:28 2010
From: charlessolar at gmail.com (Charles Solar)
Date: Tue, 12 Jan 2010 14:06:28 -0600
Subject: [C++-sig] Member function bp::optional? Or workaround?
In-Reply-To: <4B4CD531.3060802@resophonic.com>
References: <708326ee1001120708h62fbaad8hb43db3c5ec37b5cf@mail.gmail.com> 
	<201001122021.39062.hans_meine@gmx.net>
	<708326ee1001121146n11a748cdj8bbd32e30cfafa61@mail.gmail.com> 
	<4B4CD531.3060802@resophonic.com>
Message-ID: <708326ee1001121206j46e89b7fyf9db1d8c63f93c13@mail.gmail.com>

Looks perfect, thanks much.

On Tue, Jan 12, 2010 at 2:01 PM, troy d. straszheim <troy at resophonic.com>wrote:

> Charles Solar wrote:
>
>> Well I want to define the overloads myself anyway, I just do not know how
>> to properly setup the small wrapper that will work.  In the doc it tells you
>> how to make flat function wrappers, but nothing on member function wrappers.
>>  I am unsure how I am supposed to handle the this pointer, for example.
>> So idealy what I would like to know is what the macro does so I can
>> manually write out the wrappers myself.
>>
>> And I have tried using Py++, it does not handle the default argument, it
>> just puts the call in there and python still requires the param to be there.
>>
>>
> Personally I hate that overload-wrapping macro thing.  Here's one way to do
> it:
>
> struct T
> {
>  void bar(int i) { cout << "bar(" << i << ")\n"; }
>  void bar() { cout << "bar\n"; }
> };
>
> //
> // use 'self' for 'this' for the sake of pythonicism
> //
> void bar_int (T* self, int i) { self->bar(i); }
> void bar_void(T* self)        { self->bar();  }
>
> BOOST_PYTHON_MODULE(foop)
> {
>  class_<T>("T")
>    .def("bar", bar_int)
>    .def("bar", bar_void)
>    ;
> }
>
> -t
>
>
> _______________________________________________
> 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/20100112/93697a36/attachment-0001.htm>

From roman.yakovenko at gmail.com  Tue Jan 12 21:09:53 2010
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 12 Jan 2010 22:09:53 +0200
Subject: [C++-sig] Member function bp::optional? Or workaround?
In-Reply-To: <708326ee1001121146n11a748cdj8bbd32e30cfafa61@mail.gmail.com>
References: <708326ee1001120708h62fbaad8hb43db3c5ec37b5cf@mail.gmail.com>
	<201001122021.39062.hans_meine@gmx.net>
	<708326ee1001121146n11a748cdj8bbd32e30cfafa61@mail.gmail.com>
Message-ID: <7465b6171001121209r4dcd8357ra394226080ee4d3@mail.gmail.com>

On Tue, Jan 12, 2010 at 9:46 PM, Charles Solar <charlessolar at gmail.com> wrote:
> And I have tried using Py++, it does not handle the default argument, it
> just puts the call in there and python still requires the param to be there.

Can you explain what you mean?

The following code was generated by Py++:

    bp::class_< overloads_macro::Foo >( "Foo")
        .def(
            "bar"
            , (int ( ::overloads_macro::Foo::* )( ::overloads_macro::Y
const * ) )( &::overloads_macro::Foo::bar )
            , ( bp::arg("arg0")=bp::object() ) )
        .def(
            "bar"
            , (int ( ::overloads_macro::Foo::* )( ::overloads_macro::X
const &,::overloads_macro::Y const * ) )( &::overloads_macro::Foo::bar
)
            , ( bp::arg("arg0"), bp::arg("arg1")=bp::object() ) );


as you can see ( and try ) the generated code does not "requires" the
user to pass default arguments.

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

From hans_meine at gmx.net  Tue Jan 12 21:09:39 2010
From: hans_meine at gmx.net (Hans Meine)
Date: Tue, 12 Jan 2010 21:09:39 +0100
Subject: [C++-sig] Member function bp::optional? Or workaround?
In-Reply-To: <708326ee1001121146n11a748cdj8bbd32e30cfafa61@mail.gmail.com>
References: <708326ee1001120708h62fbaad8hb43db3c5ec37b5cf@mail.gmail.com>
	<201001122021.39062.hans_meine@gmx.net>
	<708326ee1001121146n11a748cdj8bbd32e30cfafa61@mail.gmail.com>
Message-ID: <201001122109.40705.hans_meine@gmx.net>

On Dienstag 12 Januar 2010, Charles Solar wrote:
> Well I want to define the overloads myself anyway, I just do not know how
> to properly setup the small wrapper that will work.

Oh, OK.

> In the doc it tells 
> you how to make flat function wrappers, but nothing on member function
> wrappers.

I am 100% sure that they're documented, yet I am not very lucky with the docs 
either.

> I am unsure how I am supposed to handle the this pointer, for 
> example.

You can reference the member function pointer like this (untested of course):

.def("bar", (void(Foo::*)(const Y*))&T::bar)
.def("bar", (void(Foo::*)(const X&,const Y*))&T::bar)

> So idealy what I would like to know is what the macro does so I can
> manually write out the wrappers myself.
>
> And I have tried using Py++, it does not handle the default argument, it
> just puts the call in there and python still requires the param to be
> there.

Really?  AFAIK, it should work.

You should be able to write (arg("some_x"), arg("some_y") = NULL)) as 
extra .def() argument.

Maybe py++ does not output the default args when the arguments are unnamed?
(I wouldn't know how, either.)

-- 
Ciao, /  /                                                    .o.
     /--/                                                     ..o
    /  / ANS                                                  ooo
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part.
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100112/ad3aa351/attachment.pgp>

From charlessolar at gmail.com  Tue Jan 12 21:27:41 2010
From: charlessolar at gmail.com (Charles Solar)
Date: Tue, 12 Jan 2010 14:27:41 -0600
Subject: [C++-sig] Member function bp::optional? Or workaround?
In-Reply-To: <201001122109.40705.hans_meine@gmx.net>
References: <708326ee1001120708h62fbaad8hb43db3c5ec37b5cf@mail.gmail.com> 
	<201001122021.39062.hans_meine@gmx.net>
	<708326ee1001121146n11a748cdj8bbd32e30cfafa61@mail.gmail.com> 
	<201001122109.40705.hans_meine@gmx.net>
Message-ID: <708326ee1001121227g49a60babw28c42aa196a1bc53@mail.gmail.com>

Hmm, well if the arg definition matters like you two are suggesting it does,
then maybe the python code from Py++ does work.

Right now it generates code that I use, but I do not include the bp::object,
so my function in boost python looks like

int (Foo::*bar1)( const Y*) = &Foo::bar;
int (Foo::*bar2)(const X&, const Y*) = &Foo::bar;
...
.def( "bar", bar1 )
.def( "bar", bar2 )

That requires python to supply Y* but if I understand what you two are
saying

.def( "bar", bar1, ( bp::arg("Y") = bp::object() ) )
.def( "bar", bar2, ( bp::arg("X"), bp::arg(Y) = bp::object() ) )

would make python not require that default arg.  If that is the case then i
guess all is well.  When I saw Py++ generate that I figured bp::object() was
just some weird thing I did not have to worry about.

the documentation page I have been using and could not find an example of
member function default params is here
http://www.boost.org/doc/libs/1_41_0/libs/python/doc/tutorial/doc/html/python/functions.html#python.default_arguments
perhaps it would be nice to note this very handy bp::object somewhere.

Thanks

On Tue, Jan 12, 2010 at 2:09 PM, Hans Meine <hans_meine at gmx.net> wrote:

> On Dienstag 12 Januar 2010, Charles Solar wrote:
> > Well I want to define the overloads myself anyway, I just do not know how
> > to properly setup the small wrapper that will work.
>
> Oh, OK.
>
> > In the doc it tells
> > you how to make flat function wrappers, but nothing on member function
> > wrappers.
>
> I am 100% sure that they're documented, yet I am not very lucky with the
> docs
> either.
>
> > I am unsure how I am supposed to handle the this pointer, for
> > example.
>
> You can reference the member function pointer like this (untested of
> course):
>
> .def("bar", (void(Foo::*)(const Y*))&T::bar)
> .def("bar", (void(Foo::*)(const X&,const Y*))&T::bar)
>
> > So idealy what I would like to know is what the macro does so I can
> > manually write out the wrappers myself.
> >
> > And I have tried using Py++, it does not handle the default argument, it
> > just puts the call in there and python still requires the param to be
> > there.
>
> Really?  AFAIK, it should work.
>
> You should be able to write (arg("some_x"), arg("some_y") = NULL)) as
> extra .def() argument.
>
> Maybe py++ does not output the default args when the arguments are unnamed?
> (I wouldn't know how, either.)
>
> --
> Ciao, /  /                                                    .o.
>     /--/                                                     ..o
>    /  / ANS                                                  ooo
>
> _______________________________________________
> 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/20100112/78f441ca/attachment.htm>

From blp330 at msn.com  Wed Jan 13 07:57:31 2010
From: blp330 at msn.com (blp330)
Date: Tue, 12 Jan 2010 22:57:31 -0800 (PST)
Subject: [C++-sig] [boost.python] How can I wrap operator==?
In-Reply-To: <7465b6171001121155s25319ffbt724aac143530af2f@mail.gmail.com>
References: <27105703.post@talk.nabble.com>
	<7465b6171001121155s25319ffbt724aac143530af2f@mail.gmail.com>
Message-ID: <27140350.post@talk.nabble.com>


Um... Sorry, I mean I have a class Document, which defined

class Document
{
public:
 ....

 bool operator==(const Document& other) const
 {
  return Compare(other);
 }
};

It simply call Compare method, I need to something more when I use it in
python code but I don't want to change my Document class implementation
because it can be used by other language that do not need extra code.

Py++ generates:
Document_exposer.def( bp::self == bp::self );

For other Document method I can change Py++ generated code to do something
more, for example:
Py++:
   {   
            typedef ::StringPtr ( ::Document::*Thumb_function_type )(
int,int ) const;
            
            Document_exposer.def( 
                "Thumb"
                //, Thumb_function_type( &::Document::Thumb )
                , &Document__ThumbWrap
                , ( bp::arg("width"), bp::arg("height") ) );
        
        }
I can define a Document__ThumbWrap function to write my extra code in it.

But operator==, I don't know how to do that.
I can't simply change
Document_exposer.def( bp::self == bp::self );
to
Document_exposer.def( 
     bp::self == bp::self
     , &Document__EqualWrap
 );
to write my extra code in Document__EqualWrap.

Thank you very much.
-- 
View this message in context: http://old.nabble.com/-boost.python--How-can-I-wrap-operator%3D%3D--tp27105703p27140350.html
Sent from the Python - c++-sig mailing list archive at Nabble.com.


From roman.yakovenko at gmail.com  Wed Jan 13 08:24:33 2010
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Wed, 13 Jan 2010 09:24:33 +0200
Subject: [C++-sig] [boost.python] How can I wrap operator==?
In-Reply-To: <27140350.post@talk.nabble.com>
References: <27105703.post@talk.nabble.com>
	<7465b6171001121155s25319ffbt724aac143530af2f@mail.gmail.com>
	<27140350.post@talk.nabble.com>
Message-ID: <7465b6171001122324q1a96d81cm54d43d3c613b5cbf@mail.gmail.com>

On Wed, Jan 13, 2010 at 8:57 AM, blp330 <blp330 at msn.com> wrote:
>
> Um... Sorry, I mean I have a class Document, which defined
>
> class Document
> {
> public:
> ?....
>
> ?bool operator==(const Document& other) const
> ?{
> ?return Compare(other);
> ?}
> };
>
> It simply call Compare method, I need to something more when I use it in
> python code but I don't want to change my Document class implementation
> because it can be used by other language that do not need extra code.
>
> Py++ generates:
> Document_exposer.def( bp::self == bp::self );
>
> For other Document method I can change Py++ generated code to do something
> more, for example:
> Py++:
> ? {
> ? ? ? ? ? ?typedef ::StringPtr ( ::Document::*Thumb_function_type )(
> int,int ) const;
>
> ? ? ? ? ? ?Document_exposer.def(
> ? ? ? ? ? ? ? ?"Thumb"
> ? ? ? ? ? ? ? ?//, Thumb_function_type( &::Document::Thumb )
> ? ? ? ? ? ? ? ?, &Document__ThumbWrap
> ? ? ? ? ? ? ? ?, ( bp::arg("width"), bp::arg("height") ) );
>
> ? ? ? ?}
> I can define a Document__ThumbWrap function to write my extra code in it.
>
> But operator==, I don't know how to do that.
> I can't simply change
> Document_exposer.def( bp::self == bp::self );
> to
> Document_exposer.def(
> ? ? bp::self == bp::self
> ? ? , &Document__EqualWrap
> ?);
> to write my extra code in Document__EqualWrap.

If I understand you right, you need:

1. define new global function, which takes Document& as the first argument

bool Document_EqualWrap( const Document& x, const Document& y ){...}

2. Expose it:

...
.def( "__eq__", &Document_EqualWrap )

Is not the working code, but it should give you a direction.

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

From sipickles at googlemail.com  Wed Jan 13 10:00:54 2010
From: sipickles at googlemail.com (Simon Pickles)
Date: Wed, 13 Jan 2010 09:00:54 +0000
Subject: [C++-sig] Conversion problem
Message-ID: <4B4D8BC6.6050401@googlemail.com>

Hello,

I wonder if someone could help me with a tricky conversion between c++ 
and python types (well, tricky for me!)

I have a c++ library that I would like to use in python via Boost::Python

One function fills a c++ unsigned char* buffer. It has two overloads:

        unsigned char * getPixels(int deviceID);
        bool getPixels(int id, unsigned char * pixels);

How would I go about getting that data into a python object, a string in 
particular? I seem to face several problems:
/////////////////////////
1) unsigned char* does not have a to_python converter to Py_String.

I tried this:

struct uchar_ptr_to_str
{
    static PyObject* convert(unsigned char* p)
    {
         return incref( str(p).ptr());
    }
    static PyTypeObject const *get_pytype () {return &PyString_Type; }
};
struct uchar_to_str
{
    static PyObject* convert(unsigned char& p)
    {
         return incref( str(p).ptr());
    }
    static PyTypeObject const *get_pytype () {return &PyString_Type; }
};

BOOST_PYTHON_MODULE(pyVideoInput)
{
    to_python_converter< unsigned char*, uchar_ptr_to_str, true >();   
    to_python_converter< unsigned char, uchar_to_str, true >();

    class_<videoInput>("videoInput")
       .def("getPixels", &videoInput::getPixels, 
return_value_policy<manage_new_object>())
       ;
}


/////////////////////////
2) How do I allocate memory in a python string and then pass the pointer 
through boost python, to be filled by the c++ function?

In c++, the library would be used like this:

    videoInput vi;
    ....
    unsigned char * buffer = new unsigned char[size];
    vi.getPixels(device, buffer);

The buffer could be pretty big so I want to pass by pointer, not by value

The alternative would be to use the overload which returns the pointer, 
but I can't get that to compile with bjam. I get the dreaded:

make_instance.hpp(24) : error C2027: 
use of undefined type 'boost::STATIC_ASSERTION_FAILURE<x>

when I try to return a pointer...

Thanks for any help freeing me from my confusion!

Regards

Si






From nico_ml at mgdesign.org  Wed Jan 13 10:10:29 2010
From: nico_ml at mgdesign.org (Nicolas Lelong)
Date: Wed, 13 Jan 2010 10:10:29 +0100
Subject: [C++-sig] Returning a simple char *
In-Reply-To: <4B4C2FF6.7080307@googlemail.com>
References: <4B4C2FF6.7080307@googlemail.com>
Message-ID: <4B4D8E05.3020204@mgdesign.org>

An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100113/a7734004/attachment.htm>

From roman.yakovenko at gmail.com  Wed Jan 13 11:44:50 2010
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Wed, 13 Jan 2010 12:44:50 +0200
Subject: [C++-sig] Conversion problem
In-Reply-To: <4B4D8BC6.6050401@googlemail.com>
References: <4B4D8BC6.6050401@googlemail.com>
Message-ID: <7465b6171001130244td40a022k103eedb703b2f9f0@mail.gmail.com>

On Wed, Jan 13, 2010 at 11:00 AM, Simon Pickles
<sipickles at googlemail.com> wrote:
> Hello,
>
> I wonder if someone could help me with a tricky conversion between c++ and
> python types (well, tricky for me!)
>
> I have a c++ library that I would like to use in python via Boost::Python
>
> One function fills a c++ unsigned char* buffer. It has two overloads:
>
> ? ? ? unsigned char * getPixels(int deviceID);
> ? ? ? bool getPixels(int id, unsigned char * pixels);
>

In both cases the easiest solution would be create small wrappers
around the functions:

std::string getPixels(int i ){ ... }

and expose it.

Another solution would be to use ctypes module to call those functions.

In both cases Py++ can help you:
http://language-binding.net/pyplusplus/documentation/functions/transformation/transformation.html

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

From nico_ml at mgdesign.org  Wed Jan 13 12:17:23 2010
From: nico_ml at mgdesign.org (Nicolas Lelong)
Date: Wed, 13 Jan 2010 12:17:23 +0100
Subject: [C++-sig] Conversion problem
In-Reply-To: <4B4D8BC6.6050401@googlemail.com>
References: <4B4D8BC6.6050401@googlemail.com>
Message-ID: <4B4DABC3.3020002@mgdesign.org>

Simon,

it seems that I wrapped the same videoinput library you're using.

