From jbrandmeyer at earthlink.net  Mon Nov  1 04:18:35 2004
From: jbrandmeyer at earthlink.net (Jonathan Brandmeyer)
Date: Sun, 31 Oct 2004 22:18:35 -0500
Subject: [C++-sig] MEMBER_FUNCTION_OVERLOADS doesn't honor virtualness
Message-ID: <1099279115.7972.8.camel@illuvatar>


The subject says it all.  Attached is a test case.

Thanks,
-Jonathan Brandmeyer


-------------- next part --------------
A non-text attachment was scrubbed...
Name: idontwork.cpp
Type: text/x-c++src
Size: 583 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041031/6129b758/attachment.cpp>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: idontwork.py
Type: application/x-python
Size: 90 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041031/6129b758/attachment.bin>

From dave at boost-consulting.com  Mon Nov  1 05:10:32 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Sun, 31 Oct 2004 23:10:32 -0500
Subject: [C++-sig] Re: MEMBER_FUNCTION_OVERLOADS doesn't honor virtualness
References: <1099279115.7972.8.camel@illuvatar>
Message-ID: <u1xfeky4n.fsf@boost-consulting.com>

Jonathan Brandmeyer <jbrandmeyer at earthlink.net> writes:

> The subject says it all.  Attached is a test case.
>
> Thanks,
> -Jonathan Brandmeyer
>
>
>
> #include <boost/python.hpp>
> #include <iostream>
>
> class base
> {
>  public:
>     virtual void print_me(unsigned int i = 0)
>     {  std::cout << "base: " << i << "\n"; }
> };
>
> class derived : public base
> {
>  public:
>     void print_me( unsigned int i)
>     {  std::cout << "derived: " << i << "\n"; }
> };
>
> BOOST_PYTHON_MEMBER_FUNCTION_OVERLOADS( print_me, base::print_me, 0, 1)
This won't compile without an ampersand-------------^

> BOOST_PYTHON_MODULE( idontwork)
> {
>     using namespace boost::python;
>     class_<base>("base")
>         .def( "print_me", &base::print_me, print_me())
>         ;
>
>     class_<derived, bases<base> >("derived")
>         ;
> }
>
>
> from idontwork import *
> x = base()
> y = derived()
>
> x.print_me()
> y.print_me()
> y.print_me(1)


What result are you expecting from this?  Can you rewrite the example
so that it asserts when you don't get the expected result.

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From jbrandmeyer at earthlink.net  Mon Nov  1 12:07:30 2004
From: jbrandmeyer at earthlink.net (Jonathan Brandmeyer)
Date: Mon, 01 Nov 2004 06:07:30 -0500
Subject: [C++-sig] Re: MEMBER_FUNCTION_OVERLOADS doesn't honor virtualness
In-Reply-To: <u1xfeky4n.fsf@boost-consulting.com>
References: <1099279115.7972.8.camel@illuvatar>
	<u1xfeky4n.fsf@boost-consulting.com>
Message-ID: <1099307250.9270.11.camel@illuvatar>

On Sun, 2004-10-31 at 23:10 -0500, David Abrahams wrote:
> Jonathan Brandmeyer <jbrandmeyer at earthlink.net> writes:
> 
> > The subject says it all.  Attached is a test case.
> >

> > BOOST_PYTHON_MEMBER_FUNCTION_OVERLOADS( print_me, base::print_me, 0, 1)
> This won't compile without an ampersand-------------^

Are you sure?  For me (GCC 3.3.4), it doesn't compile with the
address-of operator.  Tests in libs/python/test/keywords.cpp and
libs/python/test/args.cpp do not use the ampersand.

> What result are you expecting from this?  Can you rewrite the example
> so that it asserts when you don't get the expected result.

Sure.  I have attached a replacement for the Python file.  The problem
is that derived.print_me() prints "base: " some_number, when I expect to
get "derived: " some_number.

-Jonathan
-------------- next part --------------
A non-text attachment was scrubbed...
Name: idontwork.py
Type: application/x-python
Size: 433 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041101/4d77c9f1/attachment.bin>

From dave at boost-consulting.com  Mon Nov  1 15:10:15 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Mon, 01 Nov 2004 09:10:15 -0500
Subject: [C++-sig] Re: MEMBER_FUNCTION_OVERLOADS doesn't honor virtualness
References: <1099279115.7972.8.camel@illuvatar>
	<u1xfeky4n.fsf@boost-consulting.com>
	<1099307250.9270.11.camel@illuvatar>
Message-ID: <uk6t5k6d4.fsf@boost-consulting.com>

Jonathan Brandmeyer <jbrandmeyer at earthlink.net> writes:

> On Sun, 2004-10-31 at 23:10 -0500, David Abrahams wrote:
>> Jonathan Brandmeyer <jbrandmeyer at earthlink.net> writes:
>> 
>> > The subject says it all.  Attached is a test case.
>> >
>
>> > BOOST_PYTHON_MEMBER_FUNCTION_OVERLOADS( print_me, base::print_me, 0, 1)
>> This won't compile without an ampersand-------------^
>
> Are you sure?  For me (GCC 3.3.4), it doesn't compile with the
> address-of operator.  Tests in libs/python/test/keywords.cpp and
> libs/python/test/args.cpp do not use the ampersand.
>
>> What result are you expecting from this?  Can you rewrite the example
>> so that it asserts when you don't get the expected result.
>
> Sure.  I have attached a replacement for the Python file.  

That test doesn't work; you can't expect C++ stream I/O to be seen by
Python.  Try the enclosed.

> The problem
> is that derived.print_me() prints "base: " some_number, when I expect to
> get "derived: " some_number.

Just drop the base:: qualification from
BOOST_PYTHON_MEMBER_FUNCTION_OVERLOADS and you'll be fine.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: theTest.cpp
Type: text/x-cplusplus
Size: 628 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041101/f8825dcd/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: theTest.py
Type: application/octet-stream
Size: 447 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041101/f8825dcd/attachment.obj>
-------------- next part --------------

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com

From ldfwyn at 163.com  Mon Nov  1 16:41:00 2004
From: ldfwyn at 163.com (Li Dongfeng)
Date: Mon, 1 Nov 2004 10:41:0 -0500
Subject: [C++-sig] Re: regarding your post entitled " Numeric--factory
	function" onBoost.Python
Message-ID: <20041101154122.E18EC1E4002@bag.python.org>

Hi Faheem,

>   boost::python::numeric::array arr(dim1,dim2);
--------------------------------------^^^^^^^^^
The problem with you code is the array constructor function usage.
That function cannot be called with two integers. See the numarray
(or Numeric) docs for correct usage.

   I tested some solution but cannot reach an elegant one.
An elegant solution should use the numarray(or Numeric) API
directly to access the array contents in C++.
But I have only got a make-shift solution: use the boost::python
indexing suite. After you have used the indexing suite to map C++ vector
to Python sequence type, you can use a C++ vector as a python sequence
in C++ to initialize a numarray.array.

  To map a C++ std::vector<double> to a Python sequence type named
``DoubleVec'', just make this boost.python extension:

------------------------
#include <boost/python/suite/indexing/vector_indexing_suite.hpp>
#include <boost/python/implicit.hpp>
#include <boost/python/numeric.hpp>
#include <boost/python/tuple.hpp>
#include <boost/python/module.hpp>
#include <boost/python/def.hpp>
#include <vector>
BOOST_PYTHON_MODULE(_numpyext)
{
  class_<std::vector<double> >("DoubleVec")
    .def(vector_indexing_suite<std::vector<double> >());
}
---------------
	
Then in a C++ extension module, use:

---------------
// std::vector conversion to numarray.array
numeric::array vec2array_double(std::vector<double> x){
  return numeric::array(x, "Float64");
}
BOOST_PYTHON_MODULE(_numpyext)
{
  def("_cppvec2array", &vec2array_double);
}
---------------

this really works, the ``x'' in the numeric::array function call
is recognized by the numarray module constructor function ``array''
as a python sequence of type ``DoubleVec''. 
This function is meant to return a array
in python, so in python we can wrap this function up:

-------------------
from _numpyext import _cppvec2array
def cpp2nar(x, shape=None):
    """Convert a C++ <std>vector to a numarray.array.

    If shp is provided then reshape."""
    xar = _cppvec2array(x)
    if shape is not None:
        xar.setshape(shape)
    return xar
def nar2cpp(x, totype=None):
    """Convert a numarray.array to a 1D C++ std::vector.

    xar should be a double type numarray.array
    """
    xv = DoubleVec()
    xv[:] = x.flat
    xv.shape = x.shape
    ## C++ vector may not have a shape attribute,
    ## but we attach it in python.
    ## In C++ although we only use the data as one-dim,
    ## but when converting back to python we can
    ## use the shape infomation.
    ## return value is a C++ vector
    return xv
--------------------------
We also give a conversion utility to convert numarray.array to C++ vector.

Below is a simple test with C++ extension and python wrapper:
----------------------------
// A test that vector indexing works.
std::vector<double> test_log(std::vector<double>& x){
  std::vector<double>::iterator it;
  for(it = x.begin(); it < x.end(); it++)
    *it = log(*it);

  return x;
}
BOOST_PYTHON_MODULE(_testnum)
{
  def("test_log", test_log);
}
---------------------

--------------------
def test():
    x = array([1, 4, 10.5, 100.0], shape=(2,2), type="Float64")
    print x

    xx = nar2cpp(x, totype="Float64")  ## converted to C++ vector
    yy = test_log(xx)     ## done in C++ vector
    y = cpp2nar(yy, shape=x.shape)   ## back to numarray.array

    print y
    return
----------------------------

These code are not complete, they are simplified.
The includes and imports are omitted.  You should
modify as you need. I have not use the module
in production use. I can send you the comple code
files if you like.



Cheers,			
			 
????????Li Dongfeng
????????ldfwyn at 163.com
??????????2004-11-01
   

======= 2004-10-30 01:55:00 =======

>Dear Li Dongfeng,
>
>I saw your post on 
>http://mail.python.org/pipermail/c++-sig/2004-August/007857.html I'm 
>trying to do very similar things. Did you meet with any success with your 
>efforts? In particular, I've had difficulty in creating a numeric array in 
>C++ and passing it back to python. I get the impression that these factory 
>functions have something to do with it, but I am not clear. Are these 
>supposed to be used as constructors for arrays? Can you provide an 
>example?
>
>A function like the following compiles without problems, but bombs when I 
>try to use it. Can you help me?
>
>Thanks in advance.                                           Faheem Mitha.
>
>****************************************************************************
>// Convert (C++) numeric matrix to (Python) Numarray numeric array.
>boost::python::object cppmat2pynumarr(Array2D<double> A)
>{
>   int i, j;
>   int dim1 = A.dim1();
>   int dim2 = A.dim2();
>
>   boost::python::numeric::array arr(dim1,dim2);
>   for(i=0; i<dim1; i++)
>     for(j=0; j<dim2; j++)
>       arr[make_tuple(i,j)] = A[i][j];
>   return arr;
>}







From chris at niekel.net  Tue Nov  2 10:43:05 2004
From: chris at niekel.net (Chris Niekel)
Date: Tue, 2 Nov 2004 10:43:05 +0100
Subject: [C++-sig] got stuck starting with boost.python
Message-ID: <20041102094305.GP1252@kira.niekel.net>

Hi,

I just started with boost.python, and I'm already stuck.
The C++ classes provide a 'readonly' view of the objects, and the user of
these classes can't create the objects, they're created and destroyed by some
friend-class (say, Manager).

So, there's a class Entity 
{
protected:
    Entity();
    ~Entity();

    std::string getId();
    ...
};

class Manager
{
public:
    std::map<std::string, Entity*> getEntityMap();
    ...
}

First problem was that the Entity destructor had to be public. 
So I changed that in the headerfile 
Q1: is there a better way to fix that?

I've translated the getEntityMap into a getEntities, using a new function
that creates a boost::python::list and walks through the map, adding the
elements. That works now.

What I want to do next is:

class_<Entity, Entity*>("Entity", no_init)
    .def_readonly("id", &Entity::getId)
    ;

That gives me a weird errors:
/usr/people1/smf/include/boost-1_31/boost/python/data_members.hpp:285:
conversion
   from `boost::is_member_pointer<const
   std::string&(Entity::*)() 
      const>' to non-scalar type `boost::mpl::bool_<false>' requested

which I don't understand. So working around the problem, I
    .def("id", &Entity::getId)
That also gives an error:
boost/python/detail/invoke.hpp:89: no match 
   for call to `(const 
      boost::python::detail::specify_a_return_value_policy_to_wrap_functions_returning<const
         std::string&>) (const std::basic_string<char,
         std::char_traits<char>, 
            std::allocator<char> >&)'

Also beyond my comprehension. If I do:
    .def("id", &Entity::getId, return_internal_reference<>())
it compiles, but I get the runtime exception:
TypeError: No Python class registered for C++ class std::string.

So I've now created an extra method in the Entity: const char*
getIdCStr() which returns the id.c_str(). That sorta works.
I can do in python: for i in getEntities(): print i.id()

With .add_property("id", &Entity::getIdCStr) it works as I want it. 

Q2: This requires me to add methods to the Entity class (there are several
std::string attributes I need).  Is there a more elegant way to solve this?

Hope anyone can help me. I worked through the tutorial, but after jumping
to my own code, I noticed quite some bumps in my road.

Regards,
    Chris Niekel


-- 
    I've been down so long, if I'd cheer up, I'd still be depressed.
            - Lisa Simpson, Moanin' Lisa Blues.


From dave at boost-consulting.com  Tue Nov  2 16:18:31 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 02 Nov 2004 10:18:31 -0500
Subject: [C++-sig] Re: got stuck starting with boost.python
References: <20041102094305.GP1252@kira.niekel.net>
Message-ID: <u4qk8ffeg.fsf@boost-consulting.com>

Chris Niekel <chris at niekel.net> writes:

> Hi,
>
> I just started with boost.python, and I'm already stuck.
> The C++ classes provide a 'readonly' view of the objects, and the user of
> these classes can't create the objects, they're created and destroyed by some
> friend-class (say, Manager).
>
> So, there's a class Entity 
> {
> protected:
>     Entity();
>     ~Entity();
>
>     std::string getId();
>     ...
> };
>
> class Manager
> {
> public:
>     std::map<std::string, Entity*> getEntityMap();
>     ...
> }
>
> First problem was that the Entity destructor had to be public. 
> So I changed that in the headerfile 
> Q1: is there a better way to fix that?

Use boost::noncopyable when wrapping Entity.

> I've translated the getEntityMap into a getEntities, using a new function
> that creates a boost::python::list and walks through the map, adding the
> elements. That works now.
>
> What I want to do next is:
>
> class_<Entity, Entity*>("Entity", no_init)
>     .def_readonly("id", &Entity::getId)
>     ;
>
> That gives me a weird errors:
> /usr/people1/smf/include/boost-1_31/boost/python/data_members.hpp:285:
> conversion
>    from `boost::is_member_pointer<const
>    std::string&(Entity::*)() 
>       const>' to non-scalar type `boost::mpl::bool_<false>' requested

def_readonly only works on pointers-to-data-members. Try 
                                       ^^^^

    .add_property("id", &Entity::getId)

you may yet need to use a return value policy as in:

    .add_property(
        "id"
      , make_function(
             &Entity::getId
           , return_value_policy<copy_const_reference>()))

> which I don't understand. So working around the problem, I
>     .def("id", &Entity::getId)
> That also gives an error:
> boost/python/detail/invoke.hpp:89: no match 
>    for call to `(const 
>       boost::python::detail::specify_a_return_value_policy_to_wrap_functions_returning<const
>          std::string&>) (const std::basic_string<char,
>          std::char_traits<char>, 
>             std::allocator<char> >&)'
>
> Also beyond my comprehension. 

As the message says, you have to "specify a return value policy to wrap
functions returning const std::string&."  I suggest you use 

  return_value_policy<copy_const_reference>

> If I do:
>     .def("id", &Entity::getId, return_internal_reference<>())
> it compiles, but I get the runtime exception:
> TypeError: No Python class registered for C++ class std::string.

Right.  You're not wrapping std::string with a class_<...> (and you
don't want to) so Boost.Python has no way of creating a python object
that refers to the referenced std::string.

> So I've now created an extra method in the Entity: const char*
> getIdCStr() which returns the id.c_str(). That sorta works.
> I can do in python: for i in getEntities(): print i.id()
>
> With .add_property("id", &Entity::getIdCStr) it works as I want it. 

Why didn't you use add_property in the first place?  I ask because it
speaks to a deficiency in the documentation.

> Q2: This requires me to add methods to the Entity class (there are
> several std::string attributes I need).  Is there a more elegant way
> to solve this?

A free function taking an initial X& parameter always works just as
well as a member function of X.

HTH,

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From john at arbash-meinel.com  Tue Nov  2 23:52:09 2004
From: john at arbash-meinel.com (John A Meinel)
Date: Tue, 02 Nov 2004 16:52:09 -0600
Subject: [C++-sig] Custom Exception objects
Message-ID: <41880F99.20609@arbash-meinel.com>

I was wondering how to handle custom exception objects using boost::python.

I know about doing this:

static void translateConfigError(ConfigError const &e) {
	PyErr_SetString(PyExc_RuntimeError, e.what());
}
register_exception_translator<ConfigError>(&translateConfigError);

However, as far as I know, this only lets you translate a C++ exception 
class into a standard Python exception class.

Since I have already wrapped ConfigError as a boost::python class, is it 
possible to do something more like:

static void translateConfigError(ConfigError const &e) {
	PyErr_SetString(my_python_exception, e.what());
}

where my_python_exception was somehow related to ConfigError?

It is nice that I can at least generate KeyError, or ValueError, 
SyntaxError, etc, but it would be nice if I could generate ConfigError.