I wrapped the getPixels the following way : python script is responsible 
of the memory allocation for pixels buffer (in a correctly sized string).

The getPixels function is wrapped as follows :

namespace {

   bool videoInput_getPixels(videoInput& input, int device_id, 
python::object memory_buffer)
   {
     PyObject* pyObject = memory_buffer.ptr();
     if (PyString_CheckExact(pyObject))
     {
       Py_ssize_t string_size = PyString_Size(pyObject);
       if (string_size >= Py_ssize_t(input.getWidth(device_id)) * 
Py_ssize_t(input.getHeight(device_id)) * Py_ssize_t(3))
       {
         unsigned char* pixels = reinterpret_cast<unsigned 
char*>(PyString_AsString(pyObject));

         return input.getPixels(device_id, pixels, false, false);
       }
     }
     return false;
   }

};

python::class_<videoInput, boost::noncopyable> klass("VideoInput");
klass.def("getPixels", videoInput_getPixels);

This certainly lacks some error checking for the actual 'memory_buffer' 
parameter type, but can give you a clue.

HTH,

Nicolas


From tim at tizmoi.net  Wed Jan 13 19:24:19 2010
From: tim at tizmoi.net (Tim Couper)
Date: Wed, 13 Jan 2010 18:24:19 +0000
Subject: [C++-sig] Simple , ,
	I think .. type conversion question C++ -> python
In-Reply-To: <4B4DABC3.3020002@mgdesign.org>
References: <4B4D8BC6.6050401@googlemail.com> <4B4DABC3.3020002@mgdesign.org>
Message-ID: <4B4E0FD3.9040106@tizmoi.net>

OK. Here's what I've been stuck with all today .. I have a 3rd party C++ 
program function which returns a boost::variant  (and its inverse)

my_variant my_variant_of_string(const std::string& str)

This one takes a string & returns a variant, and am trying to wrap this 
in python, so that there I can have

 >>> my_variant_of_string('hello')
'hello'

 >>> my_variant_of_string('1.2')
1.2

 >>> my_variant_of_string('10')
10

Simple, eh? .. The reduced code below compiles, creating the python 
function correctly, but I have not successfully defined 
to_python_value/to_python_converter statement(s), so the python 
predictably fails with a ".. no to_python .. converter for .. 
boost::variant .." when run.

I can find lots of helpful references to such converters of classes (eg 
http://misspent.wordpress.com/2009/09/27/how-to-write-boost-python-converters/),
but cannot see how I should use the to_python_value functionality to 
express the "convert the output of this function to a string, int or 
double as appropriate".

Thanks

Tim "novice, but learning" Couper


-- my_variant.cpp --

#include <sstream>

typedef boost::variant<int, double, std::string> my_variant;

template <class T>

     bool from_string(T& t,const std::string& s, std::ios_base& 
(*f)(std::ios_base&))
     {
         std::istringstream iss(s);
         return !(iss >> f >> t).fail();
      }

namespace var {

     my_variant my_variant_of_string(const std::string& str)
        // return a float, int or string depending on input parameter 
content
       {
         my_variant param;

         double dval; int ival;
         if (str.find(".") != std::string::npos) {
             if (from_string<double>(dval,str,std::dec)) {
                 param = dval;
             }
         }
         else if (from_string<int>(ival,str,std::dec) ) {
             param = ival;
         }
         else {
             param= str;
         }
         return param;
      }

}// var

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

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

     def("my_variant_of_string", &var::my_variant_of_string);

}

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100113/97739819/attachment.htm>

From troy at resophonic.com  Wed Jan 13 20:10:26 2010
From: troy at resophonic.com (troy d. straszheim)
Date: Wed, 13 Jan 2010 14:10:26 -0500
Subject: [C++-sig] Simple , ,
 I think .. type conversion question C++ -> python
In-Reply-To: <4B4E0FD3.9040106@tizmoi.net>
References: <4B4D8BC6.6050401@googlemail.com> <4B4DABC3.3020002@mgdesign.org>
	<4B4E0FD3.9040106@tizmoi.net>
Message-ID: <4B4E1AA2.8060805@resophonic.com>

Tim Couper wrote:
> OK. Here's what I've been stuck with all today .. I have a 3rd party C++ 
> program function which returns a boost::variant  (and its inverse)
> 
> my_variant my_variant_of_string(const std::string& str)
> 
> This one takes a string & returns a variant, and am trying to wrap this 
> in python, so that there I can have
> 
>  >>> my_variant_of_string('hello')
> 'hello'
> 
>  >>> my_variant_of_string('1.2')
> 1.2
> 
>  >>> my_variant_of_string('10')
> 10
> 
> Simple, eh? ...  

How about something like (not tested):

using boost::python::object;

typedef variant<...> variant_t;

variant_t makes_variant_from(string s); // defined elsewhere

//
// visitor for converting contents of visitor to object
//
struct vc : boost::static_visitor<object>
{
   template <typename T>
   object operator()(const T& v) const
   {
      return v;
   }
};

bp::object
thunk(string s)
{
   variant_t v = makes_variant_from(s);
   return boost::apply_visitor(vc(), v);
}

BOOST_PYTHON_MODULE(mod)
{
   def("f", &thunk);
}


You could probably use this to just register a converter from variant to 
vanilla boost::python::object as well, if this is more convenient...

-t




From tim at tizmoi.net  Wed Jan 13 20:24:50 2010
From: tim at tizmoi.net (Tim Couper)
Date: Wed, 13 Jan 2010 19:24:50 +0000
Subject: [C++-sig] Simple , ,
 I think .. type conversion question C++ -> python
In-Reply-To: <4B4E1AA2.8060805@resophonic.com>
References: <4B4D8BC6.6050401@googlemail.com>
	<4B4DABC3.3020002@mgdesign.org>	<4B4E0FD3.9040106@tizmoi.net>
	<4B4E1AA2.8060805@resophonic.com>
Message-ID: <4B4E1E02.9030400@tizmoi.net>

Troy

Thanks .. your hints are most helpful .. I'll be back :-)

Tim

On 13/01/2010 19:10, troy d. straszheim wrote:
> Tim Couper wrote:
>> OK. Here's what I've been stuck with all today .. I have a 3rd party 
>> C++ program function which returns a boost::variant  (and its inverse)
>>
>> my_variant my_variant_of_string(const std::string& str)
>>
>> This one takes a string & returns a variant, and am trying to wrap 
>> this in python, so that there I can have
>>
>> >>> my_variant_of_string('hello')
>> 'hello'
>>
>> >>> my_variant_of_string('1.2')
>> 1.2
>>
>> >>> my_variant_of_string('10')
>> 10
>>
>> Simple, eh? ... 
>
> How about something like (not tested):
>
> using boost::python::object;
>
> typedef variant<...> variant_t;
>
> variant_t makes_variant_from(string s); // defined elsewhere
>
> //
> // visitor for converting contents of visitor to object
> //
> struct vc : boost::static_visitor<object>
> {
>   template <typename T>
>   object operator()(const T& v) const
>   {
>      return v;
>   }
> };
>
> bp::object
> thunk(string s)
> {
>   variant_t v = makes_variant_from(s);
>   return boost::apply_visitor(vc(), v);
> }
>
> BOOST_PYTHON_MODULE(mod)
> {
>   def("f", &thunk);
> }
>
>
> You could probably use this to just register a converter from variant 
> to vanilla boost::python::object as well, if this is more convenient...
>
> -t
>
>
>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>
>
>
> No virus found in this incoming message.
> Checked by AVG - www.avg.com
> Version: 9.0.725 / Virus Database: 270.14.138/2618 - Release Date: 01/13/10 07:35:00
>
>    
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100113/ae6296f0/attachment.htm>

From hans_meine at gmx.net  Wed Jan 13 13:54:24 2010
From: hans_meine at gmx.net (Hans Meine)
Date: Wed, 13 Jan 2010 13:54:24 +0100
Subject: [C++-sig] [boost.python] How can I wrap operator==?
In-Reply-To: <7465b6171001122324q1a96d81cm54d43d3c613b5cbf@mail.gmail.com>
References: <27105703.post@talk.nabble.com> <27140350.post@talk.nabble.com>
	<7465b6171001122324q1a96d81cm54d43d3c613b5cbf@mail.gmail.com>
Message-ID: <201001131354.25672.hans_meine@gmx.net>

On Mittwoch 13 Januar 2010, Roman Yakovenko wrote:
> On Wed, Jan 13, 2010 at 8:57 AM, blp330 <blp330 at msn.com> wrote:
> > class Document
> > {
> > public:
> >  ....
> >
> >  bool operator==(const Document& other) const
> >  {
> >  return Compare(other);
> >  }
> [...]
> If I understand you right, you need:
>
> 1. define new global function, which takes Document& as the first argument
>
> bool Document_EqualWrap( const Document& x, const Document& y ){...}

Not necessary AFAIK, a member function should work, too.
(i.e. directly exposing &Document::operator==)

> 2. Expose it:
>
> ...
> .def( "__eq__", &Document_EqualWrap )
>
> Is not the working code, but it should give you a direction.

Yes, I think that's what he was missing.

-- 
Ciao, /  /                                                    .o.
     /--/                                                     ..o
    /  / ANS                                                  ooo
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part.
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100113/9ac1e4c6/attachment-0001.pgp>

From roman.yakovenko at gmail.com  Wed Jan 13 21:39:53 2010
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Wed, 13 Jan 2010 22:39:53 +0200
Subject: [C++-sig] A nice Boost.Python and Py++ usage example
Message-ID: <7465b6171001131239r4fec3efcy6b0b3792baadf92f@mail.gmail.com>

http://frontiersin.org/psychology/psychology/paper/10.3389/neuro.11/011.2009/html/

The articles talks about some science fiction ( for me:-) ), but also
has few paragraphs related to Boost.Python and Py++.

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

From jjcarles at jocajica.com  Wed Jan 13 22:06:40 2010
From: jjcarles at jocajica.com (Joan Carles Jimenez)
Date: Wed, 13 Jan 2010 22:06:40 +0100
Subject: [C++-sig] Boost and wxWidgets
Message-ID: <1263416800.3462.71.camel@servidor.jocajica>

Hi.

  I am testing the following source code:



--------
Canvas.h
--------

#ifndef _CANVAS_H_
#define _CANVAS_H_

#include <string>
#include "wx/wx.h"

class CanvasFrame
{
    protected:
        wxFrame* _frame;
        
    public:
        CanvasFrame(const std::string& pTitle);  
        ~CanvasFrame(void);  
        
        void Plot(void);
};

#endif /* _CANVAS_H_ */

----------
Canvas.cpp
----------

#include <Python.h>
#include "canvas.h"

CanvasFrame::CanvasFrame(const std::string& pTitle)
{
    _frame = new wxFrame(NULL, wxID_ANY,
wxString::FromAscii(pTitle.c_str()));
}

CanvasFrame::~CanvasFrame(void)
{
    if(_frame) {
        delete _frame;
    }
}

void CanvasFrame::Plot(void)
{
}

-----------
wrapper.cpp
-----------

#include <boost/python.hpp>
#include "canvas.h"

using namespace boost::python;

BOOST_PYTHON_MODULE(canvas)
{
    class_<CanvasFrame>("CanvasFrame", init<const std::string&>())
        .def("Plot", &CanvasFrame::Plot)
    ;
}



And to generate the module "canvas.so" I execute:



g++ -shared -g -I. -I/usr/include/python2.6 -lpython2.6 canvas.cpp
wrapper.cpp /usr/lib64/libboost_python.so -fPIC `wx-config --cxxflags`
`wx-config --libs` -o canvas.so

"canvas.so" is generated without any problems but when I execute in the
python console the next code:



>>> import canvas
>>> a = canvas.CanvasFrame('My Plot Window')



I have the next errors:



(process:3521): GLib-GObject-CRITICAL **: gtype.c:2458: initialization
assertion failed, use IA__g_type_init() prior to this function

(process:3521): GLib-CRITICAL **: g_once_init_leave: assertion
`initialization_value != 0' failed

(process:3521): Gdk-CRITICAL **: gdk_cursor_new_for_display: assertion
`GDK_IS_DISPLAY (display)' failed

(process:3521): GLib-GObject-CRITICAL **: gtype.c:2458: initialization
assertion failed, use IA__g_type_init() prior to this function



And more, and more, and more. I have watched the examples of 

http://wiki.python.org/moin/boost.python/ExportingClasses

but I have not seen any problem. They can help me? Thanks!


P.D. Sorry for my english!



From sipickles at googlemail.com  Thu Jan 14 09:23:49 2010
From: sipickles at googlemail.com (Simon Pickles)
Date: Thu, 14 Jan 2010 08:23:49 +0000
Subject: [C++-sig] Conversion problem
In-Reply-To: <4B4DABC3.3020002@mgdesign.org>
References: <4B4D8BC6.6050401@googlemail.com> <4B4DABC3.3020002@mgdesign.org>
Message-ID: <4B4ED495.8030505@googlemail.com>

Thanks Nicolas,

I learnt a lot from your way of doing it!

Simon

On 13/01/2010 11:17, Nicolas Lelong wrote:
> Simon,
>
> it seems that I wrapped the same videoinput library you're using.
>
> I wrapped the getPixels the following way : python script is 
> responsible of the memory allocation for pixels buffer (in a correctly 
> sized string).
>
> The getPixels function is wrapped as follows :
>
> namespace {
>
>   bool videoInput_getPixels(videoInput& input, int device_id, 
> python::object memory_buffer)
>   {
>     PyObject* pyObject = memory_buffer.ptr();
>     if (PyString_CheckExact(pyObject))
>     {
>       Py_ssize_t string_size = PyString_Size(pyObject);
>       if (string_size >= Py_ssize_t(input.getWidth(device_id)) * 
> Py_ssize_t(input.getHeight(device_id)) * Py_ssize_t(3))
>       {
>         unsigned char* pixels = reinterpret_cast<unsigned 
> char*>(PyString_AsString(pyObject));
>
>         return input.getPixels(device_id, pixels, false, false);
>       }
>     }
>     return false;
>   }
>
> };
>
> python::class_<videoInput, boost::noncopyable> klass("VideoInput");
> klass.def("getPixels", videoInput_getPixels);
>
> This certainly lacks some error checking for the actual 
> 'memory_buffer' parameter type, but can give you a clue.
>
> HTH,
>
> Nicolas
>
> _______________________________________________
> 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/20100114/b2bee01a/attachment.htm>

From brokenn at gmail.com  Thu Jan 14 09:53:06 2010
From: brokenn at gmail.com (Brian O'Kennedy)
Date: Thu, 14 Jan 2010 08:53:06 +0000
Subject: [C++-sig] Boost and wxWidgets
In-Reply-To: <1263416800.3462.71.camel@servidor.jocajica>
References: <1263416800.3462.71.camel@servidor.jocajica>
Message-ID: <cd543dea1001140053u7fd06335lb85f5e3e2beb1eba@mail.gmail.com>

I suspect your problem is unrelated to boost.python. I've never used
wxWidgets, but a quick google suggests it should be initialised before being
used.

*http://docs.wxwidgets.org/stable/wx_appinifunctions.html*
<http://docs.wxwidgets.org/stable/wx_appinifunctions.html>  int wxEntry(int&
* argc*, wxChar ***argv*)

hope this helps,
 Brian


2010/1/13 Joan Carles Jimenez <jjcarles at jocajica.com>

> Hi.
>
>  I am testing the following source code:
>
>
>
> --------
> Canvas.h
> --------
>
> #ifndef _CANVAS_H_
> #define _CANVAS_H_
>
> #include <string>
> #include "wx/wx.h"
>
> class CanvasFrame
> {
>    protected:
>        wxFrame* _frame;
>
>    public:
>        CanvasFrame(const std::string& pTitle);
>        ~CanvasFrame(void);
>
>        void Plot(void);
> };
>
> #endif /* _CANVAS_H_ */
>
> ----------
> Canvas.cpp
> ----------
>
> #include <Python.h>
> #include "canvas.h"
>
> CanvasFrame::CanvasFrame(const std::string& pTitle)
> {
>    _frame = new wxFrame(NULL, wxID_ANY,
> wxString::FromAscii(pTitle.c_str()));
> }
>
> CanvasFrame::~CanvasFrame(void)
> {
>    if(_frame) {
>        delete _frame;
>    }
> }
>
> void CanvasFrame::Plot(void)
> {
> }
>
> -----------
> wrapper.cpp
> -----------
>
> #include <boost/python.hpp>
> #include "canvas.h"
>
> using namespace boost::python;
>
> BOOST_PYTHON_MODULE(canvas)
> {
>    class_<CanvasFrame>("CanvasFrame", init<const std::string&>())
>        .def("Plot", &CanvasFrame::Plot)
>    ;
> }
>
>
>
> And to generate the module "canvas.so" I execute:
>
>
>
> g++ -shared -g -I. -I/usr/include/python2.6 -lpython2.6 canvas.cpp
> wrapper.cpp /usr/lib64/libboost_python.so -fPIC `wx-config --cxxflags`
> `wx-config --libs` -o canvas.so
>
> "canvas.so" is generated without any problems but when I execute in the
> python console the next code:
>
>
>
> >>> import canvas
> >>> a = canvas.CanvasFrame('My Plot Window')
>
>
>
> I have the next errors:
>
>
>
> (process:3521): GLib-GObject-CRITICAL **: gtype.c:2458: initialization
> assertion failed, use IA__g_type_init() prior to this function
>
> (process:3521): GLib-CRITICAL **: g_once_init_leave: assertion
> `initialization_value != 0' failed
>
> (process:3521): Gdk-CRITICAL **: gdk_cursor_new_for_display: assertion
> `GDK_IS_DISPLAY (display)' failed
>
> (process:3521): GLib-GObject-CRITICAL **: gtype.c:2458: initialization
> assertion failed, use IA__g_type_init() prior to this function
>
>
>
> And more, and more, and more. I have watched the examples of
>
> http://wiki.python.org/moin/boost.python/ExportingClasses
>
> but I have not seen any problem. They can help me? Thanks!
>
>
> P.D. Sorry for my english!
>
>
> _______________________________________________
> 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/20100114/98616ba5/attachment.htm>

From jjcarles at jocajica.com  Thu Jan 14 11:41:31 2010
From: jjcarles at jocajica.com (Joan Carles Jimenez)
Date: Thu, 14 Jan 2010 11:41:31 +0100
Subject: [C++-sig] Boost and wxWidgets
In-Reply-To: <cd543dea1001140053u7fd06335lb85f5e3e2beb1eba@mail.gmail.com>
References: <1263416800.3462.71.camel@servidor.jocajica>
	<cd543dea1001140053u7fd06335lb85f5e3e2beb1eba@mail.gmail.com>
Message-ID: <1263465691.3462.87.camel@servidor.jocajica>

Thanks Brian, but do not work. The "canvas.so" is a static library not
an application. My info is:

openSuSE 11.2 64bits
boost 1.39
gcc-c++ 4.4
python 2.6.2




On Thu, 2010-01-14 at 08:53 +0000, Brian O'Kennedy wrote:
> I suspect your problem is unrelated to boost.python. I've never used
> wxWidgets, but a quick google suggests it should be initialised before
> being used. 
> 
> 
> http://docs.wxwidgets.org/stable/wx_appinifunctions.html
>   int wxEntry(int& argc, wxChar **argv)
> 
> 
> hope this helps, 
>  Brian
> 
> 
> 
> 2010/1/13 Joan Carles Jimenez <jjcarles at jocajica.com>
>         Hi.
>         
>          I am testing the following source code:
>         
>         
>         
>         --------
>         Canvas.h
>         --------
>         
>         #ifndef _CANVAS_H_
>         #define _CANVAS_H_
>         
>         #include <string>
>         #include "wx/wx.h"
>         
>         class CanvasFrame
>         {
>            protected:
>                wxFrame* _frame;
>         
>            public:
>                CanvasFrame(const std::string& pTitle);
>                ~CanvasFrame(void);
>         
>                void Plot(void);
>         };
>         
>         #endif /* _CANVAS_H_ */
>         
>         ----------
>         Canvas.cpp
>         ----------
>         
>         #include <Python.h>
>         #include "canvas.h"
>         
>         CanvasFrame::CanvasFrame(const std::string& pTitle)
>         {
>            _frame = new wxFrame(NULL, wxID_ANY,
>         wxString::FromAscii(pTitle.c_str()));
>         }
>         
>         CanvasFrame::~CanvasFrame(void)
>         {
>            if(_frame) {
>                delete _frame;
>            }
>         }
>         
>         void CanvasFrame::Plot(void)
>         {
>         }
>         
>         -----------
>         wrapper.cpp
>         -----------
>         
>         #include <boost/python.hpp>
>         #include "canvas.h"
>         
>         using namespace boost::python;
>         
>         BOOST_PYTHON_MODULE(canvas)
>         {
>            class_<CanvasFrame>("CanvasFrame", init<const
>         std::string&>())
>                .def("Plot", &CanvasFrame::Plot)
>            ;
>         }
>         
>         
>         
>         And to generate the module "canvas.so" I execute:
>         
>         
>         
>         g++ -shared -g -I. -I/usr/include/python2.6 -lpython2.6
>         canvas.cpp
>         wrapper.cpp /usr/lib64/libboost_python.so -fPIC `wx-config
>         --cxxflags`
>         `wx-config --libs` -o canvas.so
>         
>         "canvas.so" is generated without any problems but when I
>         execute in the
>         python console the next code:
>         
>         
>         
>         >>> import canvas
>         >>> a = canvas.CanvasFrame('My Plot Window')
>         
>         
>         
>         I have the next errors:
>         
>         
>         
>         (process:3521): GLib-GObject-CRITICAL **: gtype.c:2458:
>         initialization
>         assertion failed, use IA__g_type_init() prior to this function
>         
>         (process:3521): GLib-CRITICAL **: g_once_init_leave: assertion
>         `initialization_value != 0' failed
>         
>         (process:3521): Gdk-CRITICAL **: gdk_cursor_new_for_display:
>         assertion
>         `GDK_IS_DISPLAY (display)' failed
>         
>         (process:3521): GLib-GObject-CRITICAL **: gtype.c:2458:
>         initialization
>         assertion failed, use IA__g_type_init() prior to this function
>         
>         
>         
>         And more, and more, and more. I have watched the examples of
>         
>         http://wiki.python.org/moin/boost.python/ExportingClasses
>         
>         but I have not seen any problem. They can help me? Thanks!
>         
>         
>         P.D. Sorry for my english!
>         
>         
>         _______________________________________________
>         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 hans_meine at gmx.net  Thu Jan 14 11:54:08 2010
From: hans_meine at gmx.net (Hans Meine)
Date: Thu, 14 Jan 2010 11:54:08 +0100
Subject: [C++-sig] Boost and wxWidgets
In-Reply-To: <1263465691.3462.87.camel@servidor.jocajica>
References: <1263416800.3462.71.camel@servidor.jocajica>
	<cd543dea1001140053u7fd06335lb85f5e3e2beb1eba@mail.gmail.com>
	<1263465691.3462.87.camel@servidor.jocajica>
Message-ID: <201001141154.08855.hans_meine@gmx.net>

Am Donnerstag, 14. Januar 2010 11:41:31 schrieb Joan Carles Jimenez:
> On Thu, 2010-01-14 at 08:53 +0000, Brian O'Kennedy wrote:
> > I suspect your problem is unrelated to boost.python. I've never used
> > wxWidgets, but a quick google suggests it should be initialised before
> > being used.
> [...]
> Thanks Brian, but do not work. The "canvas.so" is a static library not
> an application. 

Yes, but then you should initialize wx/gtk in the actual program using 
canvas.so, before calling the relevant canvas.so functions.

HTH,
  Hans

From rosen.diankov at gmail.com  Thu Jan 14 16:55:37 2010
From: rosen.diankov at gmail.com (Rosen Diankov)
Date: Thu, 14 Jan 2010 10:55:37 -0500
Subject: [C++-sig] Robotics Project Using Boost Python: OpenRAVE
Message-ID: <103101d71001140755k64f1b256l8b3acd5d52e9a018@mail.gmail.com>

Dear Boost Python Developers,

We've been using boost.python for about a year now for a robotics
planning and simulation environment called OpenRAVE (developed by
Rosen Diankov):
http://openrave.programmingvision.com

We are very pleased with boost python and it has opened up many new
doors for development. I saw that Boost.Python has a projects page, so
if it openrave is deemed worth of getting mentioned on it, a possible
project description would be:

"OpenRAVE is an open-source, cross-platform, plugin-based robot
planning environment for autonomous robotics. Includes many services
like collision detection, physics, (inverse) kinematics, sensors,
robot controls, complex motion planners, grasping simulation,
manipulation planning, and a powerful python scripting environment
based on boost.python. Boost.Python allows OpenRAVE to export all its
complex C++ functionality into a scripting environment where
researchers can easily test new algorithms and dynamically control the
flow of execution."

thank you,
rosen,

From nico_ml at mgdesign.org  Fri Jan 15 18:09:36 2010
From: nico_ml at mgdesign.org (Nicolas Lelong)
Date: Fri, 15 Jan 2010 18:09:36 +0100
Subject: [C++-sig] boost python calling conventions support
 patch	submission (__stdcall, __cdecl, __fastcall)
In-Reply-To: <201001101118.25701.lists_ravi@lavabit.com>
References: <4B437643.5000103@mgdesign.org>
	<201001101118.25701.lists_ravi@lavabit.com>
Message-ID: <4B50A150.1010908@mgdesign.org>

My previous mail once again got lost, gotta be more careful...

> Changes (by dave):
>
>   This looks terrific!  I'll try to get it applied ASAP.  Please poke me if
>   you don't see it in trunk by the end of the week.

Thanks ! As I didn't see the batch in trunk today, I poke you :)

 > Please file a trac ticket at svn.boost.org so that this does not get 
lost.

OK Ravi, thanks for the pointer, I should have known....

I submitted ticket #3833 (https://svn.boost.org/trac/boost/ticket/3833), 
and will be happy to have feedback about it !

Cheers,

Nicolas.


From bfitzpatrick at vtiinstruments.com  Mon Jan 18 22:48:01 2010
From: bfitzpatrick at vtiinstruments.com (Ben Fitzpatrick)
Date: Mon, 18 Jan 2010 15:48:01 -0600
Subject: [C++-sig] Pybindgen and enum types
Message-ID: <27AD07BD25612C40AF5C78CB49D09B4967CC0F69EE@34093-MBX-C08.mex07a.mlsrvr.com>

Hi everyone,

I am trying to wrap a section of C++ code with many enum types in it, of the form:

enum Blah
{
  BlahSomething1 = 1000;
  BlahSomething2 = 1001;
}

Thanks to this definition - pulled into pybindgen via automagical pygccxml parsing of some header files - code of the form
Blah here_is_a_variable;

Is recognized just fine and works great. However, code of the form
Blah* here_is_a_variable;

Is not recognized because pybindgen/pygccxml can't find an explicit definition of a Blah*.  I am wondering if there's a simple way to convert these programmatically - that is, not have to write a to_python converter for each one, especially since they're all of the generic form "(enum)*". The code I am wrapping is not guaranteed API-stable, and I'm trying to automate as much parsing as possible. Any suggestions?

Thanks,
Ben
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100118/cc3b680a/attachment.htm>

From gjcarneiro at gmail.com  Tue Jan 19 12:04:37 2010
From: gjcarneiro at gmail.com (Gustavo Carneiro)
Date: Tue, 19 Jan 2010 11:04:37 +0000
Subject: [C++-sig] Pybindgen and enum types
In-Reply-To: <27AD07BD25612C40AF5C78CB49D09B4967CC0F69EE@34093-MBX-C08.mex07a.mlsrvr.com>
References: <27AD07BD25612C40AF5C78CB49D09B4967CC0F69EE@34093-MBX-C08.mex07a.mlsrvr.com>
Message-ID: <a467ca4f1001190304w6122e980q5a828d009afb327f@mail.gmail.com>

On Mon, Jan 18, 2010 at 9:48 PM, Ben Fitzpatrick <
bfitzpatrick at vtiinstruments.com> wrote:

>  Hi everyone,
>
>
>
> I am trying to wrap a section of C++ code with many enum types in it, of
> the form:
>
>
>
> enum Blah
>
> {
>
>   BlahSomething1 = 1000;
>
>   BlahSomething2 = 1001;
>
> }
>
>
>
> Thanks to this definition - pulled into pybindgen via automagical pygccxml
> parsing of some header files - code of the form
>
> Blah here_is_a_variable;
>
>
>
> Is recognized just fine and works great. However, code of the form
>
> Blah* here_is_a_variable;
>
>
>
> Is not recognized because pybindgen/pygccxml can?t find an explicit
> definition of a Blah*.  I am wondering if there?s a simple way to convert
> these programmatically ? that is, not have to write a to_python converter
> for each one, especially since they?re all of the generic form ?(enum)*?.
> The code I am wrapping is not guaranteed API-stable, and I?m trying to
> automate as much parsing as possible. Any suggestions?
>

If I understand correctly, you are just missing the enum* type handler,
which should have been implemented in pybindgen, but I am lazy and often
only implement things as they become needed.

You can download a patch from here:

http://bazaar.launchpad.net/~gjc/pybindgen/trunk/revision/741


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



-- 
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/20100119/7e4d6451/attachment.htm>

From nico_ml at mgdesign.fr  Mon Jan 11 10:06:35 2010
From: nico_ml at mgdesign.fr (Nicolas Lelong)
Date: Mon, 11 Jan 2010 10:06:35 +0100
Subject: [C++-sig] boost python calling conventions support
 patch	submission (__stdcall, __cdecl, __fastcall)
In-Reply-To: <201001101118.25701.lists_ravi@lavabit.com>
References: <4B437643.5000103@mgdesign.org>
	<201001101118.25701.lists_ravi@lavabit.com>
Message-ID: <4B4AEA1B.2080907@mgdesign.fr>

 > Please file a trac ticket at svn.boost.org so that this does not get 
lost.

OK Ravi, thanks for the pointer, I should have known....

I submitted ticket #3833 (https://svn.boost.org/trac/boost/ticket/3833), 
and will be happy to have feedback about it !

Cheers,

Nicolas.

From mtn at xcm.ch  Mon Jan 11 17:15:16 2010
From: mtn at xcm.ch (mtn)
Date: Mon, 11 Jan 2010 08:15:16 -0800 (PST)
Subject: [C++-sig] Py++/boost::python: How to correctly wrap libraries
 (DLLs) with dependencies on each other.
Message-ID: <27112969.post@talk.nabble.com>


Hi all,

I try to embed a Python interpreter into my project using
Py++/boost::python.

I have a problem wrapping different libraries for which i own the source and
which I compile to DLL/*.so files . The libraries have dependencies on each
other but do not necessarily have to be used together.


An abstract example:

The interpreter is in a library on which all other libraries depend and
where they register themselves using init functions generated by
py++/boost::python.
Let's say I have a base library B.
Additionally a library E extending library B.

lib B:
class __exported_to_python__ Base {
  void __exported_to_pytthon__  foo(){...}
}

lib E -> depends B:
class __exported_to_python__ Extended : public B
{
  void __exported_to_python__  boo(){...}
}

It makes sense to also use B alone, therefore I would like to build a Python
wrapper for each library directly into the DLL.

Now I register my init function generate by Py++/boost::python to the
interpreter instance using `PyImport_AppendInittab`.

The thing, that I cannot generate a proper inheritance relation between the
class `Extended` and `Base` without telling Py++ to wrap both classes. But
then I have wrapped `Base` twice. Once in B and once in E.
And when boost::python tries to register some transfer functions based on
the C++ type (RTTI) it fails, as it has already been registered (because
init_B was called).

If I simply delete the duplicated classes by hand from the wrapper of
library `E` it works.
But one has to ensure that first `B` is imported to Python (`import B`) and
afterwards `E` (`import E`) otherwise boost::python does not have `Base`
registered on which `Extended` depends.


I hope I made myself clear enough now :-)
So has anybody solved this problem before me?
I did not find a source which would explain how to solve this issue with
Py++ and boost::python.

Thanks for all the fish!

solong

Martin










-- 
View this message in context: http://old.nabble.com/Py%2B%2B-boost%3A%3Apython%3A-How-to-correctly-wrap-libraries-%28DLLs%29-with-dependencies-on-each-other.-tp27112969p27112969.html
Sent from the Python - c++-sig mailing list archive at Nabble.com.


From nico_ml at mgdesign.fr  Tue Jan 12 15:06:24 2010
From: nico_ml at mgdesign.fr (Nicolas Lelong)
Date: Tue, 12 Jan 2010 15:06:24 +0100
Subject: [C++-sig] Returning a simple char *
In-Reply-To: <4B4C2FF6.7080307@googlemail.com>
References: <4B4C2FF6.7080307@googlemail.com>
Message-ID: <4B4C81E0.8070306@mgdesign.fr>

IMO, the easiest route would be to wrap the member functions :

namespace {

std::string video_getDeviceName_wrapped(int deviceID)
{
   char* name = videoInput::getDeviceName(devideID);
   std::string result(name);
   // free 'name' memory if needed
   return result;
}

}

and then ::

using namespace boost::python;
BOOST_PYTHON_MODULE(pyVideoInput)
{
    class_<videoInput>("videoInput")
        .def("getDeviceName", &video_getDeviceName_wrapped)
        .staticmethod("getDeviceName")
        ;
}

and you should be done.

From roman.yakovenko at gmail.com  Wed Jan 20 09:22:30 2010
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Wed, 20 Jan 2010 10:22:30 +0200
Subject: [C++-sig] Py++/boost::python: How to correctly wrap libraries
	(DLLs) with dependencies on each other.
In-Reply-To: <27112969.post@talk.nabble.com>
References: <27112969.post@talk.nabble.com>
Message-ID: <7465b6171001200022m11b2611ft70038cb026e0883c@mail.gmail.com>

On Mon, Jan 11, 2010 at 6:15 PM, mtn <mtn at xcm.ch> wrote:
>
> Hi all,
>
> I try to embed a Python interpreter into my project using
> Py++/boost::python.
>
> I have a problem wrapping different libraries for which i own the source and
> which I compile to DLL/*.so files . The libraries have dependencies on each
> other but do not necessarily have to be used together.
>
>
> An abstract example:
>
> The interpreter is in a library on which all other libraries depend and
> where they register themselves using init functions generated by
> py++/boost::python.
> Let's say I have a base library B.
> Additionally a library E extending library B.
>
> lib B:
> class __exported_to_python__ Base {
> ?void __exported_to_pytthon__ ?foo(){...}
> }
>
> lib E -> depends B:
> class __exported_to_python__ Extended : public B
> {
> ?void __exported_to_python__ ?boo(){...}
> }
>
> It makes sense to also use B alone, therefore I would like to build a Python
> wrapper for each library directly into the DLL.
>
> Now I register my init function generate by Py++/boost::python to the
> interpreter instance using `PyImport_AppendInittab`.
>
> The thing, that I cannot generate a proper inheritance relation between the
> class `Extended` and `Base` without telling Py++ to wrap both classes. But
> then I have wrapped `Base` twice. Once in B and once in E.
> And when boost::python tries to register some transfer functions based on
> the C++ type (RTTI) it fails, as it has already been registered (because
> init_B was called).
>
> If I simply delete the duplicated classes by hand from the wrapper of
> library `E` it works.
> But one has to ensure that first `B` is imported to Python (`import B`) and
> afterwards `E` (`import E`) otherwise boost::python does not have `Base`
> registered on which `Extended` depends.