John
=:->
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 253 bytes
Desc: OpenPGP digital signature
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041102/da9db3a7/attachment.pgp>

From mrovner at propel.com  Wed Nov  3 00:16:35 2004
From: mrovner at propel.com (Mike Rovner)
Date: Tue, 02 Nov 2004 15:16:35 -0800
Subject: [C++-sig] Re: Custom Exception objects
In-Reply-To: <41880F99.20609@arbash-meinel.com>
References: <41880F99.20609@arbash-meinel.com>
Message-ID: <cm94es$lop$1@sea.gmane.org>

John A Meinel wrote:
> I was wondering how to handle custom exception objects using boost::python.

> Since I have already wrapped ConfigError as a boost::python class, is it 
> possible to do something more like:

suppose you have

object my_python_exception =
class_<ConfigError>("ConfigError", init<char*>)
    ...
    ;

> static void translateConfigError(ConfigError const &e) {
>     PyErr_SetString(my_python_exception, e.what());
> }

than above statement will be perfectly valid

HTH,
Mike



From blanding at localnet.com  Wed Nov  3 01:15:41 2004
From: blanding at localnet.com (Doug Blanding)
Date: Tue, 2 Nov 2004 19:15:41 -0500
Subject: [C++-sig] help running boost-python tutorials
Message-ID: <200411021915.41602.blanding@localnet.com>

I have downloaded boost_1_31_0 and have successfully gone through all of the 
steps of the installation, but when I try to run the tutorial examples, I get 
the following error when I try to import the extension module into python:

ImportError: libboost_python.so.1.31.0: cannot open shared object file: No 
such file or directory

I have Python2.2 on RH9 linux.
I installed bjam as an rpm.
I set the following environment variables:
        export PYTHON_ROOT=/usr
        export PYTHON_VERSION=2.2
Then I compiled boost using:
        bjam -sTOOLS=gcc stage
Next, I cd to file:  /dnld/boost_1_31_0/libs/python/example
and ran the command:
[doug at maven example]$ bjam -sTOOLS=gcc test
...patience...
...found 2025 targets...
...updating 40 targets...
MkDir1 bin/boost
... etc. ...
execute-test bin/example/test1.test/gcc/debug/test1.run
**passed** bin/example/test1.test/gcc/debug/test1.test
gcc-C++-action 
bin/example/getting_started2.so/gcc/debug/shared-linkable-true/getting_started2.o
gcc-Link-action 
bin/example/getting_started2.so/gcc/debug/shared-linkable-true/getting_started2.so
Chmod1 
bin/example/getting_started2.so/gcc/debug/shared-linkable-true/getting_started2.so
execute-test bin/example/test2.test/gcc/debug/test2.run
**passed** bin/example/test2.test/gcc/debug/test2.test
...updated 40 targets...

I then get the import error when I try to import getting_started1 or 
getting_started2.

ImportError: libboost_python.so.1.31.0: cannot open shared object file: No 
such file or directory

I have tried copying the files libboost_python.so.1.31.0 and 
libboost_python.so into various places so they can be found by python, but 
without success.

Can you tell me what I'm doing wrong?
Doug Blanding


From mrovner at propel.com  Wed Nov  3 01:38:23 2004
From: mrovner at propel.com (Mike Rovner)
Date: Tue, 02 Nov 2004 16:38:23 -0800
Subject: [C++-sig] Re: help running boost-python tutorials
In-Reply-To: <200411021915.41602.blanding@localnet.com>
References: <200411021915.41602.blanding@localnet.com>
Message-ID: <cm9988$vae$1@sea.gmane.org>

Doug Blanding wrote:
> ImportError: libboost_python.so.1.31.0: cannot open shared object file: No 
> such file or directory
> 
> I have tried copying the files libboost_python.so.1.31.0 and 
> libboost_python.so into various places so they can be found by python, but 
> without success.
> 
> Can you tell me what I'm doing wrong?
> Doug Blanding

did you try ldconfig after install/copy?

HTH,
Mike



From faheem at email.unc.edu  Wed Nov  3 02:29:19 2004
From: faheem at email.unc.edu (Faheem Mitha)
Date: Tue, 2 Nov 2004 20:29:19 -0500 (EST)
Subject: [C++-sig] Re: regarding your post entitled " Numeric--factory
 function" onBoost.Python
In-Reply-To: <20041101154122.E18EC1E4002@bag.python.org>
References: <20041101154122.E18EC1E4002@bag.python.org>
Message-ID: <Pine.LNX.4.61.0411021813240.20079@Chrestomanci>


Hi Li,

Thanks for your helpful reply. Sorry you had difficulties posting to my 
address. If you continue to have problems, please let me know and I will 
report it.

On Mon, 1 Nov 2004, Li Dongfeng wrote:

> Hi Faheem,
>
>>   boost::python::numeric::array arr(dim1,dim2);
> --------------------------------------^^^^^^^^^
> The problem with you code is the array constructor function usage.
> That function cannot be called with two integers. See the numarray
> (or Numeric) docs for correct usage.

Well, it was not intended to be working code, just an example of what I 
had in mind.

>   I tested some solution but cannot reach an elegant one.
> An elegant solution should use the numarray(or Numeric) API
> directly to access the array contents in C++.
> But I have only got a make-shift solution: use the boost::python
> indexing suite. After you have used the indexing suite to map C++ vector
> to Python sequence type, you can use a C++ vector as a python sequence
> in C++ to initialize a numarray.array.
>
>  To map a C++ std::vector<double> to a Python sequence type named
> ``DoubleVec'', just make this boost.python extension:
>
> ------------------------
> #include <boost/python/suite/indexing/vector_indexing_suite.hpp>
> #include <boost/python/implicit.hpp>
> #include <boost/python/numeric.hpp>
> #include <boost/python/tuple.hpp>
> #include <boost/python/module.hpp>
> #include <boost/python/def.hpp>
> #include <vector>
> BOOST_PYTHON_MODULE(_numpyext)
> {
>  class_<std::vector<double> >("DoubleVec")
>    .def(vector_indexing_suite<std::vector<double> >());
> }
> ---------------
>
> Then in a C++ extension module, use:
>
> ---------------
> // std::vector conversion to numarray.array
> numeric::array vec2array_double(std::vector<double> x){
>  return numeric::array(x, "Float64");
> }
> BOOST_PYTHON_MODULE(_numpyext)
> {
>  def("_cppvec2array", &vec2array_double);
> }
> ---------------
>
> this really works, the ``x'' in the numeric::array function call
> is recognized by the numarray module constructor function ``array''
> as a python sequence of type ``DoubleVec''.

I don't see why, if the second argument is supposed to be a list, why you 
can't just create a Boost::python::list in C++, and then pass it in as a 
argument. Isn't that equivalent? Eg.

  //Convert (C++) numeric matrix to (Python) Numarray numeric array.
  boost::python::object cppmat2pynumarr(Array2D<double> A)
  {
    int i, j;
    int dim1 = A.dim1();
    int dim2 = A.dim2();

    boost::python::list lst;

    for(i=0; i<dim1; i++)
      for(j=0; j<dim2; j++)
        lst.append(A[i][j]);

    return boost::python::numeric::array(lst,"Double", make_tuple(dim1,
  								dim2));

  }

This compiles without problems. However, this does not work. I get errors 
like this:

************************************************************************
Traceback (most recent call last):
   File "test.py", line 8, in ?
     concatfoobar = genepath.cpp_concatmat(foo,bar)
   File "/usr/lib/python2.3/site-packages/numarray/numarraycore.py", line 
321, in array
     type = getTypeObject(sequence, type, typecode)
   File "/usr/lib/python2.3/site-packages/numarray/numarraycore.py", line 
256, in getTypeObject
     rtype = _typeFromTypeAndTypecode(type, typecode)
   File "/usr/lib/python2.3/site-packages/numarray/numarraycore.py", line 
247, in _typeFromTypeAndTypecode
     return _nt.getType(typecode)
   File "/usr/lib/python2.3/site-packages/numarray/numerictypes.py", line 
462, in getType
     raise TypeError("Not a numeric type")
TypeError: Not a numeric type
************************************************************************

The method you suggested does work as you described it, though I am not 
clear why the above does not work, and what you suggest does. Ie

*************************************************************************
//Convert (C++) numeric matrix to (Python) Numarray numeric array.
boost::python::object cppmat2pynumarr(Array2D<double> A)
{
   int i, j;
   int dim1 = A.dim1();
   int dim2 = A.dim2();

   vector<double> vec;

   for(i=0; i<dim1; i++)
     for(j=0; j<dim2; j++)
       vec.push_back(A[i][j]);

   return  vec2array_double(vec);
}
*************************************************************************

where vec2array_double is as you defined it.

This gives

***********************************************************************
In [3]: foo
Out[3]:
array([[1, 2],
        [3, 4]])

In [4]: bar
Out[4]:
array([[0, 0],
        [0, 0]])

In [5]: concatfoobar
Out[5]: array([ 1.,  2.,  3.,  4.,  0.,  0.,  0.,  0.])
************************************************************************

However, this is not completely satisfactory. I would prefer to set the 
shape in the C++ code itself. Why is this not possible? Why is it 
necessary to do so in Python code? I don't want a Python wrapper for my 
C++ code.

I tried something like

******************************************************************
// std::vector conversion to numarray.array
boost::python::numeric::array vec2array_double(std::vector<double> x)
{
   boost::python::tuple foo = make_tuple(2,2);
   return boost::python::numeric::array(x, "Float64", foo);
}
*****************************************************************

This compiles without complaint, but does not reshape the array.

Furthermore, I am unclear on a fundamental point.

What *is* the constructor of the Numeric array in C++? From the 
documentation at http://www.boost.org/libs/python/doc/v2/numeric.html I 
thought that these were C++ functions called factory.

The documentation says

"These functions map to the underlying array type's array()  function 
family. They are not called "array" because of the C++ limitation that you 
can't define a member function with the same name as its enclosing class."

I'm not sure what this means. How are you supposed to call them? Does
numeric::array call these functions? I could not find any relevant 
examples. There are examples in the examples directory in numpy.cc, but
I am not sure what to make of

************************************************************************
object new_array()
{
     return numeric::array(
         make_tuple(
             make_tuple(1,2,3)
           , make_tuple(4,5,6)
           , make_tuple(7,8,9)
             )
         );
}
*************************************************************************

This works, but how do I generalise it to do what I want, and which 
constructor does this belong to?

and

*************************************************************************

void exercise_numarray(numeric::array& y, object check)
{
...
check(y.factory(make_tuple(1.2, 3.4), "Double", make_tuple(1,2,1)));
...
*************************************************************************

What is that y doing, if this is a constructor? Looks more like a member 
function. I've tried variations on this too, but get errors at runtime.

So you can see I am quite confused. Thanks again.

                                                                Faheem.


From faheem at email.unc.edu  Wed Nov  3 05:18:48 2004
From: faheem at email.unc.edu (Faheem Mitha)
Date: Tue, 2 Nov 2004 23:18:48 -0500 (EST)
Subject: [C++-sig] Re: regarding your post entitled " Numeric--factory
 function" onBoost.Python
In-Reply-To: <Pine.LNX.4.61.0411021813240.20079@Chrestomanci>
References: <20041101154122.E18EC1E4002@bag.python.org>
	<Pine.LNX.4.61.0411021813240.20079@Chrestomanci>
Message-ID: <Pine.LNX.4.61.0411022101390.20079@Chrestomanci>



On Tue, 2 Nov 2004, Faheem Mitha wrote:

>
> Hi Li,
>
> Thanks for your helpful reply.
[snip]

Hi Li,

I solved it! (I think). I was using a function that converted a matrix to 
a list of lists, where each list corresponded to a row of the matrix. Just 
looking at the example below (in my earlier mail)

************************************************************************
object new_array()
{
     return numeric::array(
         make_tuple(
             make_tuple(1,2,3)
           , make_tuple(4,5,6)
           , make_tuple(7,8,9)
             )
         );
}
*************************************************************************

made me realise that wrapping the final return result (list of lists) in a 
numeric::array might work, since apparently the constructor accepted this 
kind of syntax. Well, apparently it does, though I don't understand why. I 
don't see this constructor documented anywhere. Also, I don't understand 
why my previous attempts don't work.

See function below.

If you can enlighten me about any of this, please do so. Thanks.

                                                                Faheem.

*************************************************************************
// Convert (C++) numeric matrix to (Python) list.
boost::python::object cppmat2pynumarr(Array2D<double> A)
{
   int i, j;
   int dim1 = A.dim1();
   int dim2 = A.dim2();

   boost::python::list lst;
   boost::python::list rowlst;

   // Copy matrix to python list
   for(i=0; i<dim1; i++)
     {
       for(j=0; j<dim2; j++)
 	{
 	  double Aij = A[i][j];
 	  rowlst.append(Aij);
 	}
       lst.append(rowlst);
       rowlst = boost::python::list();
     }
   return boost::python::numeric::array(lst);
}


From mike.spam.filter at day8.com.au  Thu Nov  4 08:17:45 2004
From: mike.spam.filter at day8.com.au (Mike Thompson)
Date: Thu, 04 Nov 2004 18:17:45 +1100
Subject: [C++-sig] Call Policy Question
Message-ID: <cmcl36$ltv$1@sea.gmane.org>


I have a member function which returns an object and I'm trying to 
define a call policy for that member function which will:
    1.  cause the returned object to be managed:
            return_value_policy<manage_new_object>()
        its a new object and needs to be managed.

    2.   register that the returned object holds a
         reference to the originating object (but not visa-versa)

but I can't work out how to combine the two.

I need 1. above because the returned object is new and needs to be managed.

I need 2. above because this returned object has references BACK
to the object who's member function supplies it.

So the returned object can be destroyed without the original being 
effected, but destruction of the original MUST wait until the returned 
object is gone.


Some code which hopes to capture this ...

---------------------------------------------------------------------
class Orig;

class Fresh {
     public:
         Fresh(Orig* o) : o_(o) {}

         Orig* o_;
};


class Orig {
     public:
         Fresh* newFresh()  { return new Fresh(this); }
};

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

     class_<Orig>("Orig", no_init)
         .def("newFresh", &Orig::newFresh, XXXXX)
         ;

---------------------------------------------------------------------
What should XXXXX be??

Help greatly appreciated.  I can't work this out from the tutorial and 
test harness code.

--
Mike



From domma at procoders.net  Thu Nov  4 11:28:28 2004
From: domma at procoders.net (Achim Domma (Procoders))
Date: Thu, 04 Nov 2004 11:28:28 +0100
Subject: [C++-sig] python extension and bjam stage
Message-ID: <418A044C.2030909@procoders.net>

Hi,

I cannot figure out how to use the stage rule with boost.python. Could 
somebody give me an example?

regards,
Achim



From dave at boost-consulting.com  Thu Nov  4 15:01:17 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Thu, 04 Nov 2004 09:01:17 -0500
Subject: [C++-sig] Re: Call Policy Question
References: <cmcl36$ltv$1@sea.gmane.org>
Message-ID: <ufz3piuhe.fsf@boost-consulting.com>

Mike Thompson <mike.spam.filter at day8.com.au> writes:

> I have a member function which returns an object and I'm trying to
> define a call policy for that member function which will:
>     1.  cause the returned object to be managed:
>             return_value_policy<manage_new_object>()
>         its a new object and needs to be managed.
>
>     2.   register that the returned object holds a
>          reference to the originating object (but not visa-versa)
>
> but I can't work out how to combine the two.
>
> I need 1. above because the returned object is new and needs to be managed.
>
> I need 2. above because this returned object has references BACK
> to the object who's member function supplies it.
>
> So the returned object can be destroyed without the original being
> effected, but destruction of the original MUST wait until the returned
> object is gone.

  return_value_policy<
        manage_new_object
      , with_custodian_and_ward<0, 1>
  >

should work.

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From achim.domma at syynx.de  Thu Nov  4 16:15:34 2004
From: achim.domma at syynx.de (Achim Domma)
Date: Thu, 04 Nov 2004 16:15:34 +0100
Subject: [C++-sig] Pyste Error
Message-ID: <418A4796.70004@syynx.de>

Hi,

could somebody explain to me, what could cause the error below? I have 
no idea what's going wrong.

regards,
Achim


Traceback (most recent call last):
   File "/usr/bin/pyste.py", line 4, in ?
     pyste.main()
   File "/usr/lib/python2.3/site-packages/Pyste/pyste.py", line 418, in main
     status = Begin()
   File "/usr/lib/python2.3/site-packages/Pyste/pyste.py", line 257, in 
Begin
     return GenerateCode(parser, module, out, interfaces, multiple)
   File "/usr/lib/python2.3/site-packages/Pyste/pyste.py", line 363, in 
GenerateCode
     exports.extend(order[interface])
KeyError: 'pyste_files/DrawablePushClipPath.pyste'



From john at johnmeinel.com  Sat Nov  6 15:51:19 2004
From: john at johnmeinel.com (John A Meinel)
Date: Sat, 06 Nov 2004 08:51:19 -0600
Subject: [C++-sig] Re: Custom Exception objects
In-Reply-To: <cm94es$lop$1@sea.gmane.org>
References: <41880F99.20609@arbash-meinel.com> <cm94es$lop$1@sea.gmane.org>
Message-ID: <418CE4E7.3080804@johnmeinel.com>

Mike Rovner wrote:
> John A Meinel wrote:
> 
>> I was wondering how to handle custom exception objects using 
>> boost::python.
> 
> 
>> Since I have already wrapped ConfigError as a boost::python class, is 
>> it possible to do something more like:
> 
> 
> suppose you have
> 
> object my_python_exception =
> class_<ConfigError>("ConfigError", init<char*>)
>    ...
>    ;
> 
>> static void translateConfigError(ConfigError const &e) {
>>     PyErr_SetString(my_python_exception, e.what());
>> }
> 
> 
> than above statement will be perfectly valid
> 
> HTH,
> Mike
> 

So am I right in saying I need something like this:

static object my_custom_exc;

static void translateCustomError(CustomError const &e) {
     PyErr_SetString(my_custom_exc, e.what());
}

void exportCustom()
{
	my_custom_exc = class_<CustomError, bases<std::runtime_error>
			, >("CustomError")
		.def("what", &CustomError::what)
	;
         register_exception_translator<CustomError>(&translateCustomError);
}

Basically, the object has to be a global since the translate function 
needs access to it.