You can import "B" from "E". You can do this by building "EImpl"
module and the "wrapping" it with package:
 E/
     __init__.py
     EImpl.[so|pyd|dll]

where __init__.py contains the following code:

import B
from EImpl import * #or something like that


>
> I hope I made myself clear enough now :-)
> So has anybody solved this problem before me?
> I did not find a source which would explain how to solve this issue with
> Py++ and boost::python.

Take a look on http://language-binding.net/pyplusplus/documentation/multi_module_development.html
. The document describes the situation you have.

If you need a good example take a look on Python-Ogre project. It uses
"multi module development" feature all the way.

HTH

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

From mtn at xcm.ch  Wed Jan 20 10:51:10 2010
From: mtn at xcm.ch (Martin Seiler)
Date: Wed, 20 Jan 2010 10:51:10 +0100
Subject: [C++-sig] Py++/boost::python: How to correctly wrap libraries
	(DLLs) with dependencies on each other.
In-Reply-To: <7465b6171001200022m11b2611ft70038cb026e0883c@mail.gmail.com>
References: <27112969.post@talk.nabble.com>
	<7465b6171001200022m11b2611ft70038cb026e0883c@mail.gmail.com>
Message-ID: <7957fb091001200151p25b588afg4429f40860b647f4@mail.gmail.com>

Thanks a lot Roman, everything answered..

On Wed, Jan 20, 2010 at 9:22 AM, Roman Yakovenko
<roman.yakovenko at gmail.com> wrote:
> On Mon, Jan 11, 2010 at 6:15 PM, mtn <mtn at xcm.ch> wrote:
>>
>> Hi all,
>>
>> I try to embed a Python interpreter into my project using
>> Py++/boost::python.
>>
>> I have a problem wrapping different libraries for which i own the source and
>> which I compile to DLL/*.so files . The libraries have dependencies on each
>> other but do not necessarily have to be used together.
>>
>>
>> An abstract example:
>>
>> The interpreter is in a library on which all other libraries depend and
>> where they register themselves using init functions generated by
>> py++/boost::python.
>> Let's say I have a base library B.
>> Additionally a library E extending library B.
>>
>> lib B:
>> class __exported_to_python__ Base {
>> ?void __exported_to_pytthon__ ?foo(){...}
>> }
>>
>> lib E -> depends B:
>> class __exported_to_python__ Extended : public B
>> {
>> ?void __exported_to_python__ ?boo(){...}
>> }
>>
>> It makes sense to also use B alone, therefore I would like to build a Python
>> wrapper for each library directly into the DLL.
>>
>> Now I register my init function generate by Py++/boost::python to the
>> interpreter instance using `PyImport_AppendInittab`.
>>
>> The thing, that I cannot generate a proper inheritance relation between the
>> class `Extended` and `Base` without telling Py++ to wrap both classes. But
>> then I have wrapped `Base` twice. Once in B and once in E.
>> And when boost::python tries to register some transfer functions based on
>> the C++ type (RTTI) it fails, as it has already been registered (because
>> init_B was called).
>>
>> If I simply delete the duplicated classes by hand from the wrapper of
>> library `E` it works.
>> But one has to ensure that first `B` is imported to Python (`import B`) and
>> afterwards `E` (`import E`) otherwise boost::python does not have `Base`
>> registered on which `Extended` depends.
>
> You can import "B" from "E". You can do this by building "EImpl"
> module and the "wrapping" it with package:
> ?E/
> ? ? __init__.py
> ? ? EImpl.[so|pyd|dll]
>
> where __init__.py contains the following code:
>
> import B
> from EImpl import * #or something like that
>
>
>>
>> I hope I made myself clear enough now :-)
>> So has anybody solved this problem before me?
>> I did not find a source which would explain how to solve this issue with
>> Py++ and boost::python.
>
> Take a look on http://language-binding.net/pyplusplus/documentation/multi_module_development.html
> . The document describes the situation you have.
>
> If you need a good example take a look on Python-Ogre project. It uses
> "multi module development" feature all the way.
>
> 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
>

From bfitzpatrick at vtiinstruments.com  Wed Jan 20 18:43:36 2010
From: bfitzpatrick at vtiinstruments.com (Ben Fitzpatrick)
Date: Wed, 20 Jan 2010 11:43:36 -0600
Subject: [C++-sig] Boost-python/pybindgen and wrapping member variables
Message-ID: <27AD07BD25612C40AF5C78CB49D09B4967CC0F6FDE@34093-MBX-C08.mex07a.mlsrvr.com>

Hi everyone,

I think I'm finally getting the hang of this, I seem to have gotten almost everything in my project working thanks to Gustavo's latest pybindgen patch. One of the remaining things I am having difficulty with is putting annotations on member variables of a class. I have something of the form:

class foo{
 <stuff>
};

class baz {
  public:
  foo* var;
}

The error I am getting from pybindgen is "WrapperWarning: Return value '::foo *' error (used in baz::var [variable]): TypeConfigurationError('Either caller_owns_return or self.reference_existing_object must be given')

 I know that the issue is that I need to stick an annotation on there. Since the pointer is allocated and tracked by the underlying C++ code, I'm pretty sure that I need to put reference_existing_object on it. I've tried inserting "{'reference_existing_object' : 'true'}" into the global_annotations table, and also inserting it into the parameter_annotations["return"] dictionary. For kicks, I even tried both at once. No matter what I do I can't get rid of the error, or see it show up in my final build, which leads me to believe there's something I'm missing. I went and looked up the documentation for reference_existing_object, but of course none of their examples show how you could use it on a member variable. What am I missing?

Thanks,
Ben Fitzpatrick
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100120/c6d85721/attachment.htm>

From gjcarneiro at gmail.com  Wed Jan 20 19:17:59 2010
From: gjcarneiro at gmail.com (Gustavo Carneiro)
Date: Wed, 20 Jan 2010 18:17:59 +0000
Subject: [C++-sig] Boost-python/pybindgen and wrapping member variables
In-Reply-To: <27AD07BD25612C40AF5C78CB49D09B4967CC0F6FDE@34093-MBX-C08.mex07a.mlsrvr.com>
References: <27AD07BD25612C40AF5C78CB49D09B4967CC0F6FDE@34093-MBX-C08.mex07a.mlsrvr.com>
Message-ID: <a467ca4f1001201017g7007f24cl9baa6ec88252333f@mail.gmail.com>

On Wed, Jan 20, 2010 at 5:43 PM, Ben Fitzpatrick <
bfitzpatrick at vtiinstruments.com> wrote:

>  Hi everyone,
>
>
>
> I think I?m finally getting the hang of this, I seem to have gotten almost
> everything in my project working thanks to Gustavo?s latest pybindgen patch.
> One of the remaining things I am having difficulty with is putting
> annotations on member variables of a class. I have something of the form:
>
>
>
> class foo{
>
>  <stuff>
>
> };
>
>
>
> class baz {
>
>   public:
>
>   foo* var;
>
> }
>
>
>
> The error I am getting from pybindgen is ?WrapperWarning: Return value
> ?::foo *? error (used in baz::var [variable]):
> TypeConfigurationError(?Either caller_owns_return or
> self.reference_existing_object must be given?)
>
>
>
>  I know that the issue is that I need to stick an annotation on there.
> Since the pointer is allocated and tracked by the underlying C++ code, I?m
> pretty sure that I need to put reference_existing_object on it. I?ve tried
> inserting ?{?reference_existing_object? : ?true?}? into the
> global_annotations table, and also inserting it into the
> parameter_annotations[?return?] dictionary. For kicks, I even tried both at
> once. No matter what I do I can?t get rid of the error, or see it show up in
> my final build, which leads me to believe there?s something I?m missing. I
> went and looked up the documentation for reference_existing_object, but of
> course none of their examples show how you could use it on a member
> variable. What am I missing?
>
You are not missing anything.  Annotations for member variables (pybindgen
calls them "instance attributes") were not implemented.

An oversight easily fixed:

http://bazaar.launchpad.net/~gjc/pybindgen/trunk/revision/742


>
>
> Thanks,
>
> Ben Fitzpatrick
>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>



-- 
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/20100120/546d1553/attachment.htm>

From bfitzpatrick at vtiinstruments.com  Wed Jan 20 20:26:03 2010
From: bfitzpatrick at vtiinstruments.com (Ben Fitzpatrick)
Date: Wed, 20 Jan 2010 13:26:03 -0600
Subject: [C++-sig] Boost-python/pybindgen and wrapping member variables
In-Reply-To: <a467ca4f1001201017g7007f24cl9baa6ec88252333f@mail.gmail.com>
References: <27AD07BD25612C40AF5C78CB49D09B4967CC0F6FDE@34093-MBX-C08.mex07a.mlsrvr.com>
	<a467ca4f1001201017g7007f24cl9baa6ec88252333f@mail.gmail.com>
Message-ID: <27AD07BD25612C40AF5C78CB49D09B4967CC0F7063@34093-MBX-C08.mex07a.mlsrvr.com>

Gustavo,

I think you win some sort of prize for not only fastest mailing list responses, but also fastest implementation of features. I reviewed the diff, and changed what I was adding to the global annotations table for these pointers to ?{?return_internal_reference?:?true?}? ? this seems to be what the new code in foo.h was doing.  I am still getting the errors, though ? is this the right way of using it?

And just to make sure, I grabbed a new checkout of pybindgen and re-installed it.

From: cplusplus-sig-bounces+bfitzpatrick=vtiinstruments.com at python.org [mailto:cplusplus-sig-bounces+bfitzpatrick=vtiinstruments.com at python.org] On Behalf Of Gustavo Carneiro
Sent: Wednesday, January 20, 2010 1:18 PM
To: Development of Python/C++ integration
Subject: Re: [C++-sig] Boost-python/pybindgen and wrapping member variables


On Wed, Jan 20, 2010 at 5:43 PM, Ben Fitzpatrick <bfitzpatrick at vtiinstruments.com<mailto:bfitzpatrick at vtiinstruments.com>> wrote:
Hi everyone,

I think I?m finally getting the hang of this, I seem to have gotten almost everything in my project working thanks to Gustavo?s latest pybindgen patch. One of the remaining things I am having difficulty with is putting annotations on member variables of a class. I have something of the form:

class foo{
 <stuff>
};

class baz {
  public:
  foo* var;
}

The error I am getting from pybindgen is ?WrapperWarning: Return value ?::foo *? error (used in baz::var [variable]): TypeConfigurationError(?Either caller_owns_return or self.reference_existing_object must be given?)

 I know that the issue is that I need to stick an annotation on there. Since the pointer is allocated and tracked by the underlying C++ code, I?m pretty sure that I need to put reference_existing_object on it. I?ve tried inserting ?{?reference_existing_object? : ?true?}? into the global_annotations table, and also inserting it into the parameter_annotations[?return?] dictionary. For kicks, I even tried both at once. No matter what I do I can?t get rid of the error, or see it show up in my final build, which leads me to believe there?s something I?m missing. I went and looked up the documentation for reference_existing_object, but of course none of their examples show how you could use it on a member variable. What am I missing?
You are not missing anything.  Annotations for member variables (pybindgen calls them "instance attributes") were not implemented.

An oversight easily fixed:

http://bazaar.launchpad.net/~gjc/pybindgen/trunk/revision/742


Thanks,
Ben Fitzpatrick

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



--
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/20100120/cc580117/attachment-0001.htm>

From enrico.ng at lmco.com  Wed Jan 20 23:51:05 2010
From: enrico.ng at lmco.com (Ng, Enrico)
Date: Wed, 20 Jan 2010 17:51:05 -0500
Subject: [C++-sig] Arrays to python
Message-ID: <D2FE1F40C1088A4B9F9CC9EECE74D90AC008A38D@HVXMSPA.us.lmco.com>

I have C code similar to the following:

float image_data[IMAGE_SIZE];
const float *get_data()

get_data returns a pointer to image_data.  I'd like to expose get_data to python and have it return a copy of the array in some Python type.  (it is actually a 2D array of image data)


In the wrapper code, I tried something like:

numeric::array  py_get_data()
{
  return numeric::array(get_data());
}

With 
  .def("get_data", &wrap::py_get_data)


But I get this error:

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


I thought I had done this in the past but I must be missing something.  Any help would be much appreciated.


From gjcarneiro at gmail.com  Thu Jan 21 00:28:47 2010
From: gjcarneiro at gmail.com (Gustavo Carneiro)
Date: Wed, 20 Jan 2010 23:28:47 +0000
Subject: [C++-sig] Boost-python/pybindgen and wrapping member variables
In-Reply-To: <27AD07BD25612C40AF5C78CB49D09B4967CC0F7063@34093-MBX-C08.mex07a.mlsrvr.com>
References: <27AD07BD25612C40AF5C78CB49D09B4967CC0F6FDE@34093-MBX-C08.mex07a.mlsrvr.com>
	<a467ca4f1001201017g7007f24cl9baa6ec88252333f@mail.gmail.com>
	<27AD07BD25612C40AF5C78CB49D09B4967CC0F7063@34093-MBX-C08.mex07a.mlsrvr.com>
Message-ID: <a467ca4f1001201528l345603a2tb4f041c15f5f7a96@mail.gmail.com>

On Wed, Jan 20, 2010 at 7:26 PM, Ben Fitzpatrick <
bfitzpatrick at vtiinstruments.com> wrote:

>  Gustavo,
>
>
>
> I think you win some sort of prize for not only fastest mailing list
> responses, but also fastest implementation of features. I reviewed the diff,
> and changed what I was adding to the global annotations table for these
> pointers to ?{?return_internal_reference?:?true?}? ? this seems to be what
> the new code in foo.h was doing.  I am still getting the errors, though ? is
> this the right way of using it?
>
I am not sure what is wrong.  Assuming you are using a "pre_scan_hook", and
that you are doing something like
global_annotations[?return_internal_reference?]
= True, in the proper attribute definition, then it should work.  Is it the
same error message, or a different error?


>
> And just to make sure, I grabbed a new checkout of pybindgen and
> re-installed it.
>
>
>
> *From:* cplusplus-sig-bounces+bfitzpatrick=vtiinstruments.com at python.org[mailto:
> cplusplus-sig-bounces+bfitzpatrick <cplusplus-sig-bounces%2Bbfitzpatrick>=
> vtiinstruments.com at python.org] *On Behalf Of *Gustavo Carneiro
> *Sent:* Wednesday, January 20, 2010 1:18 PM
> *To:* Development of Python/C++ integration
> *Subject:* Re: [C++-sig] Boost-python/pybindgen and wrapping member
> variables
>
>
>
>
>
> On Wed, Jan 20, 2010 at 5:43 PM, Ben Fitzpatrick <
> bfitzpatrick at vtiinstruments.com> wrote:
>
> Hi everyone,
>
>
>
> I think I?m finally getting the hang of this, I seem to have gotten almost
> everything in my project working thanks to Gustavo?s latest pybindgen patch.
> One of the remaining things I am having difficulty with is putting
> annotations on member variables of a class. I have something of the form:
>
>
>
> class foo{
>
>  <stuff>
>
> };
>
>
>
> class baz {
>
>   public:
>
>   foo* var;
>
> }
>
>
>
> The error I am getting from pybindgen is ?WrapperWarning: Return value
> ?::foo *? error (used in baz::var [variable]):
> TypeConfigurationError(?Either caller_owns_return or
> self.reference_existing_object must be given?)
>
>
>
>  I know that the issue is that I need to stick an annotation on there.
> Since the pointer is allocated and tracked by the underlying C++ code, I?m
> pretty sure that I need to put reference_existing_object on it. I?ve tried
> inserting ?{?reference_existing_object? : ?true?}? into the
> global_annotations table, and also inserting it into the
> parameter_annotations[?return?] dictionary. For kicks, I even tried both at
> once. No matter what I do I can?t get rid of the error, or see it show up in
> my final build, which leads me to believe there?s something I?m missing. I
> went and looked up the documentation for reference_existing_object, but of
> course none of their examples show how you could use it on a member
> variable. What am I missing?
>
> You are not missing anything.  Annotations for member variables (pybindgen
> calls them "instance attributes") were not implemented.
>
>
>
> An oversight easily fixed:
>
>
>
> http://bazaar.launchpad.net/~gjc/pybindgen/trunk/revision/742
>
>
>
>
>
> Thanks,
>
> Ben Fitzpatrick
>
>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>
>
>
>
> --
> Gustavo J. A. M. Carneiro
> INESC Porto, Telecommunications and Multimedia Unit
> "The universe is always one step beyond logic." -- Frank Herbert
>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>



-- 
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/20100120/6bf015e0/attachment.htm>

From talljimbo at gmail.com  Thu Jan 21 02:02:45 2010
From: talljimbo at gmail.com (Jim Bosch)
Date: Wed, 20 Jan 2010 17:02:45 -0800
Subject: [C++-sig] Arrays to python
In-Reply-To: <D2FE1F40C1088A4B9F9CC9EECE74D90AC008A38D@HVXMSPA.us.lmco.com>
References: <D2FE1F40C1088A4B9F9CC9EECE74D90AC008A38D@HVXMSPA.us.lmco.com>
Message-ID: <1264035765.10584.67.camel@mu>

On Wed, 2010-01-20 at 17:51 -0500, Ng, Enrico wrote:
> I have C code similar to the following:
> 
> float image_data[IMAGE_SIZE];
> const float *get_data()
> 
> get_data returns a pointer to image_data.  I'd like to expose get_data to python and have it return a copy of the array in some Python type.  (it is actually a 2D array of image data)
> 

I have to admit I'm not really familiar with using
boost::python::numeric; I don't know how well it works with numeric and
numarray's successor, numpy, and I pretty much use numpy exclusively (no
one is supporting the other two anymore).