John
=:->
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 253 bytes
Desc: OpenPGP digital signature
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041106/0248768f/attachment.pgp>

From nicodemus at esss.com.br  Sun Nov  7 06:38:49 2004
From: nicodemus at esss.com.br (Nicodemus)
Date: Sun, 07 Nov 2004 02:38:49 -0300
Subject: [C++-sig] Pyste Error
In-Reply-To: <418A4796.70004@syynx.de>
References: <418A4796.70004@syynx.de>
Message-ID: <418DB4E9.4080108@esss.com.br>

Hi Achim,

Can you post a small example that reproduces this error? It would be 
very helpful!

Regards,
Bruno.

Achim Domma wrote:

> Hi,
>
> could somebody explain to me, what could cause the error below? I have 
> no idea what's going wrong.
>
> regards,
> Achim
>
>
> Traceback (most recent call last):
>   File "/usr/bin/pyste.py", line 4, in ?
>     pyste.main()
>   File "/usr/lib/python2.3/site-packages/Pyste/pyste.py", line 418, in 
> main
>     status = Begin()
>   File "/usr/lib/python2.3/site-packages/Pyste/pyste.py", line 257, in 
> Begin
>     return GenerateCode(parser, module, out, interfaces, multiple)
>   File "/usr/lib/python2.3/site-packages/Pyste/pyste.py", line 363, in 
> GenerateCode
>     exports.extend(order[interface])
> KeyError: 'pyste_files/DrawablePushClipPath.pyste'
>
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>



From jdhunter at ace.bsd.uchicago.edu  Mon Nov  8 16:27:25 2004
From: jdhunter at ace.bsd.uchicago.edu (John Hunter)
Date: Mon, 08 Nov 2004 09:27:25 -0600
Subject: [C++-sig] typedef'd unnamed enum / pyste
Message-ID: <m3hdo0mkde.fsf@peds-pc311.bsd.uchicago.edu>


I am using boost_python 1.31.0 with pyste and gccxml-0.6.0, and am
trying to wrap an enum defined like

    typedef unsigned char cover_type;    //----cover_type
    enum
    {
        cover_shift = 8,                 //----cover_shift
        cover_size  = 1 << cover_shift,  //----cover_size 
        cover_mask  = cover_size - 1,    //----cover_mask 
        cover_none  = 0,                 //----cover_none 
        cover_full  = cover_mask         //----cover_full 
    };


How does one wrap an enum like this in pyste?  If I try the naive,

Enum("agg::cover_type", "agg_basics.h")

I get an exception in pyste, included below -- the enum object appears
to not have the values attribute.

FYI, other functions that I need to wrap will pass cover_type args as
well as cover_type arrays.

Thanks for any help, I've read though some of the archives on how to
handle unnamed enums, but wasn't able to grok the answers to figure
out what I should do.

JDH


peds-pc311:~/python/projects/aggboost> pyste.py -I/usr/local/include/agg --module=bagg --out=src/basics.cpp pyste/basics.pyste
gccxml -I "/tmp" -I "/usr/local/include/agg"  /tmp/tmpBsdOix.h -fxml=/tmp/tmpDRjYOM.xml
/tmp/tmpDRjYOM.xml
Traceback (most recent call last):
  File "/usr/local/bin/pyste.py", line 4, in ?
    pyste.main()
  File "/usr/local/lib/python2.3/site-packages/Pyste/pyste.py", line 405, in main
    status = Begin()
  File "/usr/local/lib/python2.3/site-packages/Pyste/pyste.py", line 244, in Begin
    return GenerateCode(parser, module, out, interfaces, multiple)
  File "/usr/local/lib/python2.3/site-packages/Pyste/pyste.py", line 372, in GenerateCode
    export.GenerateCode(codeunit, exported_names)
  File "/usr/local/lib/python2.3/site-packages/Pyste/Exporter.py", line 50, in GenerateCode
    self.Export(codeunit, exported_names)
  File "/usr/local/lib/python2.3/site-packages/Pyste/EnumExporter.py", line 44, in Export
    for name in self.enum.values:
AttributeError: 'Typedef' object has no attribute 'values'
[2]  - Done                          sudo emacs -q /usr/local/lib/python2.3/site-packages/Pyste/EnumExporter.py


From jdhunter at ace.bsd.uchicago.edu  Mon Nov  8 16:58:49 2004
From: jdhunter at ace.bsd.uchicago.edu (John Hunter)
Date: Mon, 08 Nov 2004 09:58:49 -0600
Subject: [C++-sig] pyste and template functions
Message-ID: <m3zn1sl4cm.fsf@peds-pc311.bsd.uchicago.edu>


I am using pyste to wrap a template function.  I have used Template
successfully to wrap a templated class, but this does not appear to
support template functions (at least I didn't see anything in the
docs).

I can manually do it with module_code


  module_code('def("intersect_rectangles", &agg::intersect_rectangles<agg::rect>);');
  module_code('def("intersect_rectangles_d", &agg::intersect_rectangles<agg::rect_d>);');


Is there a way to do this directly in pyste?

Thanks!
JDH

boost_python 1.31.0 with pyste and gccxml-0.6.0


From jdhunter at ace.bsd.uchicago.edu  Mon Nov  8 18:46:19 2004
From: jdhunter at ace.bsd.uchicago.edu (John Hunter)
Date: Mon, 08 Nov 2004 11:46:19 -0600
Subject: [C++-sig] pyste + wrapping + overloads
Message-ID: <m3u0s0dyj8.fsf@peds-pc311.bsd.uchicago.edu>


I am trying to wrap a class method in pyste that has a signature

    void path_storage::add_poly(const double* vertices, unsigned num,
                                bool solid_path, unsigned end_flags)

vertices is a length 2*num array of x,y doubles x0, y0, x1, y1, ....

I am using the following wrapper

  void add_poly_wrapper(agg::path_storage* ps, list vertices, 
			bool solid_path = false,
			unsigned end_flags = agg::path_flags_none) {
    size_t N = extract<size_t>(vertices.attr("length")());
    double * pverts = new double[2*N];
    for (size_t i=0; i<N; i++) {
      tuple xy  = extract<tuple>(vertices[i]);
      *pverts++ = extract<double>(xy[0]);
      *pverts++ = extract<double>(xy[1]);
    }
    pverts -= 2*N;  //rewind
    ps->add_poly(pverts, N, solid_path, end_flags);
    
  }

The first problem I've encountered in pyste is that the standard way of
specifying wrappers

    Include("path_storage_wrap.h")
    PS = Class("agg::path_storage", "agg_path_storage.h")
    set_wrapper(PS.add_poly, "add_poly_wrapper")

doesn't work because there is no PS.add_poly, since pyste defines it
with the overload macro


  BOOST_PYTHON_MEMBER_FUNCTION_OVERLOADS(agg_path_storage_add_poly_overloads_2_4, add_poly, 2, 4)


How does one specify wrappers for overloaded functions?  I don't know
enough about the overload macro internals to figure out how to
proceed...

Thanks,
JDH

boost_python 1.31.0 with pyste and gccxml-0.6.0


From mrovner at propel.com  Mon Nov  8 19:09:36 2004
From: mrovner at propel.com (Mike Rovner)
Date: Mon, 08 Nov 2004 10:09:36 -0800
Subject: [C++-sig] Re: Custom Exception objects
In-Reply-To: <418CE4E7.3080804@johnmeinel.com>
References: <41880F99.20609@arbash-meinel.com> <cm94es$lop$1@sea.gmane.org>
	<418CE4E7.3080804@johnmeinel.com>
Message-ID: <cmocod$orf$1@sea.gmane.org>

John A Meinel wrote:
> Basically, the object has to be a global since the translate function 
> needs access to it.

I'm not sure what are you asking. Let's return to basic.

a. If you have C++ exception that you need to register_exception_translator.

b. If you want to expose your exception, you wrap it in a class_.

c. Scope rules are basically normal python rules: in order to use 
exception you need to have in in your local or global namespace.

Combining all three you example is a possible implementation.

Best,
Mike



From jdhunter at ace.bsd.uchicago.edu  Mon Nov  8 23:08:25 2004
From: jdhunter at ace.bsd.uchicago.edu (John Hunter)
Date: Mon, 08 Nov 2004 16:08:25 -0600
Subject: [C++-sig] pyste and operator*
Message-ID: <m3y8hc0zae.fsf@peds-pc311.bsd.uchicago.edu>


I have a class agg::path_storage which nests another class
agg::path_storage::const_iterator that defines operator*().  When I
run pyste on this file, I get a compile time error

src/test.cpp: In function `void init_module_tagg()':
src/test.cpp:22: error: no match for 'operator*' in '*
   boost::python::self_ns::self'
error: command 'gcc' failed with exit status 1

The part of the boost generated code that is failing is the .def(
*self ) line in 

    class_< agg::path_storage::const_iterator >("const_iterator", init<  >())
        .def(init< const agg::path_storage::const_iterator& >())
        .def( *self )
    ;

I tried removing operator* in pyste, but did not succeed.  Here is my
complete test.pyste file

PS = Class("agg::path_storage", "test.h")
CI = Class("agg::path_storage::const_iterator", "test.h")
exclude(CI.operator['*'])


and here is the minimal test.h example code that is failing

namespace agg
{
    struct vertex_type
    {
        double   x, y;
        unsigned cmd;

        vertex_type() {}
        vertex_type(double x_, double y_, unsigned cmd_) : 
            x(x_), y(y_), cmd(cmd_) {}
    };

    class path_storage
    {
    public:

        class const_iterator
        {

        public:
            const_iterator() {}
	    const vertex_type& operator*() const { return m_vertex; }

        private:
            vertex_type         m_vertex;
        };

        ~path_storage() {};
        path_storage() {};

    };
}


Thanks,
JDH

boost_python 1.31.0 with pyste and gccxml-0.6.0


From jdhunter at ace.bsd.uchicago.edu  Tue Nov  9 00:40:46 2004
From: jdhunter at ace.bsd.uchicago.edu (John Hunter)
Date: Mon, 08 Nov 2004 17:40:46 -0600
Subject: [C++-sig] pyste and return reference problem
Message-ID: <m3sm7jx62p.fsf@peds-pc311.bsd.uchicago.edu>


I have found what looks like pyste bug.  pyste wraps the following
struct

  struct C {
    const C& func(double val) {return *this;}
    double func() {return 0.0;}
  };

as

    class_< C >("C", init<  >())
        .def(init< const C& >())
        .def("func", (const C& (C::*)(double) )&C::func, return_value_policy< copy_const_reference >())
        .def("func", (double (C::*)() )&C::func, return_value_policy< copy_const_reference >())
    ;
 


Note that (double (C::*)() )&C::func has a call policy of
return_value_policy< copy_const_reference >()).  

If you reverse the order of the method definitions in C, ie


  struct C {
    double func() {return 0.0;}
    const C& func(double val) {return *this;}
  };

pyste gets it right

    class_< C >("C", init<  >())
        .def(init< const C& >())
        .def("func", (double (C::*)() )&C::func)
        .def("func", (const C& (C::*)(double) )&C::func, return_value_policy< copy_const_reference >())
    ;


This leads me to a question related to one of my earlier questions:
when wrapping overloaded functions as above, is there a way to
differentially set the call policies in pyste for the different
signatures.  I know one can use as described in the docs

    set_policy(f, return_internal_reference())
    set_policy(C.foo, return_value_policy(manage_new_object))

How does one do this for overloaded functions?
 
Sorry for the deluge of mail!  I'm going home now :-)

JDH

boost_python 1.31.0 with pyste and gccxml-0.6.0


From dholth at fastmail.fm  Tue Nov  9 02:34:46 2004
From: dholth at fastmail.fm (Daniel Holth)
Date: Mon, 08 Nov 2004 20:34:46 -0500
Subject: [C++-sig] Microsoft Visual C++ Toolkit 2003 (command line)
	boost.python
Message-ID: <1099964086.5632.208244874@webmail.messagingengine.com>

Can someone help me to compile boost.python with this freely
downloadable Microsoft compiler,
http://msdn.microsoft.com/visualc/vctoolkit2003/? I have not yet been
able to get boost to find this compiler that installs itself in
c:\Program Files\Microsoft Visual C++ Toolkit 2003\bin 

Thanks,

Daniel


From caustin at gmail.com  Tue Nov  9 05:14:02 2004
From: caustin at gmail.com (Chad Austin)
Date: Mon, 8 Nov 2004 22:14:02 -0600
Subject: [C++-sig] vector of shared_ptr ?
Message-ID: <2096514e0411082014a1d1120@mail.gmail.com>

I am trying to export a vector of smart pointers to Python using
vector_indexing_suite.  Here is a test case:


// Begin C++ Code
#include <boost/shared_ptr.hpp>
#include <boost/python.hpp>
#include <boost/python/suite/indexing/vector_indexing_suite.hpp>
#include <vector>
using namespace boost;
using namespace boost::python;


class A {
};
typedef shared_ptr<A> APtr;

typedef std::vector<APtr> AList;

BOOST_PYTHON_MODULE(ptrvector) {
    class_<A, APtr>("A")
        ;
    register_ptr_to_python<APtr>();

    class_<AList>("AList")
        .def(vector_indexing_suite<AList>())
        ;
}
// End C++ Code


I would expect the following test case to succeed:

# Begin Python Code
from ptrvector import *
a = AList()
a[:] = [A()]
for b in a:
    print b
# End Python Code

However, I get an error:

$ python test.py
Traceback (most recent call last):
  File "test.py", line 4, in ?
    for b in a:
TypeError: No Python class registered for C++ class boost::shared_ptr<A>


Is this a known bug?  Is there anything I can do to work around or fix this?

Thanks,
Chad


From roman.yakovenko at gmail.com  Tue Nov  9 05:50:18 2004
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 9 Nov 2004 06:50:18 +0200
Subject: [C++-sig] typedef'd unnamed enum / pyste
In-Reply-To: <m3hdo0mkde.fsf@peds-pc311.bsd.uchicago.edu>
References: <m3hdo0mkde.fsf@peds-pc311.bsd.uchicago.edu>
Message-ID: <7465b61704110820505e09c961@mail.gmail.com>

On Mon, 08 Nov 2004 09:27:25 -0600, John Hunter
<jdhunter at nitace.bsd.uchicago.edu> wrote:
> 
> I am using boost_python 1.31.0 with pyste and gccxml-0.6.0, and am
> trying to wrap an enum defined like
> 
>     typedef unsigned char cover_type;    //----cover_type

Pyste is right cover_type is typedef to unsigned char. You try to
expose it as enum ???
You made a mistake.

>     enum
>     {
>         cover_shift = 8,                 //----cover_shift
>         cover_size  = 1 << cover_shift,  //----cover_size
>         cover_mask  = cover_size - 1,    //----cover_mask
>         cover_none  = 0,                 //----cover_none
>         cover_full  = cover_mask         //----cover_full
>     };
> 
> How does one wrap an enum like this in pyste?  If I try the naive,
> 
> Enum("agg::cover_type", "agg_basics.h")

Enum( "cover_mask", "agg_basics.h" )  should work 

> 
> I get an exception in pyste, included below -- the enum object appears
> to not have the values attribute.
> 
> FYI, other functions that I need to wrap will pass cover_type args as
> well as cover_type arrays.
> 
> Thanks for any help, I've read though some of the archives on how to
> handle unnamed enums, but wasn't able to grok the answers to figure
> out what I should do.
> 
> JDH
> 
> peds-pc311:~/python/projects/aggboost> pyste.py -I/usr/local/include/agg --module=bagg --out=src/basics.cpp pyste/basics.pyste
> gccxml -I "/tmp" -I "/usr/local/include/agg"  /tmp/tmpBsdOix.h -fxml=/tmp/tmpDRjYOM.xml
> /tmp/tmpDRjYOM.xml
> Traceback (most recent call last):
>   File "/usr/local/bin/pyste.py", line 4, in ?
>     pyste.main()
>   File "/usr/local/lib/python2.3/site-packages/Pyste/pyste.py", line 405, in main
>     status = Begin()
>   File "/usr/local/lib/python2.3/site-packages/Pyste/pyste.py", line 244, in Begin
>     return GenerateCode(parser, module, out, interfaces, multiple)
>   File "/usr/local/lib/python2.3/site-packages/Pyste/pyste.py", line 372, in GenerateCode
>     export.GenerateCode(codeunit, exported_names)
>   File "/usr/local/lib/python2.3/site-packages/Pyste/Exporter.py", line 50, in GenerateCode
>     self.Export(codeunit, exported_names)
>   File "/usr/local/lib/python2.3/site-packages/Pyste/EnumExporter.py", line 44, in Export
>     for name in self.enum.values:
> AttributeError: 'Typedef' object has no attribute 'values'
> [2]  - Done                          sudo emacs -q /usr/local/lib/python2.3/site-packages/Pyste/EnumExporter.py
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>


From roman.yakovenko at gmail.com  Tue Nov  9 05:53:37 2004
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 9 Nov 2004 06:53:37 +0200
Subject: [C++-sig] pyste and template functions
In-Reply-To: <m3zn1sl4cm.fsf@peds-pc311.bsd.uchicago.edu>
References: <m3zn1sl4cm.fsf@peds-pc311.bsd.uchicago.edu>
Message-ID: <7465b6170411082053b0531ab@mail.gmail.com>

On Mon, 08 Nov 2004 09:58:49 -0600, John Hunter
<jdhunter at nitace.bsd.uchicago.edu> wrote:
> 
> I am using pyste to wrap a template function.  I have used Template
> successfully to wrap a templated class, but this does not appear to
> support template functions (at least I didn't see anything in the
> docs).
> 
> I can manually do it with module_code
> 
>   module_code('def("intersect_rectangles", &agg::intersect_rectangles<agg::rect>);');
>   module_code('def("intersect_rectangles_d", &agg::intersect_rectangles<agg::rect_d>);');

My guess - this has nothing to do with Pyste, but with C++.  
I think you should have instantiate the function explicitly in code.
only then you can expose
it. But I could be wrong. 