But you can do this by using the numpy C API directly:

http://docs.scipy.org/doc/numpy/reference/c-api.array.html#from-scratch

and you can wrap the PyObject* you get back in a boost::python::object
like so:

#include <numpy/arrayobject.h>

boost::python::object py_get_data() {
    npy_intp shape[1] = { IMAGE_SIZE };    
    npy_intp strides[1] = { 1 };
    boost::python::handle<> array(
        PyArray_New(&PyArray_Type, 1, shape, NPY_FLOAT, strides,
                    get_data(), sizeof(float), NPY_C_ARRAY_RO, NULL)
    );  // handle is a C++ smart pointer for PyObject*
    return boost::python::object(array);
}

It's not elegant, but it should work.  It should even handle exceptions
in PyArray_New properly.  But you also have to do some work in the
module init function (in other words, inside the BOOST_PYTHON_MODULE
macro):

http://docs.scipy.org/doc/numpy/reference/c-api.array.html#importing-the-api

As a bit of self-promotion, I should mention I've written a fairly
extensive C++ multidimensional array library that provides automatic
boost::python converters to numpy, and might do what you're looking for
better:

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

Documentation for the Python stuff isn't great, but I'd be happy to
answer any questions.


Jim Bosch




From micdestefano at gmail.com  Thu Jan 21 09:32:57 2010
From: micdestefano at gmail.com (Michele De Stefano)
Date: Thu, 21 Jan 2010 09:32:57 +0100
Subject: [C++-sig] Arrays to python
In-Reply-To: <D2FE1F40C1088A4B9F9CC9EECE74D90AC008A38D@HVXMSPA.us.lmco.com>
References: <D2FE1F40C1088A4B9F9CC9EECE74D90AC008A38D@HVXMSPA.us.lmco.com>
Message-ID: <c410adc91001210032w6cd9ef7esa61d96f4ec643b68@mail.gmail.com>

If you can build a wrapper function which returns a
boost::numeric::ublas::vector<float>, you can use the wrappers into
the mds-utils library (http://code.google.com/p/mds-utils/  ...  it's
my library). If you build the doxygen documentation, you'll find a
"Python C++ extensions utilities": use the "vector.hpp" and look at
the usage examples.

Bye.
Michele

2010/1/20 Ng, Enrico <enrico.ng at lmco.com>:
> I have C code similar to the following:
>
> float image_data[IMAGE_SIZE];
> const float *get_data()
>
> get_data returns a pointer to image_data. ?I'd like to expose get_data to python and have it return a copy of the array in some Python type. ?(it is actually a 2D array of image data)
>
>
> In the wrapper code, I tried something like:
>
> numeric::array ?py_get_data()
> {
> ?return numeric::array(get_data());
> }
>
> With
> ?.def("get_data", &wrap::py_get_data)
>
>
> But I get this error:
>
> TypeError: No to_python (by-value) converter found for C++ type: float
>
>
> I thought I had done this in the past but I must be missing something. ?Any help would be much appreciated.
>
> _______________________________________________
> Cplusplus-sig mailing list
> Cplusplus-sig at python.org
> http://mail.python.org/mailman/listinfo/cplusplus-sig
>



-- 
Michele De Stefano
http://www.linkedin.com/in/micdestefano
http://code.google.com/p/mds-utils
http://xoomer.virgilio.it/michele_de_stefano

From bfitzpatrick at vtiinstruments.com  Thu Jan 21 14:43:17 2010
From: bfitzpatrick at vtiinstruments.com (Ben Fitzpatrick)
Date: Thu, 21 Jan 2010 07:43:17 -0600
Subject: [C++-sig] Boost-python/pybindgen and wrapping member variables
In-Reply-To: <a467ca4f1001201528l345603a2tb4f041c15f5f7a96@mail.gmail.com>
References: <27AD07BD25612C40AF5C78CB49D09B4967CC0F6FDE@34093-MBX-C08.mex07a.mlsrvr.com>
	<a467ca4f1001201017g7007f24cl9baa6ec88252333f@mail.gmail.com>
	<27AD07BD25612C40AF5C78CB49D09B4967CC0F7063@34093-MBX-C08.mex07a.mlsrvr.com>
	<a467ca4f1001201528l345603a2tb4f041c15f5f7a96@mail.gmail.com>
Message-ID: <27AD07BD25612C40AF5C78CB49D09B4967CC0F72A9@34093-MBX-C08.mex07a.mlsrvr.com>

Interesting, I changed it to the format you suggest and it seems to work. However, I changed it back to this way:
global_annotations = {?return_internal_reference?:?true?}

and it stopped working. I confirmed there was nothing in there before I added my annotation, so that seems strange to me. On the other hand, your solution works, so I?ll just use that.

Thanks again,
Ben


From: cplusplus-sig-bounces+bfitzpatrick=vtiinstruments.com at python.org [mailto:cplusplus-sig-bounces+bfitzpatrick=vtiinstruments.com at python.org] On Behalf Of Gustavo Carneiro
Sent: Wednesday, January 20, 2010 6:29 PM
To: Development of Python/C++ integration
Subject: Re: [C++-sig] Boost-python/pybindgen and wrapping member variables


On Wed, Jan 20, 2010 at 7:26 PM, Ben Fitzpatrick <bfitzpatrick at vtiinstruments.com<mailto:bfitzpatrick at vtiinstruments.com>> wrote:
Gustavo,

I think you win some sort of prize for not only fastest mailing list responses, but also fastest implementation of features. I reviewed the diff, and changed what I was adding to the global annotations table for these pointers to ?{?return_internal_reference?:?true?}? ? this seems to be what the new code in foo.h was doing.  I am still getting the errors, though ? is this the right way of using it?
I am not sure what is wrong.  Assuming you are using a "pre_scan_hook", and that you are doing something like global_annotations[?return_internal_reference?] = True, in the proper attribute definition, then it should work.  Is it the same error message, or a different error?


And just to make sure, I grabbed a new checkout of pybindgen and re-installed it.

From: cplusplus-sig-bounces+bfitzpatrick=vtiinstruments.com<http://vtiinstruments.com>@python.org<http://python.org> [mailto:cplusplus-sig-bounces+bfitzpatrick<mailto:cplusplus-sig-bounces%2Bbfitzpatrick>=vtiinstruments.com<http://vtiinstruments.com>@python.org<http://python.org>] On Behalf Of Gustavo Carneiro
Sent: Wednesday, January 20, 2010 1:18 PM
To: Development of Python/C++ integration
Subject: Re: [C++-sig] Boost-python/pybindgen and wrapping member variables


On Wed, Jan 20, 2010 at 5:43 PM, Ben Fitzpatrick <bfitzpatrick at vtiinstruments.com<mailto:bfitzpatrick at vtiinstruments.com>> wrote:
Hi everyone,

I think I?m finally getting the hang of this, I seem to have gotten almost everything in my project working thanks to Gustavo?s latest pybindgen patch. One of the remaining things I am having difficulty with is putting annotations on member variables of a class. I have something of the form:

class foo{
 <stuff>
};

class baz {
  public:
  foo* var;
}

The error I am getting from pybindgen is ?WrapperWarning: Return value ?::foo *? error (used in baz::var [variable]): TypeConfigurationError(?Either caller_owns_return or self.reference_existing_object must be given?)

 I know that the issue is that I need to stick an annotation on there. Since the pointer is allocated and tracked by the underlying C++ code, I?m pretty sure that I need to put reference_existing_object on it. I?ve tried inserting ?{?reference_existing_object? : ?true?}? into the global_annotations table, and also inserting it into the parameter_annotations[?return?] dictionary. For kicks, I even tried both at once. No matter what I do I can?t get rid of the error, or see it show up in my final build, which leads me to believe there?s something I?m missing. I went and looked up the documentation for reference_existing_object, but of course none of their examples show how you could use it on a member variable. What am I missing?
You are not missing anything.  Annotations for member variables (pybindgen calls them "instance attributes") were not implemented.

An oversight easily fixed:

http://bazaar.launchpad.net/~gjc/pybindgen/trunk/revision/742


Thanks,
Ben Fitzpatrick

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



--
Gustavo J. A. M. Carneiro
INESC Porto, Telecommunications and Multimedia Unit
"The universe is always one step beyond logic." -- Frank Herbert

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



--
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/20100121/9555a679/attachment-0001.htm>

From enrico.ng at lmco.com  Thu Jan 21 15:10:41 2010
From: enrico.ng at lmco.com (Ng, Enrico)
Date: Thu, 21 Jan 2010 09:10:41 -0500
Subject: [C++-sig] Arrays to python
In-Reply-To: <1264035765.10584.67.camel@mu>
References: <D2FE1F40C1088A4B9F9CC9EECE74D90AC008A38D@HVXMSPA.us.lmco.com>
	<1264035765.10584.67.camel@mu>
Message-ID: <D2FE1F40C1088A4B9F9CC9EECE74D90AC008A7AD@HVXMSPA.us.lmco.com>

That was just my example, I do not require it to go into a numeric array.  It just seemed like the way I was suppose to do it based on the documentation.  As long as I can access the data in any form in python, I'll be happy.  And I only need to read the data, I do not need to modify the array.

I did see your library but I was hoping to be able to do this without incorporating another library. 

I had hoped that there was something simple after seeing the numeric part in the documentation.

--
Enrico Ng
 

-----Original Message-----
From: Jim Bosch [mailto:talljimbo at gmail.com] 
Sent: Wednesday, January 20, 2010 8:03 PM
To: cplusplus-sig at python.org
Cc: Ng, Enrico
Subject: Re: [C++-sig] Arrays to python

On Wed, 2010-01-20 at 17:51 -0500, Ng, Enrico wrote:
> I have C code similar to the following:
> 
> float image_data[IMAGE_SIZE];
> const float *get_data()
> 
> get_data returns a pointer to image_data.  I'd like to expose get_data to python and have it return a copy of the array in some Python type.  (it is actually a 2D array of image data)
> 

I have to admit I'm not really familiar with using
boost::python::numeric; I don't know how well it works with numeric and
numarray's successor, numpy, and I pretty much use numpy exclusively (no
one is supporting the other two anymore).

But you can do this by using the numpy C API directly:

http://docs.scipy.org/doc/numpy/reference/c-api.array.html#from-scratch

and you can wrap the PyObject* you get back in a boost::python::object
like so:

#include <numpy/arrayobject.h>

boost::python::object py_get_data() {
    npy_intp shape[1] = { IMAGE_SIZE };    
    npy_intp strides[1] = { 1 };
    boost::python::handle<> array(
        PyArray_New(&PyArray_Type, 1, shape, NPY_FLOAT, strides,
                    get_data(), sizeof(float), NPY_C_ARRAY_RO, NULL)
    );  // handle is a C++ smart pointer for PyObject*
    return boost::python::object(array);
}

It's not elegant, but it should work.  It should even handle exceptions
in PyArray_New properly.  But you also have to do some work in the
module init function (in other words, inside the BOOST_PYTHON_MODULE
macro):

http://docs.scipy.org/doc/numpy/reference/c-api.array.html#importing-the-api

As a bit of self-promotion, I should mention I've written a fairly
extensive C++ multidimensional array library that provides automatic
boost::python converters to numpy, and might do what you're looking for
better:

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

Documentation for the Python stuff isn't great, but I'd be happy to
answer any questions.


Jim Bosch




From talljimbo at gmail.com  Thu Jan 21 22:17:08 2010
From: talljimbo at gmail.com (Jim Bosch)
Date: Thu, 21 Jan 2010 13:17:08 -0800
Subject: [C++-sig] Arrays to python
In-Reply-To: <D2FE1F40C1088A4B9F9CC9EECE74D90AC008A7AD@HVXMSPA.us.lmco.com>
References: <D2FE1F40C1088A4B9F9CC9EECE74D90AC008A38D@HVXMSPA.us.lmco.com>
	<1264035765.10584.67.camel@mu>
	<D2FE1F40C1088A4B9F9CC9EECE74D90AC008A7AD@HVXMSPA.us.lmco.com>
Message-ID: <1264108628.13173.71.camel@mu>

On Thu, 2010-01-21 at 09:10 -0500, Ng, Enrico wrote:
> I had hoped that there was something simple after seeing the numeric part in the documentation.

Well, if you're content with a 1-D array and speed isn't a big issue,
you can just copy the elements of your array into a Python list like
this:

boost::python::list py_get_data() {
    boost::python::list r;
    for (int n=0; n<IMAGE_SIZE; ++n) {
        r.append(get_data()[n]);
    }
    return r;
}

If you want two dimensions, you could make a list-of-lists.
I don't think there's any way to get a C array into a Python object
without lots of explicit copying aside from using numpy.


Jim Bosch



From enrico.ng at lmco.com  Thu Jan 21 22:31:12 2010
From: enrico.ng at lmco.com (Ng, Enrico)
Date: Thu, 21 Jan 2010 16:31:12 -0500
Subject: [C++-sig] Arrays to python
In-Reply-To: <1264108628.13173.71.camel@mu>
References: <D2FE1F40C1088A4B9F9CC9EECE74D90AC008A38D@HVXMSPA.us.lmco.com>
	<1264035765.10584.67.camel@mu>
	<D2FE1F40C1088A4B9F9CC9EECE74D90AC008A7AD@HVXMSPA.us.lmco.com>
	<1264108628.13173.71.camel@mu>
Message-ID: <D2FE1F40C1088A4B9F9CC9EECE74D90AC00EF6F8@HVXMSPA.us.lmco.com>

Thanks, this is pretty much what I ended up doing.
Once I got that into python, I converted it to a narray

 

-----Original Message-----
From: cplusplus-sig-bounces+enrico.ng=lmco.com at python.org [mailto:cplusplus-sig-bounces+enrico.ng=lmco.com at python.org] On Behalf Of Jim Bosch
Sent: Thursday, January 21, 2010 4:17 PM
To: Development of Python/C++ integration
Subject: Re: [C++-sig] Arrays to python

On Thu, 2010-01-21 at 09:10 -0500, Ng, Enrico wrote:
> I had hoped that there was something simple after seeing the numeric part in the documentation.

Well, if you're content with a 1-D array and speed isn't a big issue,
you can just copy the elements of your array into a Python list like
this:

boost::python::list py_get_data() {
    boost::python::list r;
    for (int n=0; n<IMAGE_SIZE; ++n) {
        r.append(get_data()[n]);
    }
    return r;
}

If you want two dimensions, you could make a list-of-lists.
I don't think there's any way to get a C array into a Python object
without lots of explicit copying aside from using numpy.


Jim Bosch


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

From daniel.lidstrom at sbg.se  Fri Jan 22 16:33:42 2010
From: daniel.lidstrom at sbg.se (=?iso-8859-1?Q?Daniel_Lidstr=F6m?=)
Date: Fri, 22 Jan 2010 16:33:42 +0100
Subject: [C++-sig] [Boost.Python] Recommended file structure layout?
Message-ID: <FE8B524F1BF0854187E434B85105AC693AB7C5@dorado.ad.sbg.se>

Hello,

I am very interested in using Boost.Python to test my C++ codebase.
However, I would like to have some tips on a typical layout of the
files in my project. My codebase today is completely C++-based and
I also have tests using cppunit. Here's my file structure of the
GFL library (Windows dll):

GFL/
   include/
           GFL   <- headers go here
   src/          <- C++ source goes here (creates GFL.dll)
   Test/         <- cppunit tests go here (Test.exe links with GFL.dll)

Now I need to create the python->C++ bridge using Boost.Python.
This has to be separate from the GFL dll as I don't want to link
with python in the main dll at this stage. Here's what I am thinking:

GFL/
   include/
           GFL
   src/
   Test/
   PyBridge/     <- source files with Boost.Python bridge

This would create two libraries, GFL.dll and GFL.PyBridge.dll. I am
unsure of the naming convention. Now, I probably would put python
test files in PyTest, like so:

GFL/
   include/
           GFL
   src/
   Test/
   PyBridge/
   PyTest/       <- nose tests here :-)

What do you think about this structure? Anybody here doing similar things?
I would really appreciate some feedback! :-)

Now to another question (sorry if I am too off-topic on this one).
We are using TeamCity for continuous integration. My goal is to integrate
nose-tests using TeamCity. I can already vision how that would work.
But I would also like to make GFL.dll and GFL.PyBridge.dll available to
my integration&verification team, so that they can create their own python
scripts for verification tests. Is there anyone here who has this kind of
solution in place? If so, how do you make the compiled binaries easily
available to the I&V team? If this is really too off-topic I'll take my
question to the TeamCity forums (might do it anyway).

Thanks in advance!

Regards,

Daniel Lidstr?m
Stockholm, Sweden

From the.groovy.gregory at googlemail.com  Fri Jan 22 19:33:09 2010
From: the.groovy.gregory at googlemail.com (MrF)
Date: Fri, 22 Jan 2010 19:33:09 +0100
Subject: [C++-sig] [Boost Python] Extended class loosing attributes when
 saving as pointer to BaseClass in C++
Message-ID: <4B59EF65.5070009@gmail.com>

Hey,

My problem is the following:

I do have a C++ class, "BaseClass", that I am extending in Python 
("SubClass"). Also there is another C++ class, let's call it 
"BaseClassManager" which has a vector of BaseClass* (well, actually it's 
holding shared_ptr's). Instances of "SubClass" are created in Python, 
but ownership has been transferred to C++, as "BaseClassManager" holds 
all the objects.

Here is the code that illustrates this:

Wrapped BaseClassManager
------------------------------------
class_<BaseClassManager>("GameBaseX")
             .def("addObject", &BaseClassManager::addObject) // takes 
shared_ptr< BaseClass>&
             .def("getObjectByName", 
&BaseClassManager::getObjectByName); // takes a string, returns 
shared_ptr<BaseClass>


WrapperClass for Base (needed for transferring ownership)
------------------------------------
class BaseClassWrap : public BaseClass{
     public:
        BaseClassWrap( PyObject* self_, const std::string& name)
             : BaseClass(name),self(self_) { Py_INCREF(self); }
         BaseClassWrap( PyObject* self_, const BaseClass& copy ) : 
BaseClass(copy), self(self_) { Py_INCREF(self); }
         ~BaseClassWrap() { Py_DECREF(self); }

         PyObject *self;
     };

Wrapped BaseClass
------------------------------------
class_<BaseClass, shared_ptr<BaseClassWrap> >("BaseClass",
             init<const std::string&>())
             .add_property("name", make_function(&BaseClass::getName, 
return_value_policy<copy_const_reference>()) );

// both needed for using shared_ptr's together with the wrapper-class
register_ptr_to_python<shared_ptr<BaseClass> >();    // to_python converter
implicitly_convertible<shared_ptr<BaseClassWrap>, shared_ptr<BaseClass> >();

Subclass
------------------------------------
class Subclass(BaseClass):
     def __init__(self, name):
         BaseClass.__init__(self, name)
         self.foo = "bar"

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

Imagine the following Python code:

obj = SubClass("superclass")
BaseClassManager.addObject(obj)

# and somewhere else

obj = BaseClassManager.getObjectByName("superclass")
print obj.name     # prints "superclass" as expected
print obj.foo         # AttributeError: 'BaseClass' object has no 
attribute 'foo'

Seems all attributes got lost, be it a member-variable or methods I 
defined in SubClass. Strangely obj is in both cases the same pointer (f. 
ex. "SubClass object at 0x02343345").

Is there something I can do to preserve the attributes? Otherwise it is 
kind of senseless to extend a C++ class in Python if it can't have its 
own attributes.

Thanks a lot for the help.

Cheers.


From talljimbo at gmail.com  Fri Jan 22 21:05:32 2010
From: talljimbo at gmail.com (Jim Bosch)
Date: Fri, 22 Jan 2010 12:05:32 -0800
Subject: [C++-sig] [Boost Python] Extended class loosing attributes when
 saving as pointer to BaseClass in C++
In-Reply-To: <4B59EF65.5070009@gmail.com>
References: <4B59EF65.5070009@gmail.com>
Message-ID: <1264190732.19156.17.camel@mu>

I think I can explain what's going on, and I've got an idea for a
partial solution.

But first, try making BaseClassWrap multiply inherit from BaseClass and
boost::python::wrapper<BaseClass>.  That might fix everything.  If not,
read on...

When boost::python converts a C++ value to Python, it looks up the C++
RTTI type name in a registry and finds a converter for that type.  Those
converters generally make a new Python object, and put a
boost::python::instance_holder in it, containing a C++ object by value
or [smart] pointer.  But what you want is to extract the PyObject* from
the C++ object and just incref and return that.  I think you could do
this by writing your own Python version of BaseClassManager that checks
for this possibility:


object PyBaseClassManager_getObjectByName(
    BaseClassManager & manager,
    std::string const & name
) {
    shared_ptr<BaseClass> r = manager.getObjectByName(name);
    shared_ptr<BaseClassWrap> w =
        dynamic_pointer_cast<BaseClassWrap>(r);
    if (w) {
        handle<> h(w->self);
        return object(h);
    }
    return object(r);
}

...unfortunately, you'd have to do that with ANY C++ function that
returns a shared_ptr<BaseClass>.  A better solution would be to write a
custom to-python conversion for shared_ptr<BaseClass>, or if necessary a
custom ResultConverterGenerator to do that work, but that's a little
more complicated.

This is probably something boost::python should do automatically if
BaseClassWrap is a subclass of wrapper<BaseClass>.  If it doesn't, and
you can get me a complete example that demonstrates the problem, I'll
see about making a patch to make that happen.

Jim Bosch


From the.groovy.gregory at googlemail.com  Fri Jan 22 22:13:40 2010
From: the.groovy.gregory at googlemail.com (MrF)
Date: Fri, 22 Jan 2010 22:13:40 +0100
Subject: [C++-sig] [Boost Python] Extended class loosing attributes when
 saving as pointer to BaseClass in C++
In-Reply-To: <1264190732.19156.17.camel@mu>
References: <4B59EF65.5070009@gmail.com> <1264190732.19156.17.camel@mu>
Message-ID: <4B5A1504.9090507@gmail.com>


Thanks for help!

> A better solution would be to write a
> custom to-python conversion for shared_ptr<BaseClass>
>    

Tried it and works perfectly. Good hint!

> This is probably something boost::python should do automatically if
> BaseClassWrap is a subclass of wrapper<BaseClass>.  If it doesn't, and
> you can get me a complete example that demonstrates the problem, I'll
> see about making a patch to make that happen

I'll have some free days in February so I'll  put this into a small 
project for you to play around with, so the whole procedure can be made 
a little easier.

Thanks a lot for your help!

Cheers.


From yunlin07 at gmail.com  Sat Jan 23 02:29:00 2010
From: yunlin07 at gmail.com (lin yun)
Date: Fri, 22 Jan 2010 17:29:00 -0800
Subject: [C++-sig] Please help: boost python 1.41 extension problem
Message-ID: <e3c115371001221729j200cc479yf75a644aede2dbcd@mail.gmail.com>

Hi,

I am trying to install boost 1.41 on Redhat Enterprise Linux 5.3 and I got
problem with the extension test. Following the instructions here:
http://www.boost.org/doc/libs/1_41_0/libs/python/doc/building.html

   - I built boost python using bjam.
   - cd into the libs/python/example/quickstart/ directory
   - bjam toolset=gcc --verbose-test test

Now I see embedding test passes while extension test failed. I got the same
problem with boost 1.35, 1.37 on Fedora 7. Below is the error messages:


../../../../boost/type_traits/detail/cv_traits_impl.hpp:37: internal
compiler error: in make_rtl_for_nonlocal_decl, at cp/decl.c:5067

Please submit a full bug report,

with preprocessed source if appropriate.

See <URL:http://bugzilla.redhat.com/bugzilla> for instructions.

Preprocessed source stored into /tmp/ccIhs3ut.out file, please attach this
to your bugreport.

"g++" -ftemplate-depth-128 -O0 -fno-inline -Wall -g -fPIC -I"../../../.."
-I"/usr/include/python2.4" -c -o "bin/gcc-4.1.2/debug/extending.o"
"extending.cpp"

...failed gcc.compile.c++ bin/gcc-4.1.2/debug/extending.o...

...skipped <pbin/gcc-4.1.2/debug>extending.so for lack of
<pbin/gcc-4.1.2/debug>extending.o...

...skipped <pbin/test_ext.test/gcc-4.1.2/debug>test_ext for lack of
<pbin/gcc-4.1.2/debug>extending.so...

Does anyone encounter the same issue as me ? Anyone can help ?

Thanks a lot,

Yun
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100122/901d7901/attachment-0001.htm>

From gerasimov at mlab.cs.msu.su  Sat Jan 23 16:59:33 2010
From: gerasimov at mlab.cs.msu.su (sergun)
Date: Sat, 23 Jan 2010 15:59:33 +0000 (UTC)
Subject: [C++-sig] copy boost::python::object crashes
Message-ID: <loom.20100123T165831-858@post.gmane.org>

I have a problem with storing reference to Python object inside C++.

The following Boost.Python wrapping class:



#include <boost/python.hpp>

using namespace boost;
using namespace boost::python;

namespace cmodule
{

object y;

void setY(object y1) { y = y1; }

}

using namespace cmodule;

BOOST_PYTHON_MODULE(cmodule)
{
	def("setY",setY);
}




Crashes with ?Segmentation fault? on so simple python fragment.




import cmodule

class Foo:
    pass

cmodule.setY(Foo())




When I comment ?y = y1;? assignment everything is fine.

What I?m doing wrong?

Python version: 2.6
Boost version: 1.41
OS: FreeBSD 8.0

Sergey Gerasimov.



From talljimbo at gmail.com  Sat Jan 23 22:38:47 2010
From: talljimbo at gmail.com (Jim Bosch)
Date: Sat, 23 Jan 2010 13:38:47 -0800
Subject: [C++-sig] copy boost::python::object crashes
In-Reply-To: <loom.20100123T165831-858@post.gmane.org>
References: <loom.20100123T165831-858@post.gmane.org>
Message-ID: <1264282727.3199.6.camel@bishop>

On Sat, 2010-01-23 at 15:59 +0000, sergun wrote:
> I have a problem with storing reference to Python object inside C++.
> 
> The following Boost.Python wrapping class:
> 
> 
> 
> #include <boost/python.hpp>
> 
> using namespace boost;
> using namespace boost::python;
> 
> namespace cmodule
> {
> 
> object y;
> 
> void setY(object y1) { y = y1; }
> 
> }
> 
> using namespace cmodule;
> 
> BOOST_PYTHON_MODULE(cmodule)
> {
> 	def("setY",setY);
> }
> 
> 
> 
> 
> Crashes with ?Segmentation fault? on so simple python fragment.
> 

This might be a static initialization problem: you declare "y", but it's
not clear from your code fragment whether it gets initialized.  I would
have expected that to result in a linker error, but the way Python does
its dynamic loading might have suppressed that.

It's probably a good idea to avoid global variables in source files
anyhow.  Wrapping it in a function:

object & getY() {
    static object y;
    return y;
}

void setY(object y1) { getY() = y1; }

might eliminate your segfault.


Jim Bosch


From dave at boostpro.com  Sun Jan 24 03:10:31 2010
From: dave at boostpro.com (David Abrahams)
Date: Sat, 23 Jan 2010 21:10:31 -0500
Subject: [C++-sig] boost python calling conventions
	support	patch	submission (__stdcall, __cdecl, __fastcall)
In-Reply-To: <4B50A150.1010908@mgdesign.org>
References: <4B437643.5000103@mgdesign.org>
	<201001101118.25701.lists_ravi@lavabit.com>
	<4B50A150.1010908@mgdesign.org>
Message-ID: <m2d410ckx4.wl%dave@boostpro.com>

At Fri, 15 Jan 2010 18:09:36 +0100,
Nicolas Lelong wrote:
> 
> My previous mail once again got lost, gotta be more careful...
> 
> > Changes (by dave):
> >
> >   This looks terrific!  I'll try to get it applied ASAP.  Please poke me if
> >   you don't see it in trunk by the end of the week.
> 
> Thanks ! As I didn't see the batch in trunk today, I poke you :)

Applied; thanks for your patience.

-- 
Dave Abrahams           Meet me at BoostCon: http://www.boostcon.com
BoostPro Computing
http://www.boostpro.com


From gerasimov at mlab.cs.msu.su  Sun Jan 24 10:30:22 2010
From: gerasimov at mlab.cs.msu.su (sergun)
Date: Sun, 24 Jan 2010 09:30:22 +0000 (UTC)
Subject: [C++-sig] copy boost::python::object crashes
References: <loom.20100123T165831-858@post.gmane.org>
	<1264282727.3199.6.camel@bishop>
Message-ID: <loom.20100124T102807-506@post.gmane.org>

Hi Jim,

In my opinion ?object y;? is just declaration of global object, not static 
variable of some class (in this case you are right: explicit declaration 
needed, something like: ?object MyStaticClass::y;?).

I have changed setY function to:
void setY(PyObject * y1) { y = object(handle<>(y1)); }

and it doesn?t make crash now.
I?m not sure that such code has no problems and very interested to know why 
assignment of one bp::object to another 
is not correct. I suppose bp:object counteres references correctly and the 
original object is killed only on Python.

Sergey.



From talljimbo at gmail.com  Mon Jan 25 02:32:12 2010
From: talljimbo at gmail.com (Jim Bosch)
Date: Sun, 24 Jan 2010 17:32:12 -0800
Subject: [C++-sig] copy boost::python::object crashes
In-Reply-To: <loom.20100124T102807-506@post.gmane.org>
References: <loom.20100123T165831-858@post.gmane.org>
	<1264282727.3199.6.camel@bishop>
	<loom.20100124T102807-506@post.gmane.org>
Message-ID: <1264383132.4015.96.camel@bishop>

On Sun, 2010-01-24 at 09:30 +0000, sergun wrote:
> I have changed setY function to:
> void setY(PyObject * y1) { y = object(handle<>(y1)); }
> 
> and it doesn?t make crash now.
> I?m not sure that such code has no problems and very interested to know why 
> assignment of one bp::object to another 
> is not correct. I suppose bp:object counteres references correctly and the 
> original object is killed only on Python.
> 

As it turns out, the new code is quite likely worse.  Here's what's
probably going on:

First, I should mention I get the segfault when quitting python after
running the python code, not after executing the last statement, so if
that's not what you experienced, there may be more going on.

Anyhow, with the original version, the global variable holds the only
reference to the Foo() object.  When Python exits, and the destructor
for the global variable executes, it DECREF's a Python object after a
lot of Python's machinery has already been shut down.  Interestingly, it
doesn't fail if you do setY("a string") or setY(5), but that's not much
consolation.

With the new version, the global variable ends up holding a Foo object
with ZERO reference count - I'm not sure, but I think it's already been
destroyed by the time the Python setY() call finishes, and the C++
global variable is just a ticking time bomb.  But since the reference
count is zero, nothing happens when the global's destructor goes off.

I think this all means it's probably a very bad idea to have a global or
class-static boost::python::object (or handle<>) variable.  The best
workaround I can think of right now is to have a a global PyObject*
instead, and do manual calls to Py_DECREF and Py_INCREF in the C++ setY
function.  That means the last thing left in the global will never get
destroyed, but unless it has a non-trivial Python destruct that
shouldn't matter.  If you need to make sure it gets destroyed, I think
you'll end up having to use Python's atexit.

Jim



From nico_ml at mgdesign.org  Mon Jan 25 10:47:06 2010
From: nico_ml at mgdesign.org (Nicolas Lelong)
Date: Mon, 25 Jan 2010 10:47:06 +0100
Subject: [C++-sig] boost python calling conventions
 support	patch	submission (__stdcall, __cdecl, __fastcall)
In-Reply-To: <m2d410ckx4.wl%dave@boostpro.com>
References: <4B437643.5000103@mgdesign.org>	<201001101118.25701.lists_ravi@lavabit.com>	<4B50A150.1010908@mgdesign.org>
	<m2d410ckx4.wl%dave@boostpro.com>
Message-ID: <4B5D689A.40902@mgdesign.org>

An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100125/031f19dc/attachment.htm>

From dave at boostpro.com  Mon Jan 25 16:28:55 2010
From: dave at boostpro.com (David Abrahams)
Date: Mon, 25 Jan 2010 10:28:55 -0500
Subject: [C++-sig] boost python calling
	conventions	support	patch	submission (__stdcall, __cdecl,
	__fastcall)
In-Reply-To: <4B5D689A.40902@mgdesign.org>
References: <4B437643.5000103@mgdesign.org>
	<201001101118.25701.lists_ravi@lavabit.com>
	<4B50A150.1010908@mgdesign.org> <m2d410ckx4.wl%dave@boostpro.com>
	<4B5D689A.40902@mgdesign.org>
Message-ID: <m27hr6i4p4.wl%dave@boostpro.com>

At Mon, 25 Jan 2010 10:47:06 +0100,
Nicolas Lelong wrote:
> 
> Thanks Dave,
> 
> not quite sure of it, but it looks like you added the unit tests to
> the Jamfile, but you did not add the tests source code to the svn,
> did I miss something.

Gah, I hate SVN.

> Also, the tests i submitted do not consider the compiler
> brand/version, so they may fail on compilers that do not have the
> Microsoft calling conventions. Did you consider that point ?

No!  

The next question is, should defining the macros to turn on calling
convention support cause an error on platforms where they're not
supported (which would imply XML markup for those tests and/or #ifdef
_MSC_VER in the tests themselves) or should it be benign (implying
#ifdef _MSC_VER in the library files)?