> Is there a way to do this directly in pyste?
> 
> Thanks!
> JDH
> 
> boost_python 1.31.0 with pyste and gccxml-0.6.0
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>


From roman.yakovenko at gmail.com  Tue Nov  9 05:56:59 2004
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 9 Nov 2004 06:56:59 +0200
Subject: [C++-sig] pyste and operator*
In-Reply-To: <m3y8hc0zae.fsf@peds-pc311.bsd.uchicago.edu>
References: <m3y8hc0zae.fsf@peds-pc311.bsd.uchicago.edu>
Message-ID: <7465b61704110820565ab5edcf@mail.gmail.com>

On Mon, 08 Nov 2004 16:08:25 -0600, John Hunter
<jdhunter at nitace.bsd.uchicago.edu> wrote:
> 
> I have a class agg::path_storage which nests another class
> agg::path_storage::const_iterator that defines operator*().  When I
> run pyste on this file, I get a compile time error
> 
> src/test.cpp: In function `void init_module_tagg()':
> src/test.cpp:22: error: no match for 'operator*' in '*
>    boost::python::self_ns::self'
> error: command 'gcc' failed with exit status 1
> 
> The part of the boost generated code that is failing is the .def(
> *self ) line in
> 
>     class_< agg::path_storage::const_iterator >("const_iterator", init<  >())
>         .def(init< const agg::path_storage::const_iterator& >())
>         .def( *self )
>     ;
> 
> I tried removing operator* in pyste, but did not succeed.  Here is my
> complete test.pyste file
> 
> PS = Class("agg::path_storage", "test.h")
> CI = Class("agg::path_storage::const_iterator", "test.h")
> exclude(CI.operator['*'])
> 
> and here is the minimal test.h example code that is failing
> 
> namespace agg
> {
>     struct vertex_type
>     {
>         double   x, y;
>         unsigned cmd;
> 
>         vertex_type() {}
>         vertex_type(double x_, double y_, unsigned cmd_) :
>             x(x_), y(y_), cmd(cmd_) {}
>     };
> 
>     class path_storage
>     {
>     public:
> 
>         class const_iterator
>         {
> 
>         public:
>             const_iterator() {}
>             const vertex_type& operator*() const { return m_vertex; }

This operation is meaningless \doesn't exis in python. If  you still
want to expose
it - expose it as regular function

> 
>         private:
>             vertex_type         m_vertex;
>         };
> 
>         ~path_storage() {};
>         path_storage() {};
> 
>     };
> }
> 
> Thanks,
> JDH
> 
> boost_python 1.31.0 with pyste and gccxml-0.6.0
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>


From jdhunter at ace.bsd.uchicago.edu  Tue Nov  9 15:55:46 2004
From: jdhunter at ace.bsd.uchicago.edu (John Hunter)
Date: Tue, 09 Nov 2004 08:55:46 -0600
Subject: [C++-sig] pyste and operator*
In-Reply-To: <7465b61704110820565ab5edcf@mail.gmail.com> (Roman Yakovenko's
	message of "Tue, 9 Nov 2004 06:56:59 +0200")
References: <m3y8hc0zae.fsf@peds-pc311.bsd.uchicago.edu>
	<7465b61704110820565ab5edcf@mail.gmail.com>
Message-ID: <m3wtwv3wct.fsf@peds-pc311.bsd.uchicago.edu>

>>>>> "Roman" == Roman Yakovenko <roman.yakovenko at gmail.com> writes:

    John> I tried removing operator* in pyste, but did not succeed.  Here
    John> is my complete test.pyste file
    John> 
    John> public: const_iterator() {} const vertex_type& operator*()
    John> const { return m_vertex; }

    Roman> This operation is meaningless \doesn't exis in python. 

Right, which is why I tried to remove it with 

    CI = Class("agg::path_storage::const_iterator", "test.h")
    exclude(CI.operator['*'])

Unfortunately, pyste still generates the line

        .def( *self )

which is the problem I addressed in my OP.

Thanks,
JDH


From dave at boost-consulting.com  Tue Nov  9 16:25:43 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 09 Nov 2004 10:25:43 -0500
Subject: [C++-sig] Re: pyste and template functions
References: <m3zn1sl4cm.fsf@peds-pc311.bsd.uchicago.edu>
	<7465b6170411082053b0531ab@mail.gmail.com>
Message-ID: <uvfcfxcw8.fsf@boost-consulting.com>

Roman Yakovenko <roman.yakovenko at gmail.com> writes:

> On Mon, 08 Nov 2004 09:58:49 -0600, John Hunter
> <jdhunter at nitace.bsd.uchicago.edu> wrote:
>> 
>> I am using pyste to wrap a template function.  I have used Template
>> successfully to wrap a templated class, but this does not appear to
>> support template functions (at least I didn't see anything in the
>> docs).
>> 
>> I can manually do it with module_code
>> 
>>   module_code('def("intersect_rectangles", &agg::intersect_rectangles<agg::rect>);');
>>   module_code('def("intersect_rectangles_d", &agg::intersect_rectangles<agg::rect_d>);');
>
> My guess - this has nothing to do with Pyste, but with C++.  
> I think you should have instantiate the function explicitly in code.
> only then you can expose
> it. But I could be wrong. 

That's wrong.  You can't wrap a function template, though.  You can
only wrap a particular specialization of that function template.  I'm
not sure what you're doing that isn't working, though.  You only
showed what worked.

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From dave at boost-consulting.com  Tue Nov  9 16:29:50 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 09 Nov 2004 10:29:50 -0500
Subject: [C++-sig] Re: pyste and operator*
References: <m3y8hc0zae.fsf@peds-pc311.bsd.uchicago.edu>
	<7465b61704110820565ab5edcf@mail.gmail.com>
Message-ID: <upt2nxcpd.fsf@boost-consulting.com>

Roman Yakovenko <roman.yakovenko at gmail.com> writes:

>>         class const_iterator
>>         {
>> 
>>         public:
>>             const_iterator() {}
>>             const vertex_type& operator*() const { return m_vertex; }
>
> This operation is meaningless \doesn't exis in python. If  you still
> want to expose
> it - expose it as regular function

On the other hand there ought to be some way to get pyste to expose
C++ iterators as Python iterators.  See
http://www.boost.org/libs/python/doc/v2/iterator.html

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From jdhunter at ace.bsd.uchicago.edu  Tue Nov  9 16:38:59 2004
From: jdhunter at ace.bsd.uchicago.edu (John Hunter)
Date: Tue, 09 Nov 2004 09:38:59 -0600
Subject: [C++-sig] typedef'd unnamed enum / pyste
In-Reply-To: <7465b61704110820505e09c961@mail.gmail.com> (Roman Yakovenko's
	message of "Tue, 9 Nov 2004 06:50:18 +0200")
References: <m3hdo0mkde.fsf@peds-pc311.bsd.uchicago.edu>
	<7465b61704110820505e09c961@mail.gmail.com>
Message-ID: <m3y8hb117w.fsf@peds-pc311.bsd.uchicago.edu>

>>>>> "Roman" == Roman Yakovenko <roman.yakovenko at gmail.com> writes:

    >> 
    >> typedef unsigned char cover_type; //----cover_type

    Roman> Pyste is right cover_type is typedef to unsigned char. You
    Roman> try to expose it as enum ???  You made a mistake.

Yes, I'm a moron.  Thanks!


    typedef unsigned char cover_type;    //----cover_type
    enum
    {
        cover_shift = 8,                 //----cover_shift
        cover_size  = 1 << cover_shift,  //----cover_size 
        cover_mask  = cover_size - 1,    //----cover_mask 
        cover_none  = 0,                 //----cover_none 
        cover_full  = cover_mask         //----cover_full 
    };

    Roman> Enum( "cover_mask", "agg_basics.h" ) should work

It doesn't.  I was, however, able to do it manually with

module_code("""    
    enum_<int>("cover")
        .value("cover_shift", agg::cover_shift)
        .value("cover_size",  agg::cover_size)
        .value("cover_mask",  agg::cover_mask)
        .value("cover_none",  agg::cover_none)
        .value("cover_full",  agg::cover_full)                        
        ;
        """)

but as always, I'm looking for the best pyste/boost::python way to do
it, so any other suggestions on the best way to export unnamed enums
in pyste welcome!

Thanks,
JDH


From jdhunter at ace.bsd.uchicago.edu  Tue Nov  9 16:47:46 2004
From: jdhunter at ace.bsd.uchicago.edu (John Hunter)
Date: Tue, 09 Nov 2004 09:47:46 -0600
Subject: [C++-sig] Re: pyste and template functions
In-Reply-To: <uvfcfxcw8.fsf@boost-consulting.com> (David Abrahams's message
	of "Tue, 09 Nov 2004 10:25:43 -0500")
References: <m3zn1sl4cm.fsf@peds-pc311.bsd.uchicago.edu>
	<7465b6170411082053b0531ab@mail.gmail.com>
	<uvfcfxcw8.fsf@boost-consulting.com>
Message-ID: <m3vfcf10t9.fsf@peds-pc311.bsd.uchicago.edu>

>>>>> "David" == David Abrahams <dave at boost-consulting.com> writes:

   
    David> I'm not sure what you're doing that isn't working, though.
    David> You only showed what worked.

Oops.  I tried following the approach for template classes

    Intersect = Template("agg::intersect_rectangles", "agg_basics.h")
    Intersect("agg::rect", "intersect_rectangles")
    Intersect("agg::rect_d", "intersect_rectangles_d")

where the signature for the template function intersect_rectangles is

    template<class Rect> 
    inline Rect intersect_rectangles(const Rect& r1, const Rect& r2)

pyste generates the following code (line number 321 marked for
compiler error reference below)

321-> typedef agg::intersect_rectangles< agg::rect > intersect_rectangles;
      void __instantiate_intersect_rectangles()
      { sizeof(intersect_rectangles); }


      typedef agg::intersect_rectangles< agg::rect_d > intersect_rectangles_d;
      void __instantiate_intersect_rectangles_d()
      { sizeof(intersect_rectangles_d); }

which causes gccxml to die with 

    /tmp/tmpcfn9df.h:321: error: syntax error before `;' token
    /tmp/tmpcfn9df.h: In function `void __instantiate_intersect_rectangles()':
    /tmp/tmpcfn9df.h:323: error: ISO C++ forbids applying `sizeof' to an expression
    of function type


where line 321 is indicated above.

I assumed thusly that pyste either did not handle template functions
or at least I didn't know how to handle them in pyste, so I reverted
to the module_code I mentioned above.

Thanks!
JDH


From jdhunter at ace.bsd.uchicago.edu  Tue Nov  9 18:17:47 2004
From: jdhunter at ace.bsd.uchicago.edu (John Hunter)
Date: Tue, 09 Nov 2004 11:17:47 -0600
Subject: [C++-sig] pyste instantiating templates
Message-ID: <m3y8hbuekk.fsf@peds-pc311.bsd.uchicago.edu>


I have a template class in someclass.h that is instantiated over
sometype.h.  someclass.h does not #include sometype.h.

If I try the standard way of instantiating templates in pyste

  SomeTemplate = Template("SomeClass", "someclass.h")
  TemplateSometype = SomeTemplate("Sometype", "template_sometype")

I get a gccxml error because SomeType is not known in someclass.h.

It would be nice to be able to tell pyste in the tmp*.h files it
generates to pass to gccxml, to include some additional headers.  Is
this possible?  None of Include, declaration_code or module_code seem
to fit the bill here, since I want to include code into the file that
I am actually wrapping.

I worked around it by creating a 

instantiate_someclass.h

  #include sometype.h
  #include someclass.h

and then doing

  SomeTemplate = Template("SomeClass", "instantiate_someclass.h")
  TemplateSometype = SomeTemplate("Sometype", "template_sometype")



JDH


From abentley at panoramicfeedback.com  Tue Nov  9 21:08:42 2004
From: abentley at panoramicfeedback.com (Aaron Bentley)
Date: Tue, 9 Nov 2004 20:08:42 +0000 (UTC)
Subject: [C++-sig] boost.Python/Jam with shared libraries
Message-ID: <loom.20041109T180856-543@post.gmane.org>

Hi,

I'm trying to add a python interface to some of the functionality in an existing
app.  I've managed to get a test version to compile, but the result doesn't work
because shared libraries (both commonly-available and app-specific) in are
required for our app.

I've tried modifying the Jamfile as suggested in the wiki, adding library-dir
and library-file directives, but I get 

/home/abentley/install/boost_1_31_0/tools/build/v1/boost-base.jam:1972: in
expand-target-subvariants
rule /home/abentley/programming/.libs/libpanutil.so unknown in module 

We're usin Autoconf in the rest of the project, and I must admit my approach to
Makefile.am is mostly a cargo-cult mentality-- I wouldn't know the correct way
to link shared objects by hand.

Can anyone help?

Aaron



From grafik.list at redshift-software.com  Tue Nov  9 22:52:52 2004
From: grafik.list at redshift-software.com (Rene Rivera)
Date: Tue, 09 Nov 2004 15:52:52 -0600
Subject: [C++-sig] boost.Python/Jam with shared libraries
In-Reply-To: <loom.20041109T180856-543@post.gmane.org>
References: <loom.20041109T180856-543@post.gmane.org>
Message-ID: <41913C34.8090507@redshift-software.com>

Aaron Bentley wrote:

> I'm trying to add a python interface to some of the functionality in an existing
> app.  I've managed to get a test version to compile, but the result doesn't work
> because shared libraries (both commonly-available and app-specific) in are
> required for our app.
> 
> I've tried modifying the Jamfile as suggested in the wiki, adding library-dir
> and library-file directives, but I get 

Which Jamfile?

> Can anyone help?

Probably. But we would need more information. Like answering the above 
question. And providing the Jamfile you have. At least.


-- 
-- Grafik - Don't Assume Anything
-- Redshift Software, Inc. - http://redshift-software.com
-- rrivera/acm.org - grafik/redshift-software.com - 102708583/icq


From grafik.list at redshift-software.com  Tue Nov  9 22:57:32 2004
From: grafik.list at redshift-software.com (Rene Rivera)
Date: Tue, 09 Nov 2004 15:57:32 -0600
Subject: [C++-sig] Re: python extension and bjam stage
In-Reply-To: <418A044C.2030909@procoders.net>
References: <418A044C.2030909@procoders.net>
Message-ID: <41913D4C.9080401@redshift-software.com>

Achim Domma (Procoders) wrote:

> I cannot figure out how to use the stage rule with boost.python. Could 
> somebody give me an example?

I don't have a complete example, all you should need to do is add a 
reference to the extension into the sources section of the stage target:

stage mystage : <pyd>myextension ;


-- 
-- Grafik - Don't Assume Anything
-- Redshift Software, Inc. - http://redshift-software.com
-- rrivera/acm.org - grafik/redshift-software.com - 102708583/icq



From abentley at panoramicfeedback.com  Tue Nov  9 23:14:58 2004
From: abentley at panoramicfeedback.com (Aaron Bentley)
Date: Tue, 9 Nov 2004 22:14:58 +0000 (UTC)
Subject: [C++-sig] Re: boost.Python/Jam with shared libraries
References: <loom.20041109T180856-543@post.gmane.org>
	<41913C34.8090507@redshift-software.com>
Message-ID: <loom.20041109T231245-517@post.gmane.org>

Rene Rivera <grafik.list <at> redshift-software.com> writes:

> 
> Which Jamfile?
> 
> > Can anyone help?
> 
> Probably. But we would need more information. Like answering the above 
> question. And providing the Jamfile you have. At least.
> 

Here's the Jamfile in question:


# Copyright David Abrahams 2003. See accompanying LICENSE for terms
# and conditions of use.

# This is the top of our own project tree
project-root ;

# Declares the following targets:
#    
#  1. an extension module called "getting_started1", which is
#     built from "getting_started1.cpp". Built by default
#
#  2. A test target called my-test.test which runs
#     test_getting_started1.py with the extension module above. Built
#     when out-of date, but only if invoked by name or if the global
#     "test" target is invoked.
#
#  3. A test target called my-test.run wihch runs the above test
#     unconditionally. Built only when invoked by name.
#
# To see verbose test output, add "-sPYTHON_TEST_ARGS=-v" to the bjam
# command-line before the first target.
#

# Include definitions needed for Python modules
import python ;

   
# ----- getting_started2 -------
    
# Declare a Python extension called pfengine
extension pfengine
:  # sources
   pfengine.cpp
   #libraries
:
#  <library-path> ../.libs
  <library-file> /home/abentley/programming/.libs/libpanutil.so
#  <library-file> libpantemplate.la
#  <library-file> libpanpdfgen.la 
#  <library-file> libpancompile.la 
#  <library-file> libpanconsult.la 
#  <library-file> libpanelement.la
#  <library-file> libpanlang.la 
#  <library-file> libpancore.la 
#  <library-file> libpanexport.la
#  <library-file> libpanreport.la
#  <library-file> libpancgi.la
#  <library-file> libchart.la 
#  <library-file> libecommerce.la

  # requirements and dependencies for Boost.Python extensions
  <template>@boost/libs/python/build/extension
  ;

# Declare a test for the extension module
boost-python-runtest test2
    :  # Python test driver
    test_getting_started2.py
    # extension modules to use
    <pyd>pfengine ;    

# arch-tag: ae55dfed-2576-4667-953f-d7d8f777f7c4



From grafik.list at redshift-software.com  Wed Nov 10 00:03:25 2004
From: grafik.list at redshift-software.com (Rene Rivera)
Date: Tue, 09 Nov 2004 17:03:25 -0600
Subject: [C++-sig] Re: boost.Python/Jam with shared libraries
In-Reply-To: <loom.20041109T231245-517@post.gmane.org>
References: <loom.20041109T180856-543@post.gmane.org>	<41913C34.8090507@redshift-software.com>
	<loom.20041109T231245-517@post.gmane.org>
Message-ID: <41914CBD.5060002@redshift-software.com>

Aaron Bentley wrote:

> # ----- getting_started2 -------
>     
> # Declare a Python extension called pfengine
> extension pfengine
> :  # sources
>    pfengine.cpp
>    #libraries
> :
> #  <library-path> ../.libs
>   <library-file> /home/abentley/programming/.libs/libpanutil.so
> #  <library-file> libpantemplate.la
> #  <library-file> libpanpdfgen.la 
> #  <library-file> libpancompile.la 
> #  <library-file> libpanconsult.la 
> #  <library-file> libpanelement.la
> #  <library-file> libpanlang.la 
> #  <library-file> libpancore.la 
> #  <library-file> libpanexport.la
> #  <library-file> libpanreport.la
> #  <library-file> libpancgi.la
> #  <library-file> libchart.la 
> #  <library-file> libecommerce.la

Remove the space---^-- between <library-file> and lib*.


-- 
-- Grafik - Don't Assume Anything
-- Redshift Software, Inc. - http://redshift-software.com
-- rrivera/acm.org - grafik/redshift-software.com - 102708583/icq


From john at johnmeinel.com  Wed Nov 10 00:22:46 2004
From: john at johnmeinel.com (John Meinel)
Date: Tue, 09 Nov 2004 17:22:46 -0600
Subject: [C++-sig] Re: boost.Python/Jam with shared libraries
In-Reply-To: <loom.20041109T231245-517@post.gmane.org>
References: <loom.20041109T180856-543@post.gmane.org>	<41913C34.8090507@redshift-software.com>
	<loom.20041109T231245-517@post.gmane.org>
Message-ID: <41915146.5050207@johnmeinel.com>

Aaron Bentley wrote:
> Rene Rivera <grafik.list <at> redshift-software.com> writes:
> 
[...]

> Here's the Jamfile in question:
> 
> 
> # Copyright David Abrahams 2003. See accompanying LICENSE for terms
> # and conditions of use.
> 
> # This is the top of our own project tree
> project-root ;
> 
> # Include definitions needed for Python modules
> import python ;
> 
>    
> # ----- getting_started2 -------
>     
> # Declare a Python extension called pfengine

# You might try this

extension pfengine
   :# sources
     pfengine.cpp
   :
      <library-path>../.libs
      <find-library>panutil
      $(BOOST_PYTHON_V2_PROPERTIES)
;

John
=:->
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 256 bytes
Desc: OpenPGP digital signature
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041109/5e1315b0/attachment.pgp>

From abentley at panoramicfeedback.com  Wed Nov 10 15:51:50 2004
From: abentley at panoramicfeedback.com (Aaron Bentley)
Date: Wed, 10 Nov 2004 14:51:50 +0000 (UTC)
Subject: [C++-sig] Re: boost.Python/Jam with shared libraries
References: <loom.20041109T180856-543@post.gmane.org>	<41913C34.8090507@redshift-software.com>
	<loom.20041109T231245-517@post.gmane.org>
	<41915146.5050207@johnmeinel.com>
Message-ID: <loom.20041110T150821-13@post.gmane.org>

John Meinel <john <at> johnmeinel.com> writes:

> # You might try this
> 
> extension pfengine
>    :# sources
>      pfengine.cpp
>    :
>       <library-path>../.libs
>       <find-library>panutil
>       $(BOOST_PYTHON_V2_PROPERTIES)
> ;
> 

Hi John,

Fancy meeting you here!

The eliminating-spaces route got rid of the errors, but find-library looks
cleaner.  Thanks.

So I'm much further along.  Now I'm trying to link in a static file.  It appears
to work, but then I get "symbol missing".

---begin Jamfile---
project-root ;

import python ;

# ----- getting_started2 -------

# Declare a Python extension called pfengine
extension pfengine
:  # sources
   pfengine.cpp
   #libraries
:
  <library-path>../.libs
  <find-library>panutil
  <find-library>pantemplate
  <find-library>panpdfgen
  <find-library>pancompile
  <find-library>panconsult
  <find-library>panelement
  <find-library>panlang
  <find-library>pancore
  <find-library>panexport
  <find-library>panreport
  <find-library>pancgi
  <find-library>chart
  <find-library>ecommerce
  <find-library>icuio
  <find-library>icudata
  <find-library>icutoolutil
  <find-library>icule
  <find-library>icui18n

  # requirements and dependencies for Boost.Python extensions
  <template>@boost/libs/python/build/extension
  ;

# Declare a test for the extension module
boost-python-runtest test2
    :  # Python test driver
    test_getting_started2.py
    # extension modules to use
    <pyd>pfengine ;

LinkLibraries pfengine : ../libpdf.a ;
# arch-tag: ae55dfed-2576-4667-953f-d7d8f777f7c4
---end jamfile---

---attempt to use---
>>> import pfengine
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
ImportError: /home/abentley/programming/.libs/libchart.so.0: undefined symbol:
PDF_setrgbcolor_fill
>>> 
---

PDF_setrgbcolor_fill is provided by libpdf.a

Any help appreciated.

Aaron





From grafik.list at redshift-software.com  Wed Nov 10 16:02:46 2004
From: grafik.list at redshift-software.com (Rene Rivera)
Date: Wed, 10 Nov 2004 09:02:46 -0600
Subject: [C++-sig] Re: boost.Python/Jam with shared libraries
In-Reply-To: <loom.20041110T150821-13@post.gmane.org>
References: <loom.20041109T180856-543@post.gmane.org>	<41913C34.8090507@redshift-software.com>	<loom.20041109T231245-517@post.gmane.org>	<41915146.5050207@johnmeinel.com>
	<loom.20041110T150821-13@post.gmane.org>
Message-ID: <41922D96.4070509@redshift-software.com>

Aaron Bentley wrote:

> The eliminating-spaces route got rid of the errors, but find-library looks
> cleaner.  Thanks.
> 
> So I'm much further along.  Now I'm trying to link in a static file.  It appears
> to work, but then I get "symbol missing".
...
> LinkLibraries pfengine : ../libpdf.a ;

That's Perforce/Jam syntax which is totally ignored by Boost.Build. You 
just need to add a <find-library>pdf


-- 
-- Grafik - Don't Assume Anything
-- Redshift Software, Inc. - http://redshift-software.com
-- rrivera/acm.org - grafik/redshift-software.com - 102708583/icq


From john at johnmeinel.com  Wed Nov 10 16:04:49 2004
From: john at johnmeinel.com (John Meinel)
Date: Wed, 10 Nov 2004 09:04:49 -0600
Subject: [C++-sig] Re: boost.Python/Jam with shared libraries
In-Reply-To: <loom.20041110T150821-13@post.gmane.org>
References: <loom.20041109T180856-543@post.gmane.org>	<41913C34.8090507@redshift-software.com>	<loom.20041109T231245-517@post.gmane.org>	<41915146.5050207@johnmeinel.com>
	<loom.20041110T150821-13@post.gmane.org>
Message-ID: <41922E11.5020704@johnmeinel.com>

Aaron Bentley wrote:
> John Meinel <john <at> johnmeinel.com> writes:
> 
> 
>># You might try this
>>
>>extension pfengine
>>   :# sources
>>     pfengine.cpp
>>   :
>>      <library-path>../.libs
>>      <find-library>panutil
>>      $(BOOST_PYTHON_V2_PROPERTIES)
>>;
>>
> 
> 
> Hi John,
> 
> Fancy meeting you here!
> 

I was surprised to see your name as well, one of the reasons I answered. :)

I don't really recognize your LinkLibraries command. (I've never used it.)

Does this not work:
   <find-library>pdf

I thought that worked for both dynamic and static libraries. The 
alternative in my mind would be
   <library-file>$(LIBPDF_PATH)/libpdf.a

You can either hard code $(LIBPDF_PATH), or what I prefer is to add the 
line:

local LIBPDF_PATH = 'path/to/libpdf' ;
                                     ^---- This space is important

I'm not sure why the space is required, but without it, you get problems.

I'm not a complete guru on Jamfiles. I've just fought my way through a 
lot of boost::python :).

I'm actually thinking about switching to Scons (www.scons.org), as soon 
as a friend of mine finishes porting his work to it, and can give me 
pointers how to move my stuff.

bjam does a lot of nice things, but it syntax is a little arcane to me. 
I feel like the planets have to align, while scons is at least python 
syntax. I haven't thoroughly evaluated it, though. But I know they both 
can do out of tree builds, and support automatic dependency checking.

John
=:->

> The eliminating-spaces route got rid of the errors, but find-library looks
> cleaner.  Thanks.
> 
> So I'm much further along.  Now I'm trying to link in a static file.  It appears
> to work, but then I get "symbol missing".
> 
> ---begin Jamfile---
> project-root ;
> 
> import python ;
> 
> # ----- getting_started2 -------
> 
> # Declare a Python extension called pfengine
> extension pfengine
> :  # sources
>    pfengine.cpp
>    #libraries
> :
>   <library-path>../.libs
>   <find-library>panutil
>   <find-library>pantemplate
>   <find-library>panpdfgen
>   <find-library>pancompile
>   <find-library>panconsult
>   <find-library>panelement
>   <find-library>panlang
>   <find-library>pancore
>   <find-library>panexport
>   <find-library>panreport
>   <find-library>pancgi
>   <find-library>chart
>   <find-library>ecommerce
>   <find-library>icuio
>   <find-library>icudata
>   <find-library>icutoolutil
>   <find-library>icule
>   <find-library>icui18n
> 
>   # requirements and dependencies for Boost.Python extensions
>   <template>@boost/libs/python/build/extension
>   ;
> 
> # Declare a test for the extension module
> boost-python-runtest test2
>     :  # Python test driver
>     test_getting_started2.py
>     # extension modules to use
>     <pyd>pfengine ;
> 
> LinkLibraries pfengine : ../libpdf.a ;
> # arch-tag: ae55dfed-2576-4667-953f-d7d8f777f7c4
> ---end jamfile---
> 
> ---attempt to use---
> 
>>>>import pfengine
> 
> Traceback (most recent call last):
>   File "<stdin>", line 1, in ?
> ImportError: /home/abentley/programming/.libs/libchart.so.0: undefined symbol:
> PDF_setrgbcolor_fill
> 
> ---
> 
> PDF_setrgbcolor_fill is provided by libpdf.a
> 
> Any help appreciated.
> 
> Aaron
> 
> 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 256 bytes
Desc: OpenPGP digital signature
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041110/db5e6f78/attachment.pgp>

From grafik.list at redshift-software.com  Wed Nov 10 16:21:54 2004
From: grafik.list at redshift-software.com (Rene Rivera)
Date: Wed, 10 Nov 2004 09:21:54 -0600
Subject: [C++-sig] Re: boost.Python/Jam with shared libraries
In-Reply-To: <41922E11.5020704@johnmeinel.com>
References: <loom.20041109T180856-543@post.gmane.org>	<41913C34.8090507@redshift-software.com>	<loom.20041109T231245-517@post.gmane.org>	<41915146.5050207@johnmeinel.com>	<loom.20041110T150821-13@post.gmane.org>
	<41922E11.5020704@johnmeinel.com>
Message-ID: <41923212.4090007@redshift-software.com>

John Meinel wrote:

> I thought that worked for both dynamic and static libraries. The 
> alternative in my mind would be
>   <library-file>$(LIBPDF_PATH)/libpdf.a
> 
> You can either hard code $(LIBPDF_PATH), or what I prefer is to add the 
> line:

You can also use a relative path to it:

	<library-file>../.libs/libpdf.a

> local LIBPDF_PATH = 'path/to/libpdf' ;
>                                     ^---- This space is important
> 
> I'm not sure why the space is required, but without it, you get problems.

Jam uses whitespace to separate all tokens. Hence the first problem 
Aaron had. Where the extra whitespace split what was supposed to be a 
single token.


-- 
-- Grafik - Don't Assume Anything
-- Redshift Software, Inc. - http://redshift-software.com
-- rrivera/acm.org - grafik/redshift-software.com - 102708583/icq


From abentley at panoramicfeedback.com  Wed Nov 10 16:37:02 2004
From: abentley at panoramicfeedback.com (Aaron Bentley)
Date: Wed, 10 Nov 2004 15:37:02 +0000 (UTC)
Subject: [C++-sig] Re: boost.Python/Jam with shared libraries
References: <loom.20041109T180856-543@post.gmane.org>	<41913C34.8090507@redshift-software.com>	<loom.20041109T231245-517@post.gmane.org>	<41915146.5050207@johnmeinel.com>	<loom.20041110T150821-13@post.gmane.org>
	<41922E11.5020704@johnmeinel.com>
	<41923212.4090007@redshift-software.com>
Message-ID: <loom.20041110T163029-768@post.gmane.org>

Rene Rivera <grafik.list <at> redshift-software.com> writes:

> 
> John Meinel wrote:
> 
> > I thought that worked for both dynamic and static libraries. The 
> > alternative in my mind would be
> >   <library-file>$(LIBPDF_PATH)/libpdf.a
> > 
> > You can either hard code $(LIBPDF_PATH), or what I prefer is to add the 
> > line:
> 
> You can also use a relative path to it:
> 
> 	<library-file>../.libs/libpdf.a
With <library-file> or with <find-library>, I get the same results.  Jam appears
to succeed:

---
abentley at troll:~/programming/python$ ./go.sh 
...found 1290 targets...
...updating 4 targets...
gcc-C++-action bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.o
gcc-Link-action bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.so
Chmod1 bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.so
gcc-C++-action bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.o
gcc-Link-action bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.so
Chmod1 bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.so
...updated 4 targets...
---

But I can't import it:
---
>>> import pfengine
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
ImportError: /home/abentley/programming/.libs/libchart.so.0: undefined symbol:
PDF_setrgbcolor_fill
---

I suppose it's possible that the problem is caused by trying to link a static
library into a shared library, but you'd think that would be possible.  I guess
I could try to convert libpdf.a into a shared library.  Who'd have thought I'd
have more trouble linking to C than to C++?

Aaron



From john at johnmeinel.com  Wed Nov 10 16:43:08 2004
From: john at johnmeinel.com (John Meinel)
Date: Wed, 10 Nov 2004 09:43:08 -0600
Subject: [C++-sig] Re: boost.Python/Jam with shared libraries
In-Reply-To: <loom.20041110T163029-768@post.gmane.org>
References: <loom.20041109T180856-543@post.gmane.org>	<41913C34.8090507@redshift-software.com>	<loom.20041109T231245-517@post.gmane.org>	<41915146.5050207@johnmeinel.com>	<loom.20041110T150821-13@post.gmane.org>	<41922E11.5020704@johnmeinel.com>	<41923212.4090007@redshift-software.com>
	<loom.20041110T163029-768@post.gmane.org>
Message-ID: <4192370C.2020901@johnmeinel.com>

Aaron Bentley wrote:
> Rene Rivera <grafik.list <at> redshift-software.com> writes:
> 
> 
>>John Meinel wrote:
>>

[...]

> ---
> abentley at troll:~/programming/python$ ./go.sh 
> ...found 1290 targets...
> ...updating 4 targets...
> gcc-C++-action bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.o
> gcc-Link-action bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.so
> Chmod1 bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.so
> gcc-C++-action bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.o
> gcc-Link-action bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.so
> Chmod1 bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.so
> ...updated 4 targets...
> ---
> 
> But I can't import it:
> ---
> 
>>>>import pfengine
> 
> Traceback (most recent call last):
>   File "<stdin>", line 1, in ?
> ImportError: /home/abentley/programming/.libs/libchart.so.0: undefined symbol:
> PDF_setrgbcolor_fill
> ---
> 
> I suppose it's possible that the problem is caused by trying to link a static
> library into a shared library, but you'd think that would be possible.  I guess
> I could try to convert libpdf.a into a shared library.  Who'd have thought I'd
> have more trouble linking to C than to C++?
> 
> Aaron
> 

You might want to try the "don't do it but tell me what you want to do." 
flag.

bjam -n

So that you can check the link line and make sure that it is trying to 
link with libpdf.a.

Just delete your pfengine.so.* file so that it forces a rebuild.
I think it is okay to have a shared library link a static one. Though 
I've noticed that a static library will not link with another static lib.

Example:

lib other
   : otherlib.cpp
;
lib mylib
   : mysource.cpp
     <lib>other
;

I end up with 2 libraries, but mylib does not contain the objects that 
exist in other. So that if later I try:

extension myext
    : myext.cpp
      <lib>mylib
      <lib>other # I must have this line,
		# even though it is supposedly in mylib
;

This easily could not be a bjam thing, and is simply how libraries work. 
And it might also depend on platform.

John
=:->
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 256 bytes
Desc: OpenPGP digital signature
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041110/437d7a72/attachment.pgp>

From dave at boost-consulting.com  Wed Nov 10 17:13:56 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 10 Nov 2004 11:13:56 -0500
Subject: [C++-sig] Re: boost.Python/Jam with shared libraries
References: <loom.20041109T180856-543@post.gmane.org>
Message-ID: <umzxpsmuz.fsf@boost-consulting.com>

Aaron Bentley <abentley at panoramicfeedback.com> writes:

> I'm trying to add a python interface to some of the functionality in an existing
> app.  I've managed to get a test version to compile, but the result doesn't work
> because shared libraries (both commonly-available and app-specific) in are
> required for our app.

Is http://tinyurl.com/5q9xz of any help to you?