Your opinion?

-- 
Dave Abrahams           Meet me at BoostCon: http://www.boostcon.com
BoostPro Computing
http://www.boostpro.com


From talljimbo at gmail.com  Mon Jan 25 21:20:05 2010
From: talljimbo at gmail.com (Jim Bosch)
Date: Mon, 25 Jan 2010 12:20:05 -0800
Subject: [C++-sig] Please help: boost python 1.41 extension problem
In-Reply-To: <e3c115371001221729j200cc479yf75a644aede2dbcd@mail.gmail.com>
References: <e3c115371001221729j200cc479yf75a644aede2dbcd@mail.gmail.com>
Message-ID: <1264450805.26787.17.camel@mu>

On Fri, 2010-01-22 at 17:29 -0800, lin yun wrote:
> Hi, 
>  
> I am trying to install boost 1.41 on Redhat Enterprise Linux 5.3 and I
> got problem with the extension test. Following the instructions here:
> http://www.boost.org/doc/libs/1_41_0/libs/python/doc/building.html
>       * I built boost python using bjam.
>       * cd into the libs/python/example/quickstart/ directory 
>       * bjam toolset=gcc --verbose-test test
> Now I see embedding test passes while extension test failed. I got the
> same problem with boost 1.35, 1.37 on Fedora 7. Below is the error
> messages:
>  
> ../../../../boost/type_traits/detail/cv_traits_impl.hpp:37: internal
> compiler error: in make_rtl_for_nonlocal_decl, at cp/decl.c:5067
> 
> Please submit a full bug report,
> 
> with preprocessed source if appropriate.
> 
> See <URL:http://bugzilla.redhat.com/bugzilla> for instructions.
> 

Your error message indicates a gcc internal error.  While it's probably
due at least in part to something unusual in your code or boost::python,
that kind of stuff should cause gcc to exit more gracefully (and
informatively) than this.

You might learn something by reporting your error message either to the
gcc developers (or Red Hat, since they're more directly responsible for
the binary you have).  At the very least, I imagine they'd like to hear
about it.

Jim Bosch





From yunlin07 at gmail.com  Mon Jan 25 21:38:44 2010
From: yunlin07 at gmail.com (lin yun)
Date: Mon, 25 Jan 2010 12:38:44 -0800
Subject: [C++-sig] Please help: boost python 1.41 extension problem
In-Reply-To: <1264450805.26787.17.camel@mu>
References: <e3c115371001221729j200cc479yf75a644aede2dbcd@mail.gmail.com>
	<1264450805.26787.17.camel@mu>
Message-ID: <e3c115371001251238u75ba3b48wf98669ed3655ccde@mail.gmail.com>

Hi, Jim:

Thank you very much for your reply !

I did not mention that I had no problem with Boost 1.34, on the same system.
Also I ever tried Boost 1.34 and 1.37 on Fedora 7, I got exactly the same
result: Boost 1.34 worked fine while 1.37 failed on extension test. The test
I ran was the example test from boost package:
libs/python/example/quickstart/, which was recommended in the boost python
installation instruction. There was no any personal code involved.

I wonder if you, or anyone else had related experience ?

Thanks again,

Yun



On Mon, Jan 25, 2010 at 12:20 PM, Jim Bosch <talljimbo at gmail.com> wrote:

> On Fri, 2010-01-22 at 17:29 -0800, lin yun wrote:
> > Hi,
> >
> > I am trying to install boost 1.41 on Redhat Enterprise Linux 5.3 and I
> > got problem with the extension test. Following the instructions here:
> > http://www.boost.org/doc/libs/1_41_0/libs/python/doc/building.html
> >       * I built boost python using bjam.
> >       * cd into the libs/python/example/quickstart/ directory
> >       * bjam toolset=gcc --verbose-test test
> > Now I see embedding test passes while extension test failed. I got the
> > same problem with boost 1.35, 1.37 on Fedora 7. Below is the error
> > messages:
> >
> > ../../../../boost/type_traits/detail/cv_traits_impl.hpp:37: internal
> > compiler error: in make_rtl_for_nonlocal_decl, at cp/decl.c:5067
> >
> > Please submit a full bug report,
> >
> > with preprocessed source if appropriate.
> >
> > See <URL:http://bugzilla.redhat.com/bugzilla> for instructions.
> >
>
> Your error message indicates a gcc internal error.  While it's probably
> due at least in part to something unusual in your code or boost::python,
> that kind of stuff should cause gcc to exit more gracefully (and
> informatively) than this.
>
> You might learn something by reporting your error message either to the
> gcc developers (or Red Hat, since they're more directly responsible for
> the binary you have).  At the very least, I imagine they'd like to hear
> about it.
>
> Jim Bosch
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100125/b6f2d809/attachment.htm>

From talljimbo at gmail.com  Mon Jan 25 22:06:13 2010
From: talljimbo at gmail.com (Jim Bosch)
Date: Mon, 25 Jan 2010 13:06:13 -0800
Subject: [C++-sig] Please help: boost python 1.41 extension problem
In-Reply-To: <e3c115371001251238u75ba3b48wf98669ed3655ccde@mail.gmail.com>
References: <e3c115371001221729j200cc479yf75a644aede2dbcd@mail.gmail.com>
	<1264450805.26787.17.camel@mu>
	<e3c115371001251238u75ba3b48wf98669ed3655ccde@mail.gmail.com>
Message-ID: <1264453573.27378.20.camel@mu>

On Mon, 2010-01-25 at 12:38 -0800, lin yun wrote:

> I did not mention that I had no problem with Boost 1.34, on the same
> system. Also I ever tried Boost 1.34 and 1.37 on Fedora 7, I got
> exactly the same result: Boost 1.34 worked fine while 1.37 failed on
> extension test. The test I ran was the example test from boost
> package: libs/python/example/quickstart/, which was recommended in
> the boost python installation instruction. There was no any personal
> code involved.
>  
> I wonder if you, or anyone else had related experience ?
>  

I have not, but I haven't tried boost > 1.34.1 on any Red Hat distros.

I can confirm that this example compiles successfully with Boost 1.40
and the latest trunk on Ubuntu 9.10 (gcc 4.4.1).

You may have noticed the line gcc aborts on:

>         >
>         > ../../../../boost/type_traits/detail/cv_traits_impl.hpp:37:
>         internal
>         > compiler error: in make_rtl_for_nonlocal_decl, at
>         cp/decl.c:5067
>         >

has not changed between Boost 1.34 and 1.35, but it refers to the macro
BOOST_STATIC_CONSTANT, which is defined in boost/config/suffix.hpp, and
reflects whether a compiler supports in-class initialization of static
integral constant members.  Perhaps your compiler does not fully support
this feature, but boost thinks it does?

Try adding the line

#define BOOST_NO_INCLASS_MEMBER_INITIALIZATION

to boost/config/user.hpp.  Then recompile boost.



Jim Bosch



From martin.manduch at gmail.com  Tue Jan 26 13:36:13 2010
From: martin.manduch at gmail.com (martin manduch)
Date: Tue, 26 Jan 2010 13:36:13 +0100
Subject: [C++-sig] wrapping vector of pointers
Message-ID: <92560a4e1001260436p3e547f38i3c749aa2ad09c114@mail.gmail.com>

Hi,

I need to wrap vector of pointers, because it's parameter in one function...

I have following code:

struct X
{
    X(int nn) {n=nn;}
    int n;
};


BOOST_PYTHON_MODULE(myVector)
{
    class_<X>("X", init<int>())
    .def_readwrite("n", &X::n);


    class_<std::vector<X*> >("XVec")
        .def(vector_indexing_suite<std::vector<X*> >())

    ;
}

I want to make XVec responsible for memory, which is addressed through
pointers, but I don't know how.

>>vec=XVec()
>>a=X(7)
>>vec.append(a)
>>del a
>>vec[0].n
It prints 1682309121, but I want it to print: 7

Martin

From daniel.lidstrom at sbg.se  Tue Jan 26 15:47:20 2010
From: daniel.lidstrom at sbg.se (=?iso-8859-1?Q?Daniel_Lidstr=F6m?=)
Date: Tue, 26 Jan 2010 15:47:20 +0100
Subject: [C++-sig] Problem running the quickstart
Message-ID: <FE8B524F1BF0854187E434B85105AC693AB8EA@dorado.ad.sbg.se>

Hi,

I have troubles compiling and running the quickstart:

C:\Users\Daniel.SBG\Programming\boost_1_41_0\libs\python\example\quickstart>bjam toolset=msvc-9.0 --verbose-test test --debug
-configuration
notice: found boost-build.jam at C:\Users\Daniel.SBG\Programming\boost_1_41_0\libs\python\example\quickstart\boost-build.jam
notice: loading Boost.Build from C:\Users\Daniel.SBG\Programming\boost_1_41_0\libs\python\example\quickstart\..\..\..\..\tool
s\build\v2
notice: Searching C:\Windows U:\ C:\cygwin\home\daniel C:\Users\Daniel.SBG C:\Users\Daniel.SBG\Programming\boost_1_41_0\libs\
python\example\quickstart\../../../../tools/build/v2 C:/Users/Daniel.SBG/Programming/boost_1_41_0/tools/build/v2/kernel C:/Us
ers/Daniel.SBG/Programming/boost_1_41_0/tools/build/v2/util C:/Users/Daniel.SBG/Programming/boost_1_41_0/tools/build/v2/build
 C:/Users/Daniel.SBG/Programming/boost_1_41_0/tools/build/v2/tools C:/Users/Daniel.SBG/Programming/boost_1_41_0/tools/build/v
2/contrib C:/Users/Daniel.SBG/Programming/boost_1_41_0/tools/build/v2/. for site-config configuration file site-config.jam .
notice: Loading site-config configuration file site-config.jam from C:\Users\Daniel.SBG\Programming\boost_1_41_0\libs\python\
example\quickstart\..\..\..\..\tools\build\v2\site-config.jam .
notice: Searching U:\ C:\cygwin\home\daniel C:\Users\Daniel.SBG C:\Users\Daniel.SBG\Programming\boost_1_41_0\libs\python\exam
ple\quickstart\../../../../tools/build/v2 C:/Users/Daniel.SBG/Programming/boost_1_41_0/tools/build/v2/kernel C:/Users/Daniel.
SBG/Programming/boost_1_41_0/tools/build/v2/util C:/Users/Daniel.SBG/Programming/boost_1_41_0/tools/build/v2/build C:/Users/D
aniel.SBG/Programming/boost_1_41_0/tools/build/v2/tools C:/Users/Daniel.SBG/Programming/boost_1_41_0/tools/build/v2/contrib C
:/Users/Daniel.SBG/Programming/boost_1_41_0/tools/build/v2/. for user-config configuration file user-config.jam .
notice: Loading user-config configuration file user-config.jam from C:\Users\Daniel.SBG\Programming\boost_1_41_0\libs\python\
example\quickstart\..\..\..\..\tools\build\v2\user-config.jam .
notice: Searching ../../../.. for project-config configuration file project-config.jam .
notice: Loading project-config configuration file project-config.jam from ..\..\..\..\project-config.jam .
notice: [msvc-cfg] msvc-9.0 detected, command: 'c:\Program Files\Microsoft Visual Studio 9.0\VC\bin\cl.exe'
notice: [msvc-cfg] msvc-8.0 detected, command: 'C:\Program Files\Microsoft Visual Studio 8\VC\bin\cl.exe'
notice: will use 'c:\Program Files\Microsoft Visual Studio 9.0\VC\bin\cl.exe' for msvc, condition <toolset>msvc-9.0
notice: [msvc-cfg] condition: '<toolset>msvc-9.0/<architecture>/<address-model>', setup: 'call "c:\Program Files\Microsoft Vi
sual Studio 9.0\VC\vcvarsall.bat" x86 >nul
'
notice: [msvc-cfg] condition: '<toolset>msvc-9.0/<architecture>/<address-model>32', setup: 'call "c:\Program Files\Microsoft
Visual Studio 9.0\VC\vcvarsall.bat" x86 >nul
'
notice: [msvc-cfg] condition: '<toolset>msvc-9.0/<architecture>x86/<address-model>', setup: 'call "c:\Program Files\Microsoft
 Visual Studio 9.0\VC\vcvarsall.bat" x86 >nul
'
notice: [msvc-cfg] condition: '<toolset>msvc-9.0/<architecture>x86/<address-model>32', setup: 'call "c:\Program Files\Microso
ft Visual Studio 9.0\VC\vcvarsall.bat" x86 >nul
'
notice: [msvc-cfg] condition: '<toolset>msvc-9.0/<architecture>/<address-model>64', setup: 'call "c:\Program Files\Microsoft
Visual Studio 9.0\VC\vcvarsall.bat" x86_amd64 >nul
'
notice: [msvc-cfg] condition: '<toolset>msvc-9.0/<architecture>x86/<address-model>64', setup: 'call "c:\Program Files\Microso
ft Visual Studio 9.0\VC\vcvarsall.bat" x86_amd64 >nul
'
notice: [msvc-cfg] condition: '<toolset>msvc-9.0/<architecture>ia64/<address-model>', setup: 'call "c:\Program Files\Microsof
t Visual Studio 9.0\VC\vcvarsall.bat" x86_ia64 >nul
'
notice: [msvc-cfg] condition: '<toolset>msvc-9.0/<architecture>ia64/<address-model>64', setup: 'call "c:\Program Files\Micros
oft Visual Studio 9.0\VC\vcvarsall.bat" x86_ia64 >nul
'
notice: [cmdline-cfg] Detected command-line request for msvc-9.0: toolset= msvc version= 9.0
notice: [cmdline-cfg] toolset msvc-9.0 already configured
notice: iostreams: not using zlib compression
notice: iostreams: not using zlib compression
notice: iostreams: not using bzip2 compression
notice: iostreams: not using bzip2 compression
notice: [python-cfg] Configuring python...
notice: [python-cfg] Registry indicates Python 3.0 installed at "C:\Python30\"
notice: [python-cfg] Checking interpreter command "python"...
notice: [python-cfg] running command 'DIR /-C /A:S "C:\Python30\python.exe" 2>&1'
notice: [python-cfg] running command 'python -c "from sys import *; print('version=%d.%d\nplatform=%s\nprefix=%s\nexec_prefix
=%s\nexecutable=%s' % (version_info[0],version_info[1],platform,prefix,exec_prefix,executable))" 2>&1'
notice: [python-cfg] ...requested configuration matched!
notice: [python-cfg] Details of this Python configuration:
notice: [python-cfg]   interpreter command: "python"
notice: [python-cfg]   include path: "C:\Python30\Include"
notice: [python-cfg]   library path: "C:\Python30\libs"
notice: [python-cfg]   DLL search path: "C:\Python30"
error: Unable to find file or target named
error:     'test_extending.py'
error: referred from project at
error:     '.'

There does not seem to be a script by the name test_extending.py. Did I miss something?
Thanks in advance!

Regards,

Daniel Lidstr?m
Stockholm, Sweden

From yunlin07 at gmail.com  Tue Jan 26 19:29:07 2010
From: yunlin07 at gmail.com (lin yun)
Date: Tue, 26 Jan 2010 10:29:07 -0800
Subject: [C++-sig] Please help: boost python 1.41 extension problem
In-Reply-To: <1264453573.27378.20.camel@mu>
References: <e3c115371001221729j200cc479yf75a644aede2dbcd@mail.gmail.com>
	<1264450805.26787.17.camel@mu>
	<e3c115371001251238u75ba3b48wf98669ed3655ccde@mail.gmail.com>
	<1264453573.27378.20.camel@mu>
Message-ID: <e3c115371001261029g3d28c8a9hff3a9d590be471d5@mail.gmail.com>

Wow, it worked like a magic!!!  Thank you so much, Jim!

On Mon, Jan 25, 2010 at 1:06 PM, Jim Bosch <talljimbo at gmail.com> wrote:

> On Mon, 2010-01-25 at 12:38 -0800, lin yun wrote:
>
> > I did not mention that I had no problem with Boost 1.34, on the same
> > system. Also I ever tried Boost 1.34 and 1.37 on Fedora 7, I got
> > exactly the same result: Boost 1.34 worked fine while 1.37 failed on
> > extension test. The test I ran was the example test from boost
> > package: libs/python/example/quickstart/, which was recommended in
> > the boost python installation instruction. There was no any personal
> > code involved.
> >
> > I wonder if you, or anyone else had related experience ?
> >
>
> I have not, but I haven't tried boost > 1.34.1 on any Red Hat distros.
>
> I can confirm that this example compiles successfully with Boost 1.40
> and the latest trunk on Ubuntu 9.10 (gcc 4.4.1).
>
> You may have noticed the line gcc aborts on:
>
> >         >
> >         > ../../../../boost/type_traits/detail/cv_traits_impl.hpp:37:
> >         internal
> >         > compiler error: in make_rtl_for_nonlocal_decl, at
> >         cp/decl.c:5067
> >         >
>
> has not changed between Boost 1.34 and 1.35, but it refers to the macro
> BOOST_STATIC_CONSTANT, which is defined in boost/config/suffix.hpp, and
> reflects whether a compiler supports in-class initialization of static
> integral constant members.  Perhaps your compiler does not fully support
> this feature, but boost thinks it does?
>
> Try adding the line
>
> #define BOOST_NO_INCLASS_MEMBER_INITIALIZATION
>
> to boost/config/user.hpp.  Then recompile boost.
>
>
>
> Jim Bosch
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100126/c4422f68/attachment.htm>

From talljimbo at gmail.com  Tue Jan 26 20:44:37 2010
From: talljimbo at gmail.com (Jim Bosch)
Date: Tue, 26 Jan 2010 11:44:37 -0800
Subject: [C++-sig] Problem running the quickstart
In-Reply-To: <FE8B524F1BF0854187E434B85105AC693AB8EA@dorado.ad.sbg.se>
References: <FE8B524F1BF0854187E434B85105AC693AB8EA@dorado.ad.sbg.se>
Message-ID: <1264535077.2513.2.camel@bishop>

On Tue, 2010-01-26 at 15:47 +0100, Daniel Lidstr?m wrote:
> error: Unable to find file or target named
> error:     'test_extending.py'
> error: referred from project at
> error:     '.'
> 
> There does not seem to be a script by the name test_extending.py. Did I miss something?
> Thanks in advance!
> 

That file is in boost/libs/python/quickstart/.