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From abentley at panoramicfeedback.com  Wed Nov 10 17:37:56 2004
From: abentley at panoramicfeedback.com (Aaron Bentley)
Date: Wed, 10 Nov 2004 16:37:56 +0000 (UTC)
Subject: [C++-sig] Re: boost.Python/Jam with shared libraries
References: <loom.20041109T180856-543@post.gmane.org>	<41913C34.8090507@redshift-software.com>	<loom.20041109T231245-517@post.gmane.org>	<41915146.5050207@johnmeinel.com>	<loom.20041110T150821-13@post.gmane.org>	<41922E11.5020704@johnmeinel.com>	<41923212.4090007@redshift-software.com>
	<loom.20041110T163029-768@post.gmane.org>
	<4192370C.2020901@johnmeinel.com>
Message-ID: <loom.20041110T171819-706@post.gmane.org>

John Meinel <john <at> johnmeinel.com> writes:

> You might want to try the "don't do it but tell me what you want to do." 
> flag.
> 
> bjam -n

Here's the meat of the output (with library-file):

"bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.so.1.31.0"

-L"/home/abentley/devel/.libs" -L"/home/abentley/lib/python2.3/config"

"bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.o"

"/home/abentley/devel/libpdf.a"  "/home/abentley/devel/libpdf.a"   -lpanutil

-lpantemplate -lpanpdfgen -lpancompile -lpanconsult -lpanelement -lpanlang

-lpancore -lpanexport -lpanreport -lpancgi -lchart -lecommerce -licuio -licudata

-licutoolutil -licule -licui18n  -Wl,-rpath-link,.  

-Wl,-soname,pfengine.so.1.31.0 




From john at arbash-meinel.com  Wed Nov 10 17:50:26 2004
From: john at arbash-meinel.com (John Meinel)
Date: Wed, 10 Nov 2004 10:50:26 -0600
Subject: [C++-sig] Re: boost.Python/Jam with shared libraries
In-Reply-To: <loom.20041110T171819-706@post.gmane.org>
References: <loom.20041109T180856-543@post.gmane.org>	<41913C34.8090507@redshift-software.com>	<loom.20041109T231245-517@post.gmane.org>	<41915146.5050207@johnmeinel.com>	<loom.20041110T150821-13@post.gmane.org>	<41922E11.5020704@johnmeinel.com>	<41923212.4090007@redshift-software.com>	<loom.20041110T163029-768@post.gmane.org>	<4192370C.2020901@johnmeinel.com>
	<loom.20041110T171819-706@post.gmane.org>
Message-ID: <419246D2.7060101@arbash-meinel.com>

Aaron Bentley wrote:
> John Meinel <john <at> johnmeinel.com> writes:
> 
> 
>>You might want to try the "don't do it but tell me what you want to do." 
>>flag.
>>
>>bjam -n
> 
> 
> Here's the meat of the output (with library-file):
> 
> "bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.so.1.31.0"
> 
> -L"/home/abentley/devel/.libs" -L"/home/abentley/lib/python2.3/config"
> 
> "bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.o"
> 
> "/home/abentley/devel/libpdf.a"  "/home/abentley/devel/libpdf.a"   -lpanutil
> 
> -lpantemplate -lpanpdfgen -lpancompile -lpanconsult -lpanelement -lpanlang
> 
> -lpancore -lpanexport -lpanreport -lpancgi -lchart -lecommerce -licuio -licudata
> 
> -licutoolutil -licule -licui18n  -Wl,-rpath-link,.  
> 
> -Wl,-soname,pfengine.so.1.31.0 
> 

I would have thought that you would want it to say:

-L"/home/abentley/devel" -lpdf

Where is libpdf.a anyway? Because you are including ~/devel/.libs, but 
you seem to have the actual library in ~/devel/libpdf.a

I would say the lines:

   <library-path>/home/abentley/devel
   <find-library>pdf

Would do what I'm mentioning. I don't really know how <library-file> 
works (I don't use it myself.)

I think suppling the full path to the library file should work, though. 
I'm not sure why it is supplying it twice. Unless it's just trying to 
get the linking order correct. I believe there have been bugs where 
doing "gcc file.cpp -llib1 -llib2" works, but "gcc file.cpp -llib2 
-llib1" does not (if one depends on the other.)

I'm not really sure. All of my static libs are managed by Jamfiles right 
now, so I haven't learned how to access external static libraries. I 
thought it would be the same as external dynamic ones.

John
=:->
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 256 bytes
Desc: OpenPGP digital signature
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041110/8e800001/attachment.pgp>

From grafik.list at redshift-software.com  Wed Nov 10 17:53:12 2004
From: grafik.list at redshift-software.com (Rene Rivera)
Date: Wed, 10 Nov 2004 10:53:12 -0600
Subject: [C++-sig] Re: boost.Python/Jam with shared libraries
In-Reply-To: <loom.20041110T171819-706@post.gmane.org>
References: <loom.20041109T180856-543@post.gmane.org>	<41913C34.8090507@redshift-software.com>	<loom.20041109T231245-517@post.gmane.org>	<41915146.5050207@johnmeinel.com>	<loom.20041110T150821-13@post.gmane.org>	<41922E11.5020704@johnmeinel.com>	<41923212.4090007@redshift-software.com>	<loom.20041110T163029-768@post.gmane.org>	<4192370C.2020901@johnmeinel.com>
	<loom.20041110T171819-706@post.gmane.org>
Message-ID: <41924778.6080700@redshift-software.com>

Aaron Bentley wrote:

> John Meinel <john <at> johnmeinel.com> writes:
> 
> 
>>You might want to try the "don't do it but tell me what you want to do." 
>>flag.
>>
>>bjam -n
> 
> 
> Here's the meat of the output (with library-file):
> 
> "bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.so.1.31.0"
> -L"/home/abentley/devel/.libs" -L"/home/abentley/lib/python2.3/config"
> "bin/python/pfengine.so/gcc/debug/shared-linkable-true/pfengine.o"
> "/home/abentley/devel/libpdf.a"  "/home/abentley/devel/libpdf.a"   -lpanutil
> -lpantemplate -lpanpdfgen -lpancompile -lpanconsult -lpanelement -lpanlang
> -lpancore -lpanexport -lpanreport -lpancgi -lchart -lecommerce -licuio -licudata
> -licutoolutil -licule -licui18n  -Wl,-rpath-link,.  
> -Wl,-soname,pfengine.so.1.31.0 

Looks correct, that is libpdf.a is in there. I would suspect that the 
problem is that libchart was built to link against the shared version of 
libpdf. You can verify what it needs by using the "ldd" command, 
http://www.die.net/doc/linux/man/man1/ldd.1.html -- If that's the case 
then you'll have to find/build that shared libpdf dependency.


-- 
-- Grafik - Don't Assume Anything
-- Redshift Software, Inc. - http://redshift-software.com
-- rrivera/acm.org - grafik/redshift-software.com - 102708583/icq


From nicodemus at esss.com.br  Thu Nov 11 00:18:51 2004
From: nicodemus at esss.com.br (Nicodemus)
Date: Wed, 10 Nov 2004 20:18:51 -0300
Subject: [C++-sig] pyste instantiating templates
In-Reply-To: <m3y8hbuekk.fsf@peds-pc311.bsd.uchicago.edu>
References: <m3y8hbuekk.fsf@peds-pc311.bsd.uchicago.edu>
Message-ID: <4192A1DB.3010808@esss.com.br>

John Hunter wrote:

>I have a template class in someclass.h that is instantiated over
>sometype.h.  someclass.h does not #include sometype.h.
>
>If I try the standard way of instantiating templates in pyste
>
>  SomeTemplate = Template("SomeClass", "someclass.h")
>  TemplateSometype = SomeTemplate("Sometype", "template_sometype")
>
>I get a gccxml error because SomeType is not known in someclass.h.
>
>It would be nice to be able to tell pyste in the tmp*.h files it
>generates to pass to gccxml, to include some additional headers.  Is
>this possible?  None of Include, declaration_code or module_code seem
>to fit the bill here, since I want to include code into the file that
>I am actually wrapping.
>
>I worked around it by creating a 
>
>instantiate_someclass.h
>
>  #include sometype.h
>  #include someclass.h
>
>and then doing
>
>  SomeTemplate = Template("SomeClass", "instantiate_someclass.h")
>  TemplateSometype = SomeTemplate("Sometype", "template_sometype")
>  
>

Hi Jonh,

It would be nice to have this feature, but sincerely this won't happen 
anytime soon. I have very little free time right now, and can't dedicate 
anytime to Pyste. 8/
Anyway, I plan a re-write of the whole system next year, so this feature 
has good chances of being avaiable in the new version.

Regarding your other questions, I will answer them over the weekend... 
sorry about the delay!

Regards,
Bruno.


From abentley at panoramicfeedback.com  Thu Nov 11 22:53:05 2004
From: abentley at panoramicfeedback.com (Aaron Bentley)
Date: Thu, 11 Nov 2004 21:53:05 +0000 (UTC)
Subject: [C++-sig] Re: boost.Python/Jam with shared libraries [solved]
References: <loom.20041109T180856-543@post.gmane.org>	<41913C34.8090507@redshift-software.com>	<loom.20041109T231245-517@post.gmane.org>	<41915146.5050207@johnmeinel.com>	<loom.20041110T150821-13@post.gmane.org>	<41922E11.5020704@johnmeinel.com>	<41923212.4090007@redshift-software.com>	<loom.20041110T163029-768@post.gmane.org>	<4192370C.2020901@johnmeinel.com>
	<loom.20041110T171819-706@post.gmane.org>
	<41924778.6080700@redshift-software.com>
Message-ID: <loom.20041111T223731-559@post.gmane.org>

Rene Rivera <grafik.list <at> redshift-software.com> writes:

Actually, I'm pretty sure there was never a shared version of libpdf.  Anyhow,
what I ended up doing was linking libpdf into another shared library.

Strangely, after all this, I found I needed to add a find-library for
boot_python itself.

Stuff is working now, so thanks for the help.

Now I'm trying to subclass abstract C++ classes in Python...

Aaron



From ndbecker2 at verizon.net  Fri Nov 12 16:56:18 2004
From: ndbecker2 at verizon.net (Neal D. Becker)
Date: Fri, 12 Nov 2004 10:56:18 -0500
Subject: [C++-sig] add_property + return value policy
Message-ID: <cn2mf2$vht$1@sea.gmane.org>

Is it possible to .add_property with a member function that needs a
return_value_policy (e.g, copy_const_reference)?



From gathmann at cenix-bioscience.com  Fri Nov 12 17:10:01 2004
From: gathmann at cenix-bioscience.com (F. Oliver Gathmann)
Date: Fri, 12 Nov 2004 17:10:01 +0100
Subject: [C++-sig] add_property + return value policy
In-Reply-To: <cn2mf2$vht$1@sea.gmane.org>
References: <cn2mf2$vht$1@sea.gmane.org>
Message-ID: <4194E059.30809@cenix-bioscience.com>

Neal D. Becker wrote:

>Is it possible to .add_property with a member function that needs a
>return_value_policy (e.g, copy_const_reference)?
>  
>
You can try make_function - something along the lines of

.add_property("myprop", 
              make_function(&getter, return_value_policy<copy_const_reference>()),
              &setter)


HTH,

Oliver

-- 
--------------------------------------------------------------------
F. Oliver Gathmann, Ph.D.
Director IT Unit
Cenix BioScience GmbH
Tatzberg 47                       phone: +49 (351) 4173-136
D-01307 Dresden, Germany          fax:   +49 (351) 4173-198

fingerprint: 8E0E 9A64 A07E 0D1A D302  34C2 421A AE9F 4E13 A009
public key: http://www.cenix-bioscience.com/public_keys/gathmann.gpg
--------------------------------------------------------------------




From Paul_Kunz at slac.stanford.edu  Sat Nov 13 10:53:18 2004
From: Paul_Kunz at slac.stanford.edu (Paul F. Kunz)
Date: Sat, 13 Nov 2004 01:53:18 -0800
Subject: [C++-sig] Deriving Python function from C++ base class with pure
 virtual function
References: <cn2mf2$vht$1@sea.gmane.org> <4194E059.30809@cenix-bioscience.com>
Message-ID: <200411130953.iAD9rI323669@libra3.slac.stanford.edu>

   I'm deriving a Python class from a C++ class with both concrete and
pure virtual functions.   I written a FunctionWrap C++ class that
implements the pure virtual function, operator() (double) by invoking
the call_method.   Then my Python class looks like...

from hippo import AbsFunction
class Linear ( AbsFunction ) :
    def __init__ ( self ) :
        AbsFunction.__init__(self)
        self.initialize ()
    
    def initialize ( self ) :
        self.setName ( 'linear' )
        self.setParmNames ( [ 'Intercept', 'Slope' ] )
        self.setParameters ( [ intercept, slope ] )
    
    def valueAt ( self, x ) :
        parms = self.getParameters ()
        
        return x * parms[1] + parms[0]

where AbsFunction is the Python name for the C++ base class and all
the set and get functions are implemented in the C++ base class.  The
FunctionWrap::operator ()(double) is implemented to call the Python
valueAt method.  All this works within a Python script.

   However, I want to also call operator()(double) from C++ code in a
different thread from the Python main thread.   This other thread is
started with Qt's QThread class.   When I do this I get to
FunctionWrap::operator()(double) and then seg fault with the following
back trace in gdb...

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1117711152 (LWP 4144)]
0x080fc20c in PyFrame_New (tstate=0x0, code=0x4018dce0,
globals=0x4015d79c, 
    locals=0x0) at Objects/frameobject.c:540
(gdb) bt
#0  0x080fc20c in PyFrame_New (tstate=0x0, code=0x4018dce0, 
    globals=0x4015d79c, locals=0x0) at Objects/frameobject.c:540
#1  0x080b7353 in PyEval_EvalCodeEx (co=0x4018dce0,
globals=0x4015d79c, 
    locals=0x4018dce0, args=0x406396b8, argcount=2, kws=0x0,
kwcount=0, 
    defs=0x0, defcount=0, closure=0x0) at Python/ceval.c:2455
#2  0x080fd955 in function_call (func=0x405ed48c, arg=0x406396ac,
kw=0x0)
    at Objects/funcobject.c:504
#3  0x0805b8c7 in PyObject_Call (func=0x0, arg=0x4018dce0,
kw=0x4018dce0)
    at Objects/abstract.c:1755
#4  0x08064aae in instancemethod_call (func=0x4018dce0,
arg=0x406396ac, 
    kw=0x4018dce0) at Objects/classobject.c:2433
#5  0x0805b8c7 in PyObject_Call (func=0x0, arg=0x4018dce0,
kw=0x4018dce0)
    at Objects/abstract.c:1755
#6  0x080b103b in PyEval_CallObjectWithKeywords (func=0x4018dce0, 
    arg=0x4018ce4c, kw=0x405ec5f4) at Python/ceval.c:3346
#7  0x080d89e3 in PyEval_CallMethod (obj=0x4018dce0, 
    methodname=0x4018dce0 "\001", format=0x4018dce0 "\001")
    at Python/modsupport.c:506
#8  0x407e5708 in boost::python::call_method<double, double> (
    self=0x4018dce0, name=0x4018dce0 "\001", a0=@0x4018dce0) at
handle.hpp:64
#9  0x407e4de7 in FunctionWrap::operator() (this=0x4018dce0, x=0)
    at ../../hippodraw/python/FunctionWrap.cxx:111

   Apparently I have threading problems but don't know how to
procede.   I tried wrapping call_method with
Py_ALLOW_THREADS_BEGIN/END, but that seg faults immediately.   

   Should I start the second thread with Python's own thread class?
Is there some Python C API functions I could call to read the Python
object to be called from this second thread?   Any suggestions /
advise would be appreciated.



From mike.spam.filter at day8.com.au  Sun Nov 14 11:31:09 2004
From: mike.spam.filter at day8.com.au (Mike Thompson)
Date: Sun, 14 Nov 2004 21:31:09 +1100
Subject: [C++-sig] Re: Call Policy Question
In-Reply-To: <ufz3piuhe.fsf@boost-consulting.com>
References: <cmcl36$ltv$1@sea.gmane.org> <ufz3piuhe.fsf@boost-consulting.com>
Message-ID: <cn7c5j$5s2$1@sea.gmane.org>

David Abrahams wrote:
> Mike Thompson <mike.spam.filter at day8.com.au> writes:
> 
> 
>>I have a member function which returns an object and I'm trying to
>>define a call policy for that member function which will:
>>    1.  cause the returned object to be managed:
>>            return_value_policy<manage_new_object>()
>>        its a new object and needs to be managed.
>>
>>    2.   register that the returned object holds a
>>         reference to the originating object (but not visa-versa)
>>
>>but I can't work out how to combine the two.
>>
>>I need 1. above because the returned object is new and needs to be managed.
>>
>>I need 2. above because this returned object has references BACK
>>to the object who's member function supplies it.
>>
>>So the returned object can be destroyed without the original being
>>effected, but destruction of the original MUST wait until the returned
>>object is gone.
> 
> 
>   return_value_policy<
>         manage_new_object
>       , with_custodian_and_ward<0, 1>
>   >
> 
> should work.
> 

Ahh. Of course. Thanks.

But, just to be clear, I suspect you meant to use 
'with_custodian_and_ward_postcall' ?