I don't know much about running Python in Windows, but it looks like it
was just looking in the default system Python locations.  You may have
to update your PYTHONPATH.

Jim Bosch


From talljimbo at gmail.com  Tue Jan 26 20:56:30 2010
From: talljimbo at gmail.com (Jim Bosch)
Date: Tue, 26 Jan 2010 11:56:30 -0800
Subject: [C++-sig] wrapping vector of pointers
In-Reply-To: <92560a4e1001260436p3e547f38i3c749aa2ad09c114@mail.gmail.com>
References: <92560a4e1001260436p3e547f38i3c749aa2ad09c114@mail.gmail.com>
Message-ID: <1264535790.2513.14.camel@bishop>

On Tue, 2010-01-26 at 13:36 +0100, martin manduch wrote:
> 
> I want to make XVec responsible for memory, which is addressed through
> pointers, but I don't know how.
> 

That would be difficult, especially because you can't really subclass
std::vector in a safe way.

> 
> I need to wrap vector of pointers, because it's parameter in one function...
> 

If it's only used in one function, I'd recommend writing a custom
wrapper for that function that does a manual conversion from a Python
list.  Something like this:

void py_accept_vector(boost::python::list const & sequence) {
    int size = boost::python::len(sequence);
    std::vector<X*> vector(size);
    for (int n = 0; n < size; ++n) {
        boost::python::extract<X*> converter(sequence[n]);
        if (!converter) {
            PyErr_SetString(
                PyExc_TypeError, 
                "Expected sequence of 'X' objects."
            );
	    boost::python::throw_error_already_set();
        }
        vector[n] = converter();
    }
    accept_vector(vector); // The passed vector's items are managed
                           // by Python, and cannot be relied upon
                           // to be valid after the function exits.
}

You can then wrap "py_accept_vector" in the usual way:

boost::python::def("accept_vector", py_accept_vector);

Hope that helps!


Jim Bosch


From daniel.lidstrom at sbg.se  Wed Jan 27 11:37:51 2010
From: daniel.lidstrom at sbg.se (=?iso-8859-1?Q?Daniel_Lidstr=F6m?=)
Date: Wed, 27 Jan 2010 11:37:51 +0100
Subject: [C++-sig] Problem running the quickstart
In-Reply-To: <1264535077.2513.2.camel@bishop>
References: <FE8B524F1BF0854187E434B85105AC693AB8EA@dorado.ad.sbg.se>
	<1264535077.2513.2.camel@bishop>
Message-ID: <FE8B524F1BF0854187E434B85105AC693AB929@dorado.ad.sbg.se>

> From: Jim Bosch [mailto:talljimbo at gmail.com] 
> 
> On Tue, 2010-01-26 at 15:47 +0100, Daniel Lidstr?m wrote:
> > error: Unable to find file or target named
> > error:     'test_extending.py'
> > error: referred from project at
> > error:     '.'
> > 
> > There does not seem to be a script by the name 
> test_extending.py. Did I miss something?
> > Thanks in advance!
> > 
> 
> That file is in boost/libs/python/quickstart/.
> 
> I don't know much about running Python in Windows, but it 
> looks like it
> was just looking in the default system Python locations.  You may have
> to update your PYTHONPATH.

I don't have boost/libs/python/quickstart:

C:\Users\Daniel.SBG\Programming\boost_1_41_0\libs\python>dir /B
build
doc
example
index.html
pyste
release_notes.txt
src
test
todo.html
todo.txt

Neither the example nor the quickstart folders contain test_extending.py:

C:\Users\Daniel.SBG\Programming\boost_1_41_0\libs\python\example>dir /B
boost-build.jam
getting_started1.cpp
getting_started2.cpp
Jamroot
project.zip
quickstart
README
std_pair.cpp
test_getting_started1.py
test_getting_started2.py
test_std_pair.py
tutorial

C:\Users\Daniel.SBG\Programming\boost_1_41_0\libs\python\example>cd quickstart

C:\Users\Daniel.SBG\Programming\boost_1_41_0\libs\python\example\quickstart>dir /B
bin
boost-build.jam
embedding.cpp
extending.cpp
Jamroot
script.py

Am I looking in the wrong place?

/Daniel

From talljimbo at gmail.com  Wed Jan 27 12:05:33 2010
From: talljimbo at gmail.com (Jim Bosch)
Date: Wed, 27 Jan 2010 03:05:33 -0800
Subject: [C++-sig] Problem running the quickstart
In-Reply-To: <FE8B524F1BF0854187E434B85105AC693AB929@dorado.ad.sbg.se>
References: <FE8B524F1BF0854187E434B85105AC693AB8EA@dorado.ad.sbg.se>
	<1264535077.2513.2.camel@bishop>
	<FE8B524F1BF0854187E434B85105AC693AB929@dorado.ad.sbg.se>
Message-ID: <1264590333.4375.5.camel@bishop>

On Wed, 2010-01-27 at 11:37 +0100, Daniel Lidstr?m wrote:
> Neither the example nor the quickstart folders contain test_extending.py:
...
> 
> Am I looking in the wrong place?
> 

Nope, that's right.  It should be libs/python/examples/quickstart (I
forgot the "examples" earlier).

I don't know why you didn't get that file, but here's a link to a
single-file download for it:

https://svn.boost.org/trac/boost/export/59273/tags/release/Boost_1_41_0/libs/python/example/quickstart/test_extending.py

Comparing your directory contents to mine, that seems to be all that's
missing.

Jim



From daniel.lidstrom at sbg.se  Wed Jan 27 16:27:02 2010
From: daniel.lidstrom at sbg.se (=?iso-8859-1?Q?Daniel_Lidstr=F6m?=)
Date: Wed, 27 Jan 2010 16:27:02 +0100
Subject: [C++-sig] Problem running the quickstart
In-Reply-To: <1264590333.4375.5.camel@bishop>
References: <FE8B524F1BF0854187E434B85105AC693AB8EA@dorado.ad.sbg.se>
	<1264535077.2513.2.camel@bishop>
	<FE8B524F1BF0854187E434B85105AC693AB929@dorado.ad.sbg.se>
	<1264590333.4375.5.camel@bishop>
Message-ID: <FE8B524F1BF0854187E434B85105AC693AB98B@dorado.ad.sbg.se>

> > 
> > Am I looking in the wrong place?
> > 
> 
> Nope, that's right.  It should be libs/python/examples/quickstart (I
> forgot the "examples" earlier).
> 
> I don't know why you didn't get that file, but here's a link to a
> single-file download for it:
> 
> https://svn.boost.org/trac/boost/export/59273/tags/release/Boo
> st_1_41_0/libs/python/example/quickstart/test_extending.py
> 
> Comparing your directory contents to mine, that seems to be all that's
> missing.

Thank you Jim, now I got it to compile!

/Daniel

From dave at boostpro.com  Wed Jan 27 17:34:12 2010
From: dave at boostpro.com (David Abrahams)
Date: Wed, 27 Jan 2010 11:34:12 -0500
Subject: [C++-sig] boost python
	calling	conventions	support	patch	submission (__stdcall,
	__cdecl, __fastcall)
In-Reply-To: <4B6020B3.6040401@mgdesign.fr>
References: <4B437643.5000103@mgdesign.org>
	<201001101118.25701.lists_ravi@lavabit.com>
	<4B50A150.1010908@mgdesign.org> <m2d410ckx4.wl%dave@boostpro.com>
	<4B5D689A.40902@mgdesign.org> <m27hr6i4p4.wl%dave@boostpro.com>
	<4B6020B3.6040401@mgdesign.fr>
Message-ID: <m2hbq7se0r.wl%dave@boostpro.com>

At Wed, 27 Jan 2010 12:17:07 +0100,
Nicolas Lelong wrote:
> 
> My thought was to leave the library files as-is, and to simply enable 
> the unit tests on 'toolsets' related to Microsoft compilers, and perhaps 
> add later other compilers that might support these calling conventions, 
> based on users needs.
> 
> This way, each user would be free to configure its boost.python
> without hassle, and tests would not break on 'uncertified' toolsets.

I'm not sure exactly what "enable the unit tests" means in this case.
Can you be explicit and/or preferably provide an illustrative patch?

-- 
Dave Abrahams           Meet me at BoostCon: http://www.boostcon.com
BoostPro Computing
http://www.boostpro.com


From nico_ml at mgdesign.org  Thu Jan 28 10:31:39 2010
From: nico_ml at mgdesign.org (Nicolas Lelong)
Date: Thu, 28 Jan 2010 10:31:39 +0100
Subject: [C++-sig] boost python
 calling	conventions	support	patch	submission (__stdcall, __cdecl,
 __fastcall)
In-Reply-To: <m2hbq7se0r.wl%dave@boostpro.com>
References: <4B437643.5000103@mgdesign.org>	<201001101118.25701.lists_ravi@lavabit.com>	<4B50A150.1010908@mgdesign.org>	<m2d410ckx4.wl%dave@boostpro.com>	<4B5D689A.40902@mgdesign.org>	<m27hr6i4p4.wl%dave@boostpro.com>	<4B6020B3.6040401@mgdesign.fr>
	<m2hbq7se0r.wl%dave@boostpro.com>
Message-ID: <4B61597B.7010204@mgdesign.org>

Le 27/01/2010 17:34, David Abrahams a ?crit :

> I'm not sure exactly what "enable the unit tests" means in this case.
> Can you be explicit and/or preferably provide an illustrative patch?
>    
As I'm a complete bjam noob, I made an attempt that is certainly clumsy 
and/or inelegant.

The idea would be to conditionally compile/run the test, the condition 
being the current toolset is in a given set of toolsets (currently only 
'msvc' toolset).

In the attached patch, I conditionnally define 
'BOOST_PYTHON_TEST_ENABLE' on 'msvc' toolset, and I modified the tests 
to use calling conventions only if 'BOOST_PYTHON_TEST_ENABLE' if defined.

I suppose that there's a way to instruct bjam to conditionally 
compile/run a test, that would be more clear, and would prevent to 
change the tests.

What do you think ?

Cheers,
Nicolas.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: boost.python.calling_conventions_test_compiler_selection_hack.patch
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100128/3239058f/attachment.txt>

From Matthew.Scouten at tradingtechnologies.com  Fri Jan 29 22:53:39 2010
From: Matthew.Scouten at tradingtechnologies.com (Matthew Scouten (TT))
Date: Fri, 29 Jan 2010 15:53:39 -0600
Subject: [C++-sig] FW: Boost.Python.function bug
Message-ID: <32490DFF7774554A85D65D23A9F0F9380C4DE822@chiex01>

Hey folks, 

One of my users spotted this, and I think it is an error in BP. It works
the same for any function exported. He states it more completely then I
could, So I will just forward what he sent to me:  

 

From: Ben Walker (TT) 
Sent: Friday, January 29, 2010 3:27 PM
To: Matthew Scouten (TT)
Subject: Boost.Python.function bug

 

>>> boost_func

<Boost.Python.function object at 0x011A0228>

>>> type(boost_func)

<type 'Boost.Python.function'>

>>> import types           

>>> isinstance(boost_func, types.BuiltinFunctionType)

True

>>> dir(boost_func)

['__call__', '__class__', '__cmp__', '__delattr__', '__doc__', '__eq__',
'__format__', '__ge__', '__getattribute__', '__gt__', '__hash__',
'__init__', '__le__', '__lt__', '__module__', '__name__', '__ne__',
'__new__', '__reduce__', '__reduce_ex__', '__repr__', '__self__',
'__setattr__', '__sizeof__', '__str__', '__subclasshook__']

>>> boost_func.__self__

Traceback (most recent call last):

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

AttributeError: 'Boost.Python.function' object has no attribute
'__self__'

>>> hasattr(boost_func, '__self__')

False

>>> 

 

dir() says Boost.Python.function objects have a __self__ member, but
accessing it throws an AttributeError. Furthermore, real builtin
functions have a __self__ member and some python libraries assume that
if something is an instance of a builtin function, it will have a
__self__ member. Testing via hasattr does function appropriately for the
actual behavior.

 

This was discovered while trying to use Sphinx's autodoc extension
(sphinx.pocoo.org) on a C++ library wrapped via Boost.Python.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100129/bc6c7513/attachment.htm>

From chris.lasher at gmail.com  Sun Jan 31 17:23:48 2010
From: chris.lasher at gmail.com (Chris Lasher)
Date: Sun, 31 Jan 2010 11:23:48 -0500
Subject: [C++-sig] Using code generated by Py++ as a Python extension
Message-ID: <128a885f1001310823h7d43ac4dlde26c354e744160f@mail.gmail.com>

I asked this question on Stack Overflow, but realized that perhaps few from
the Py++ community are there yet. I'm re-posting the question here; I hope
you don't mind.

http://stackoverflow.com/questions/2169948/using-code-generated-by-py-as-a-python-extension


I have a need to wrap an existing C++ library for use in Python. After
reading through this answer
http://stackoverflow.com/questions/456884/extending-python-to-swig-or-not-to-swig/456949#456949on
choosing an appropriate method to wrap C++ for use in Python, I
decided
to go with Py++.

I walked through the tutorial for Py++
http://www.language-binding.net/pyplusplus/documentation/tutorials/module_builder/module_builder.html,
using the tutorial files, and I got the expected output in
generated.cpp,
but I haven't figured out what to do in order to actually use the generated
code as an extension I can import in Python. I'm sure I have to compile the
code, now, but with what? Am I supposed to use bjam?

Thanks,
Chris
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20100131/380eb850/attachment.htm>

From roman.yakovenko at gmail.com  Sun Jan 31 19:19:56 2010
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Sun, 31 Jan 2010 20:19:56 +0200
Subject: [C++-sig] Using code generated by Py++ as a Python extension
In-Reply-To: <128a885f1001310823h7d43ac4dlde26c354e744160f@mail.gmail.com>
References: <128a885f1001310823h7d43ac4dlde26c354e744160f@mail.gmail.com>
Message-ID: <7465b6171001311019i6a0af051h5f23470cd5ebe124@mail.gmail.com>

On Sun, Jan 31, 2010 at 6:23 PM, Chris Lasher <chris.lasher at gmail.com> wrote:
> I asked this question on Stack Overflow, but realized that perhaps few from
> the Py++ community are there yet. I'm re-posting the question here; I hope
> you don't mind.
>
> http://stackoverflow.com/questions/2169948/using-code-generated-by-py-as-a-python-extension

I will answer here, would you mind to upload the answer to
stackoverflow? Thanks.

>
> I have a need to wrap an existing C++ library for use in Python.
> reading through this answer
> http://stackoverflow.com/questions/456884/extending-python-to-swig-or-not-to-swig/456949#456949
> on choosing an appropriate method to wrap C++ for use in Python, I decided
> to go with Py++.
>
> I walked through the tutorial for Py++
> http://www.language-binding.net/pyplusplus/documentation/tutorials/module_builder/module_builder.html
> , using the tutorial files, and I got the expected output in generated.cpp,
> but I haven't figured out what to do in order to actually use the generated
> code as an extension I can import in Python. I'm sure I have to compile the
> code, now, but with what? Am I supposed to use bjam?

After you generated the code you have to compile it and for this
purpose you can use your favorite build system. I use bjam only to
compile boost. After this, I prefer to use scons ( on Windows and on
Linux ).

The following is an example of sconstruct file, which is used to
compile one of the Py++ unittests:
( this is a generated code too :-) )

import sys
env = Environment()
if 'linux' not in sys.platform:
    env['MSVS'] = {'VERSION': ''}
    env['MSVS_VERSION'] = ''
    Tool('msvc')(env)
t = env.SharedLibrary( target=r'abstract_classes'
    , source=[ r'/home/roman/language-binding/sources/pyplusplus_dev/unittests/temp/abstract_classes.cpp'
]
    , LIBS=[ r"boost_python" ]
    , LIBPATH=[ r"",r"/home/roman/include/libs" ]
    , CPPPATH=[
r"/home/roman/boost_svn",r"/usr/include/python2.6",r"/home/roman/language-binding/sources/pyplusplus_dev/unittests/temp",r"/home/roman/language-binding/sources/pyplusplus_dev/unittests/data",r"/home/roman/boost_svn"
]
    , CCFLAGS=[  ]
    , SHLIBPREFIX=''
    , SHLIBSUFFIX='.so'
)

Since your code generator written in Python, you can continue where
Py++ stops and generate your favorite "make" file. You can go even
father. Py++ tests generate the code, compile, load new module and
test the functionality. All this is done in a single, stand alone
process.

HTH

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