Cheers,
Mike



From dave at boost-consulting.com  Sun Nov 14 14:10:23 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Sun, 14 Nov 2004 08:10:23 -0500
Subject: [C++-sig] Re: Call Policy Question
References: <cmcl36$ltv$1@sea.gmane.org> <ufz3piuhe.fsf@boost-consulting.com>
	<cn7c5j$5s2$1@sea.gmane.org>
Message-ID: <ufz3co9ts.fsf@boost-consulting.com>

Mike Thompson <mike.spam.filter at day8.com.au> writes:

> David Abrahams wrote:
>> Mike Thompson <mike.spam.filter at day8.com.au> writes:
>> 
>>>I have a member function which returns an object and I'm trying to
>>>define a call policy for that member function which will:
>>>    1.  cause the returned object to be managed:
>>>            return_value_policy<manage_new_object>()
>>>        its a new object and needs to be managed.
>>>
>>>    2.   register that the returned object holds a
>>>         reference to the originating object (but not visa-versa)
>>>
>>>but I can't work out how to combine the two.
>>>
>>>I need 1. above because the returned object is new and needs to be managed.
>>>
>>>I need 2. above because this returned object has references BACK
>>>to the object who's member function supplies it.
>>>
>>>So the returned object can be destroyed without the original being
>>>effected, but destruction of the original MUST wait until the returned
>>>object is gone.
>>   return_value_policy<
>>         manage_new_object
>>       , with_custodian_and_ward<0, 1>
>>   >
>> should work.
>> 
>
> Ahh. Of course. Thanks.
>
> But, just to be clear, I suspect you meant to use
> with_custodian_and_ward_postcall' ?

Right.

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From s_sourceforge at nedprod.com  Sun Nov 14 18:09:57 2004
From: s_sourceforge at nedprod.com (Niall Douglas)
Date: Sun, 14 Nov 2004 17:09:57 -0000
Subject: [C++-sig] Deriving Python function from C++ base class with pure
	virtual function
In-Reply-To: <200411130953.iAD9rI323669@libra3.slac.stanford.edu>
Message-ID: <41979165.9715.5336B62@localhost>

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 13 Nov 2004 at 1:53, Paul F. Kunz wrote:

>    However, I want to also call operator()(double) from C++ code in a
> different thread from the Python main thread.   This other thread is
> started with Qt's QThread class.   When I do this I get to
> FunctionWrap::operator()(double) and then seg fault with the following
> back trace in gdb...
> [snip]
>    Apparently I have threading problems but don't know how to
> procede.   I tried wrapping call_method with
> Py_ALLOW_THREADS_BEGIN/END, but that seg faults immediately.   
> 
>    Should I start the second thread with Python's own thread class? Is
> there some Python C API functions I could call to read the Python
> object to be called from this second thread?   Any suggestions /
> advise would be appreciated.

Before entering python, you must ensure that python's thread state 
has been set for the current thread. This implies holding python's 
Global Interpreter Lock. If you try entering python without setting 
the thread state, you'll get a warning on stderr and a segfault most 
probably.

If however you want to have multiple threads working with python 
concurrently, that is a whole new bag of worms. There is a solution 
but it involves patching BPL to release and grab the python GIL every 
time it enters and exits C/C++ code. Do say if you want to do this, 
but be warned it's not trivial.

Cheers,
Niall





-----BEGIN PGP SIGNATURE-----
Version: idw's PGP-Frontend 4.9.6.1 / 9-2003 + PGP 8.0.2

iQA/AwUBQZeRZcEcvDLFGKbPEQIcHwCg1hpD6OKp34ea0ow2GsU1CMg7uiUAoO5a
fk2O1bpfwGhTGv6/m9h9ATd0
=qgYg
-----END PGP SIGNATURE-----


From john at johnmeinel.com  Sun Nov 14 18:52:04 2004
From: john at johnmeinel.com (John Meinel)
Date: Sun, 14 Nov 2004 11:52:04 -0600
Subject: [C++-sig] Deriving Python function from C++ base class with pure
	virtual function
In-Reply-To: <41979165.9715.5336B62@localhost>
References: <41979165.9715.5336B62@localhost>
Message-ID: <41979B44.3040001@johnmeinel.com>

Niall Douglas wrote:
> On 13 Nov 2004 at 1:53, Paul F. Kunz wrote:
> 
> 

[...]

>
>>>   Should I start the second thread with Python's own thread class? Is
>>>there some Python C API functions I could call to read the Python
>>>object to be called from this second thread?   Any suggestions /
>>>advise would be appreciated.
> 
> 
> Before entering python, you must ensure that python's thread state 
> has been set for the current thread. This implies holding python's 
> Global Interpreter Lock. If you try entering python without setting 
> the thread state, you'll get a warning on stderr and a segfault most 
> probably.
> 
> If however you want to have multiple threads working with python 
> concurrently, that is a whole new bag of worms. There is a solution 
> but it involves patching BPL to release and grab the python GIL every 
> time it enters and exits C/C++ code. Do say if you want to do this, 
> but be warned it's not trivial.
> 
> Cheers,
> Niall
> 
> 

I've done some work with multiple threads in python. Here is what I did.

First, I have C++ functions that run for a long time doing analysis, and 
some that are very quick. I don't bother threading the quick ones, just 
the long running ones.

I have a Callback class which handles the reporting of how things are 
going. The C++ functions don't realize that they are being called from 
python, because their only interaction with the outside world is through 
the Callback object. The Callback is a C++ object, which I inherit from 
in python.

All of my threads are started from python, which means that they already 
have thread state associated with them, and when the call is made to the 
C++ code, they already have the GIL (or they wouldn't be running.)

I created a class which when instantiated replaces the Callback object 
with one of it's own, such that whenever a call is made, it grabs the 
GIL, and then places the call to the original Callback, and when that 
returns, it releases the GIL.

So what does this all look like.

The C++ side:

class Callback
{
public:
    int send_message(const std::string &msg) = 0;
};

class Worker
{
    Callback *cb_;
public:

    Callback *setCallback(Callback *cb)
    {
      Callback oldcb = cb_;
      cb_ = cb;
      return oldcb;
    }

    int doSomethingFast(int arg);
    int doSomethingLong(int arg);

};

class PyCallback
{
   Callback *other_;
   PyThreadState *state_;
   Worker *worker_;
public:
   PyCallback ()
     : state_(0), other_(0), worker_(0)
   { }
   ~PyCallback ()
   {
     releaseCallback();
   }

   void grabCallback(Worker& w)
   {
     other_ = w.setCallback(this);
     worker_ = &w;
     state_ = PyEval_SaveThread();
   }

   void releaseCallback()
   {
     if(other_) {
       PyEval_AcquireThread(state_);
       worker_->setCallback(other_);
       worker_ = 0;
       state_ = 0;
       other_ = 0;
     }
   }

   int send_message(const std::string &msg)
   {
     int response = 0;
     if (other_) {
       PyEval_AcquireThread(state_);
       response = other_->send_message(msg);
       PyEval_ReleaseThread(state_);
     }
     return response;
   }

};

And now the boost::python wrapper

static int workerDoSomething(Worker& w, int arg)
{
   PyCallback pcb;
   pcb.getCallback(w);
   return w.doSomething(arg);
}

void exportWorker()
{
   class_<Worker>("Worker")
     .def("doSomethingLong", &workerDoSomething)
     .def("doSomethingFast", &Worker::doSomethingFast)
   ;
}

Now, I happen to have a structure where everything that needs work done 
inherits from a "Worker" class. So they all have the setCallback() 
function, and they all use the Callback to talk about what is going on. 
Mine is also a little bit more complicated, as I have another parent 
class, which lets me write more like: (following RAII rules)

static int workerDoSomething(Worker& w, int arg)
{
   PyWrap pw(w);
   return w.doSomething(arg);
}

Also, I don't use raw pointers, everything is a boost::shared_ptr<>, but 
I figured raw pointers is a little bit easier to read (and less to type).

I could see easily turning PyCallback into a template function, such 
that Worker is not defined, and you just need a function like 
setCallback() to exist.

But this is basically what I am doing. Now probably boost::python could 
be rewritten such that it automatically does most of this stuff. (it 
already is creating function wrappers, it could probably just add some 
more threading code into them.)

But this is a way to do it without changing any of boost::python.

It depends on your framework, though.

John
=:->
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 256 bytes
Desc: OpenPGP digital signature
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041114/6f734119/attachment.pgp>

From jbrandmeyer at earthlink.net  Sun Nov 14 23:35:14 2004
From: jbrandmeyer at earthlink.net (Jonathan Brandmeyer)
Date: Sun, 14 Nov 2004 17:35:14 -0500
Subject: [C++-sig] Deriving Python function from C++ base class with
	pure virtual function
In-Reply-To: <200411130953.iAD9rI323669@libra3.slac.stanford.edu>
References: <cn2mf2$vht$1@sea.gmane.org> <4194E059.30809@cenix-bioscience.com>
	<200411130953.iAD9rI323669@libra3.slac.stanford.edu>
Message-ID: <1100471714.18970.21.camel@illuvatar>

On Sat, 2004-11-13 at 01:53 -0800, Paul F. Kunz wrote:
>    I'm deriving a Python class from a C++ class with both concrete and
> pure virtual functions.   I written a FunctionWrap C++ class that
> implements the pure virtual function, operator() (double) by invoking
> the call_method.   Then my Python class looks like...
> 
> from hippo import AbsFunction
> class Linear ( AbsFunction ) :
>     def __init__ ( self ) :
>         AbsFunction.__init__(self)
>         self.initialize ()
>     
>     def initialize ( self ) :
>         self.setName ( 'linear' )
>         self.setParmNames ( [ 'Intercept', 'Slope' ] )
>         self.setParameters ( [ intercept, slope ] )
>     
>     def valueAt ( self, x ) :
>         parms = self.getParameters ()
>         
>         return x * parms[1] + parms[0]
> 
> where AbsFunction is the Python name for the C++ base class and all
> the set and get functions are implemented in the C++ base class.  The
> FunctionWrap::operator ()(double) is implemented to call the Python
> valueAt method.  All this works within a Python script.
> 
>    However, I want to also call operator()(double) from C++ code in a
> different thread from the Python main thread.   This other thread is
> started with Qt's QThread class.   When I do this I get to
> FunctionWrap::operator()(double) and then seg fault with the following
> back trace in gdb...

See the PEP 311 interface PyGILState_Ensure()/PyGILState_Release(),
which you will use to wrap around the callback function in your C++
code.  This interface solves the thread bootstrap problem.  Furthermore,
the lock is recursive, so locking the GIL is safe when you already have
the lock.

HTH,
-Jonathan Brandmeyer



From Stefan.Quandt at bardenheuer.de  Tue Nov 16 16:29:52 2004
From: Stefan.Quandt at bardenheuer.de (Stefan Quandt)
Date: Tue, 16 Nov 2004 16:29:52 +0100
Subject: [C++-sig] BUILD
Message-ID: <cnd6dg$oqu$1@sea.gmane.org>

I try to build an extension with python-debug on Win32.

I compiled a debug version of the Python interpreter from source and
adjusted PYTHON_ROOT in my Jamrules to it's location.

Then I try to build my extendions with
  bjam -sBUILD=python-debug
When bjam tries to link boost_python.dll it complains not finding
"python23.lib".

As far as I know it should be "python23_d.lib".
Or am I missing something?

Thanks
  Stefan



From dave at boost-consulting.com  Tue Nov 16 17:25:23 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 16 Nov 2004 11:25:23 -0500
Subject: [C++-sig] Re: BUILD
References: <cnd6dg$oqu$1@sea.gmane.org>
Message-ID: <u7joliwwc.fsf@boost-consulting.com>

Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:

> I try to build an extension with python-debug on Win32.
>
> I compiled a debug version of the Python interpreter from source and
> adjusted PYTHON_ROOT in my Jamrules to it's location.
>
> Then I try to build my extendions with
>   bjam -sBUILD=python-debug
> When bjam tries to link boost_python.dll it complains not finding
> "python23.lib".
>
> As far as I know it should be "python23_d.lib".
> Or am I missing something?

   bjam -sBUILD=debug-python
                ^^^^^^^^^^^^

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



From haller at ableton.com  Tue Nov 16 21:47:46 2004
From: haller at ableton.com (Stefan Haller)
Date: Tue, 16 Nov 2004 21:47:46 +0100
Subject: [C++-sig] Custom WeakPtrs and "None"
Message-ID: <1gncrqj.51k2my18f3gmgM%haller@ableton.com>

I'm not sure if this is more of a Python question than a Boost.Python
question:

We have our own intrusive refcount mechanism in the classes that we want
to wrap with Boost.Python.  We have weak pointers to these classes that
go NULL when the pointee is deleted.

Because of the intrusive refcounts, we can easily pass raw pointers
instead of shared pointers as function arguments and return values in
C++; I find this more convenient than having to pass shared pointers all
the time.  The caller of a function-returning-a-raw-ptr is free to
decide if he wants to keep the returned object in a shared pointer or a
weak pointer.

Now, when exposing these functions to python, we want to make sure that
all such returned objects are held in weak pointers on the python side.
(We do all our application logic in C++, and merely want to add a
scripting layer in python that is allowed to manipulate the data model
in certain restricted ways; we are not going to create or delete objects
from python.)  This is easy to do by writing thin wrapper functions that
simply return a weak ptr constructed from the raw pointer, and ".def"ing
those instead of the real functions; register_ptr_to_python does the
rest, which is very nice.

The only problem is that when the object goes away, the "weak pointer"
on the Python side isn't set to "None".  Of course, the wrapped C++ weak
pointer is set to NULL, but the python variable isn't aware of that
unless you try to access properties or call methods on it, in which case
it throws an exception.  It would be very nice if a comparison with None
would return true for a weak pointer that has been set to NULL.

Is there a way to achieve that?  Does it help if I post some example
code to illustrate that lengthy prose above?


-- 
Stefan Haller
Ableton
http://www.ableton.com/


From nicodemus at esss.com.br  Wed Nov 17 00:22:47 2004
From: nicodemus at esss.com.br (Nicodemus)
Date: Tue, 16 Nov 2004 20:22:47 -0300
Subject: [C++-sig] Re: pyste and template functions
In-Reply-To: <m3vfcf10t9.fsf@peds-pc311.bsd.uchicago.edu>
References: <m3zn1sl4cm.fsf@peds-pc311.bsd.uchicago.edu>	<7465b6170411082053b0531ab@mail.gmail.com>	<uvfcfxcw8.fsf@boost-consulting.com>
	<m3vfcf10t9.fsf@peds-pc311.bsd.uchicago.edu>
Message-ID: <419A8BC7.20802@esss.com.br>


John Hunter wrote:

>I assumed thusly that pyste either did not handle template functions
>or at least I didn't know how to handle them in pyste, so I reverted
>to the module_code I mentioned above.
>

Hi Jonh,

Your guess is correct, Pyste doesn't export template functions... you 
have to do it by hand, like you did.

Regards,
Bruno.


From nicodemus at esss.com.br  Wed Nov 17 00:22:50 2004
From: nicodemus at esss.com.br (Nicodemus)
Date: Tue, 16 Nov 2004 20:22:50 -0300
Subject: [C++-sig] typedef'd unnamed enum / pyste
In-Reply-To: <m3y8hb117w.fsf@peds-pc311.bsd.uchicago.edu>
References: <m3hdo0mkde.fsf@peds-pc311.bsd.uchicago.edu>	<7465b61704110820505e09c961@mail.gmail.com>
	<m3y8hb117w.fsf@peds-pc311.bsd.uchicago.edu>
Message-ID: <419A8BCA.3040001@esss.com.br>

Hi Jonh,

Sorry about the delay: as usual, it's hard to find time to work on pyste 
or answering questions on the list. Sorry again. :(

The only way currently in Pyste to export unnamed enums is using 
AllFromHeader. Althought it is deprecated, it should work just fine for 
your case. The code it generates is different thought:

    scope().attr("cover_none") = (int)cover_none;
    scope().attr("cover_full") = (int)cover_full;
    scope().attr("cover_shift") = (int)cover_shift;
    scope().attr("cover_size") = (int)cover_size;
    scope().attr("cover_mask") = (int)cover_mask;

If you want to actually have an enum named "cover", you will have to 
export it just like you did.

Regards,
Bruno.

John Hunter wrote:

>>>>>>"Roman" == Roman Yakovenko <roman.yakovenko at gmail.com> writes:
>>>>>>
>
>    >> 
>    >> typedef unsigned char cover_type; //----cover_type
>
>    Roman> Pyste is right cover_type is typedef to unsigned char. You
>    Roman> try to expose it as enum ???  You made a mistake.
>
>Yes, I'm a moron.  Thanks!
>
>
>    typedef unsigned char cover_type;    //----cover_type
>    enum
>    {
>        cover_shift = 8,                 //----cover_shift
>        cover_size  = 1 << cover_shift,  //----cover_size 
>        cover_mask  = cover_size - 1,    //----cover_mask 
>        cover_none  = 0,                 //----cover_none 
>        cover_full  = cover_mask         //----cover_full 
>    };
>
>    Roman> Enum( "cover_mask", "agg_basics.h" ) should work
>
>It doesn't.  I was, however, able to do it manually with
>
>module_code("""    
>    enum_<int>("cover")
>        .value("cover_shift", agg::cover_shift)
>        .value("cover_size",  agg::cover_size)
>        .value("cover_mask",  agg::cover_mask)
>        .value("cover_none",  agg::cover_none)
>        .value("cover_full",  agg::cover_full)                        
>        ;
>        """)
>
>but as always, I'm looking for the best pyste/boost::python way to do
>it, so any other suggestions on the best way to export unnamed enums
>in pyste welcome!
>
>Thanks,
>JDH
>_______________________________________________
>C++-sig mailing list
>C++-sig at python.org
>http://mail.python.org/mailman/listinfo/c++-sig
>
>



From nicodemus at esss.com.br  Wed Nov 17 00:27:05 2004
From: nicodemus at esss.com.br (Nicodemus)
Date: Tue, 16 Nov 2004 20:27:05 -0300
Subject: [C++-sig] pyste and return reference problem
In-Reply-To: <m3sm7jx62p.fsf@peds-pc311.bsd.uchicago.edu>
References: <m3sm7jx62p.fsf@peds-pc311.bsd.uchicago.edu>
Message-ID: <419A8CC9.3090307@esss.com.br>

Hi Jonh,

John Hunter wrote:

>I have found what looks like pyste bug.  pyste wraps the following
>struct
>
>  struct C {
>    const C& func(double val) {return *this;}
>    double func() {return 0.0;}
>  };
>
>as
>
>    class_< C >("C", init<  >())
>        .def(init< const C& >())
>        .def("func", (const C& (C::*)(double) )&C::func, return_value_policy< copy_const_reference >())
>        .def("func", (double (C::*)() )&C::func, return_value_policy< copy_const_reference >())
>    ;
> 
>
>
>Note that (double (C::*)() )&C::func has a call policy of
>return_value_policy< copy_const_reference >()).  
>
>If you reverse the order of the method definitions in C, ie
>
>
>  struct C {
>    double func() {return 0.0;}
>    const C& func(double val) {return *this;}
>  };
>
>pyste gets it right
>
>    class_< C >("C", init<  >())
>        .def(init< const C& >())
>        .def("func", (double (C::*)() )&C::func)
>        .def("func", (const C& (C::*)(double) )&C::func, return_value_policy< copy_const_reference >())
>    ;
>

Yeah, sounds like a bug... unfortunately I don't have time to dig into 
the code to solve this one. :/

>This leads me to a question related to one of my earlier questions:
>when wrapping overloaded functions as above, is there a way to
>differentially set the call policies in pyste for the different
>signatures.  I know one can use as described in the docs
>
>    set_policy(f, return_internal_reference())
>    set_policy(C.foo, return_value_policy(manage_new_object))
>
>How does one do this for overloaded functions?
>

If I'm not mistaken Pyste doesn't currently support that. Sorry.

>Sorry for the deluge of mail!  I'm going home now :-)
>

That's alright; Sorry again for taking so long to respond.

Regards,
Bruno.


From Stefan.Quandt at bardenheuer.de  Wed Nov 17 08:28:17 2004
From: Stefan.Quandt at bardenheuer.de (Stefan Quandt)
Date: Wed, 17 Nov 2004 08:28:17 +0100
Subject: [C++-sig] Re: -sBUILD=debug-python
References: <cnd6dg$oqu$1@sea.gmane.org> <u7joliwwc.fsf@boost-consulting.com>
Message-ID: <cneuii$peq$1@sea.gmane.org>

David Abrahams wrote:

> Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:
> 
>> I try to build an extension with python-debug on Win32.
>>
>> I compiled a debug version of the Python interpreter from source and
>> adjusted PYTHON_ROOT in my Jamrules to it's location.
>>
>> Then I try to build my extendions with
>>   bjam -sBUILD=python-debug
>> When bjam tries to link boost_python.dll it complains not finding
>> "python23.lib".
>>
>> As far as I know it should be "python23_d.lib".
>> Or am I missing something?
> 
>    bjam -sBUILD=debug-python
>                 ^^^^^^^^^^^^
Ah, yes, that's what I did.
It's only written wrong in my posting.

Why dows bjam references the non-debug version python23.lib when linking
boost_python.dll?

I use Boost.build v1 with a boost snapshot from July this year (which has a
small workaround to avoid a VC6 template error).

Thank you for your response
  Stefan




From Paul_Kunz at slac.stanford.edu  Wed Nov 17 08:38:51 2004
From: Paul_Kunz at slac.stanford.edu (Paul F. Kunz)
Date: Tue, 16 Nov 2004 23:38:51 -0800
Subject: [C++-sig] Deriving Python function from C++ base class with
	pure virtual function
In-Reply-To: <1100471714.18970.21.camel@illuvatar> (message from Jonathan
	Brandmeyer on Sun, 14 Nov 2004 17:35:14 -0500)
References: <cn2mf2$vht$1@sea.gmane.org> <4194E059.30809@cenix-bioscience.com>
	<200411130953.iAD9rI323669@libra3.slac.stanford.edu>
	<1100471714.18970.21.camel@illuvatar>
Message-ID: <200411170738.iAH7cpT14438@libra3.slac.stanford.edu>

   Thanks for the advice I got.   I haven't got around to trying
anything yet as I'm travel on business in Europe.


From dave at boost-consulting.com  Wed Nov 17 20:24:58 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 17 Nov 2004 14:24:58 -0500
Subject: [C++-sig] Re: -sBUILD=debug-python
References: <cnd6dg$oqu$1@sea.gmane.org> <u7joliwwc.fsf@boost-consulting.com>
	<cneuii$peq$1@sea.gmane.org>
Message-ID: <u3bz89t2t.fsf@boost-consulting.com>

Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:

> David Abrahams wrote:
>
>> Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:
>> 
>>> I try to build an extension with python-debug on Win32.
>>>
>>> I compiled a debug version of the Python interpreter from source and
>>> adjusted PYTHON_ROOT in my Jamrules to it's location.
>>>
>>> Then I try to build my extendions with
>>>   bjam -sBUILD=python-debug
>>> When bjam tries to link boost_python.dll it complains not finding
>>> "python23.lib".
>>>
>>> As far as I know it should be "python23_d.lib".
>>> Or am I missing something?
>> 
>>    bjam -sBUILD=debug-python
>>                 ^^^^^^^^^^^^
> Ah, yes, that's what I did.
> It's only written wrong in my posting.
>
> Why dows bjam references the non-debug version python23.lib when linking
> boost_python.dll?

Oh, that's why.

You have to link with boost_python_pydebug.lib/.dll (or the
boost_python.lib/.dll that's created in a path including
.../debug-python/...) if you want to use Python debug mode.

Of course, if you use the method in libs/python/example/Jamfile you
won't have this problem
-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From dave at boost-consulting.com  Thu Nov 18 01:28:35 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 17 Nov 2004 19:28:35 -0500
Subject: [C++-sig] Re: Custom WeakPtrs and "None"
References: <1gncrqj.51k2my18f3gmgM%haller@ableton.com>
Message-ID: <u1xes57bg.fsf@boost-consulting.com>

haller at ableton.com (Stefan Haller) writes:

> The only problem is that when the object goes away, the "weak pointer"
> on the Python side isn't set to "None".  Of course, the wrapped C++ weak
> pointer is set to NULL, but the python variable isn't aware of that
> unless you try to access properties or call methods on it, in which case
> it throws an exception.  It would be very nice if a comparison with None
> would return true for a weak pointer that has been set to NULL.

Nothing will do that; how would you know all the places the Python
object is bound to an attribute name or local variable?

It's in the nature of Python.  It has "name bindings," not "variable
values."

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From ndbecker2 at verizon.net  Thu Nov 18 03:53:29 2004
From: ndbecker2 at verizon.net (Neal Becker)
Date: Wed, 17 Nov 2004 21:53:29 -0500
Subject: [C++-sig] make_constructor strangeness
Message-ID: <cnh2r9$onj$1@sea.gmane.org>

I noticed that the following didn't work:

  pyclass
    .def(init<size_type>())
    .def("__init__", make_constructor(vector_from_object<value_type>))

where size_type is some size (probably size_t), value_type could be double
or std::complex<double>, and vector_from_object has sig:
template<typename T>
inline vector<T>* vector_from_object (object& newobject) {
 
It seems if I try to construct an object using an argument of type int, the
init<size_type> is being ignored (silently), and the call goes to
vector_from_object.

I guess you can't dispatch an overload to a function that takes "object". 
Problem is, it seems to act unpredictably, and fails silently.

Unpredictably, because I used the same code for the case where value_type
was double, and also when it was std::complex<double>, and the two acted
differently.  The complex called with an int did call the first
constructor, while the double one called the second.  AFAICT, there were no
other significant differences between the two cases.




From jbrandmeyer at earthlink.net  Thu Nov 18 04:05:38 2004
From: jbrandmeyer at earthlink.net (Jonathan Brandmeyer)
Date: Wed, 17 Nov 2004 22:05:38 -0500
Subject: [C++-sig] make_constructor strangeness
In-Reply-To: <cnh2r9$onj$1@sea.gmane.org>
References: <cnh2r9$onj$1@sea.gmane.org>
Message-ID: <1100747138.23265.26.camel@illuvatar>

On Wed, 2004-11-17 at 21:53 -0500, Neal Becker wrote:
> I noticed that the following didn't work:
> 
>   pyclass
>     .def(init<size_type>())
>     .def("__init__", make_constructor(vector_from_object<value_type>))
> 
> where size_type is some size (probably size_t), value_type could be double
> or std::complex<double>, and vector_from_object has sig:
> template<typename T>
> inline vector<T>* vector_from_object (object& newobject) {
>  
> It seems if I try to construct an object using an argument of type int, the
> init<size_type> is being ignored (silently), and the call goes to
> vector_from_object.
> 
> I guess you can't dispatch an overload to a function that takes "object". 
> Problem is, it seems to act unpredictably, and fails silently.
> 
> Unpredictably, because I used the same code for the case where value_type
> was double, and also when it was std::complex<double>, and the two acted
> differently.  The complex called with an int did call the first
> constructor, while the double one called the second.  AFAICT, there were no
> other significant differences between the two cases.

I believe that overload resolution is performed in reverse order of the
execution of class_<>::def() (same goes for free functions).  Since the
function passed with make_constructor takes a boost::python::object,
that function will _always_ win because any invocation from Python with
only one argument will match, and that signature was added last.  If you
reverse the order such that function taking a boost::python::object gets
wrapped with def first, I think it will work.

HTH,
-Jonathan



From haller at ableton.com  Thu Nov 18 14:41:07 2004
From: haller at ableton.com (Stefan Haller)
Date: Thu, 18 Nov 2004 14:41:07 +0100
Subject: [C++-sig] Re: Custom WeakPtrs and "None"
In-Reply-To: <u1xes57bg.fsf@boost-consulting.com>
Message-ID: <1gngadz.ycr6onpmqkzkM%haller@ableton.com>

David Abrahams <dave at boost-consulting.com> wrote:

> haller at ableton.com (Stefan Haller) writes:
> 
> > The only problem is that when the object goes away, the "weak pointer"
> > on the Python side isn't set to "None".  Of course, the wrapped C++ weak
> > pointer is set to NULL, but the python variable isn't aware of that
> > unless you try to access properties or call methods on it, in which case
> > it throws an exception.  It would be very nice if a comparison with None
> > would return true for a weak pointer that has been set to NULL.
> 
> Nothing will do that; how would you know all the places the Python
> object is bound to an attribute name or local variable?
> 
> It's in the nature of Python.  It has "name bindings," not "variable
> values."

Yes, I don't want the Python variable itself to be set to None
literally, I know there's no way to do that.  I just want it to *look* a
bit more like None if the WeakPtr is NULL.  For example, I was hoping
that it is possible to play tricks with overloading __cmp__ (or some
such) to make the expression (weakptr == None) evaluate to True if
weakptr is not None, but its pointee is NULL.

A similar problem is comparison of two weak ptrs; they compare
different, even if they point to the same underlying object.  It would
be nice to compare pointees, not pointers here.


-- 
Stefan Haller
Ableton
http://www.ableton.com/


From Stefan.Quandt at bardenheuer.de  Thu Nov 18 15:51:36 2004
From: Stefan.Quandt at bardenheuer.de (Stefan Quandt)
Date: Thu, 18 Nov 2004 15:51:36 +0100
Subject: [C++-sig] Re: -sBUILD=debug-python
References: <cnd6dg$oqu$1@sea.gmane.org> <u7joliwwc.fsf@boost-consulting.com>
	<cneuii$peq$1@sea.gmane.org> <u3bz89t2t.fsf@boost-consulting.com>
Message-ID: <cnicto$s64$1@sea.gmane.org>

David Abrahams wrote:

> Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:
> 
>> David Abrahams wrote:
>>
>>> Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:
>>> 
>>>> I try to build an extension with python-debug on Win32.
>>>>
>>>> I compiled a debug version of the Python interpreter from source and
>>>> adjusted PYTHON_ROOT in my Jamrules to it's location.
>>>>
>>>> Then I try to build my extendions with
>>>>   bjam -sBUILD=python-debug
>>>> When bjam tries to link boost_python.dll it complains not finding
>>>> "python23.lib".
>>>>
>>>> As far as I know it should be "python23_d.lib".
>>>> Or am I missing something?
>>> 
>>>    bjam -sBUILD=debug-python
>>>                 ^^^^^^^^^^^^
>> Ah, yes, that's what I did.
>> It's only written wrong in my posting.
>>
>> Why dows bjam references the non-debug version python23.lib when linking
>> boost_python.dll?
> 
> Oh
> , that's why. 
> 
> You have to link with boost_python_pydebug.lib/.dll (or the
> boost_python.lib/.dll that's created in a path including
> .../debug-python/...) if you want to use Python debug mode.
As I wrote bjam tries to create a normal "boost_python.dll"

> Of course, if you use the method in libs/python/example/Jamfile you
> won't have this problem
In fact my Jamfile is derived from the given example and it's very similar
in structure.
My extension target rule differs only in 
 1) specifying multiple files in the source argument and
 2) having a requirements argument with <library-path> and <find-library>
items).

So what "mechanism" would bjam bring to create
boost_python_pydebug.lib/.dll?
After reading
  file:/home/qua/boost-2004-07-05/libs/python/doc/building.html#variants
I assumed it would be as simple as using the "-sBUILD=debug-python" option.

Thanks
  Stefan



From dave at boost-consulting.com  Thu Nov 18 21:24:06 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Thu, 18 Nov 2004 15:24:06 -0500
Subject: [C++-sig] Re: Custom WeakPtrs and "None"
References: <u1xes57bg.fsf@boost-consulting.com>
	<1gngadz.ycr6onpmqkzkM%haller@ableton.com>
Message-ID: <usm76zz15.fsf@boost-consulting.com>

haller at ableton.com (Stefan Haller) writes:

> David Abrahams <dave at boost-consulting.com> wrote:
>
>> haller at ableton.com (Stefan Haller) writes:
>> 
>> > The only problem is that when the object goes away, the "weak pointer"
>> > on the Python side isn't set to "None".  Of course, the wrapped C++ weak
>> > pointer is set to NULL, but the python variable isn't aware of that
>> > unless you try to access properties or call methods on it, in which case
>> > it throws an exception.  It would be very nice if a comparison with None
>> > would return true for a weak pointer that has been set to NULL.
>> 
>> Nothing will do that; how would you know all the places the Python
>> object is bound to an attribute name or local variable?
>> 
>> It's in the nature of Python.  It has "name bindings," not "variable
>> values."
>
> Yes, I don't want the Python variable itself to be set to None
> literally, I know there's no way to do that.  I just want it to *look* a
> bit more like None if the WeakPtr is NULL.  For example, I was hoping
> that it is possible to play tricks with overloading __cmp__ (or some
> such) to make the expression (weakptr == None) evaluate to True if
> weakptr is not None, but its pointee is NULL.

You could play that trick manually.  I would definitely *not* want to
do that automatically in the library.

Note that if you wrap this function:

     X* identity(X* x) { return x; }

you actually *will* get None back if you pass it one of these
released weak_ptrs.

> A similar problem is comparison of two weak ptrs; they compare
> different, even if they point to the same underlying object.  It would
> be nice to compare pointees, not pointers here.

That's a very separate but interesting issue.  Perhaps when two C++
objects held by pointer are compared in Python we ought to see if the
held pointers refer to the same thing.

It becomes strange when (multiple) inheritance is involved.  Probably
we'd want to return False unless the two Python objects had the same
type.

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From dave at boost-consulting.com  Thu Nov 18 21:29:38 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Thu, 18 Nov 2004 15:29:38 -0500
Subject: [C++-sig] Re: -sBUILD=debug-python
References: <cnd6dg$oqu$1@sea.gmane.org> <u7joliwwc.fsf@boost-consulting.com>
	<cneuii$peq$1@sea.gmane.org> <u3bz89t2t.fsf@boost-consulting.com>
	<cnicto$s64$1@sea.gmane.org>
Message-ID: <umzxezyrx.fsf@boost-consulting.com>

Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:

> David Abrahams wrote:
>
>> Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:
>> 
>>> David Abrahams wrote:
>>>
>>>> Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:
>>>> 
>>>>> I try to build an extension with python-debug on Win32.
>>>>>
>>>>> I compiled a debug version of the Python interpreter from source and
>>>>> adjusted PYTHON_ROOT in my Jamrules to it's location.
>>>>>
>>>>> Then I try to build my extendions with
>>>>>   bjam -sBUILD=python-debug
>>>>> When bjam tries to link boost_python.dll it complains not finding
>>>>> "python23.lib".
>>>>>
>>>>> As far as I know it should be "python23_d.lib".
>>>>> Or am I missing something?
>>>> 
>>>>    bjam -sBUILD=debug-python
>>>>                 ^^^^^^^^^^^^
>>> Ah, yes, that's what I did.
>>> It's only written wrong in my posting.
>>>
>>> Why dows bjam references the non-debug version python23.lib when linking
>>> boost_python.dll?
>> 
>> Oh
>> , that's why. 
>> 
>> You have to link with boost_python_pydebug.lib/.dll (or the
>> boost_python.lib/.dll that's created in a path including
>> .../debug-python/...) if you want to use Python debug mode.
> As I wrote bjam tries to create a normal "boost_python.dll"

That's fine, if you're using the method in
libs/python/example/Jamfile as you say.  That boost_python.dll is
actually compiled against the "_d" version of the Python library.

>> Of course, if you use the method in libs/python/example/Jamfile you
>> won't have this problem
>
> In fact my Jamfile is derived from the given example and it's very similar
> in structure.
> My extension target rule differs only in 
>  1) specifying multiple files in the source argument and
>  2) having a requirements argument with <library-path> and <find-library>
> items).

What libraries are you looking for?  I hope not Python or boost_python!

> So what "mechanism" would bjam bring to create
> boost_python_pydebug.lib/.dll?
>
> After reading
>   file:/home/qua/boost-2004-07-05/libs/python/doc/building.html#variants
> I assumed it would be as simple as using the "-sBUILD=debug-python" option.

That file only gets created when you build the "bin-stage" target, I
think.

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From amp at singingwizard.org  Fri Nov 19 00:17:50 2004
From: amp at singingwizard.org (Arthur Peters)
Date: Thu, 18 Nov 2004 17:17:50 -0600
Subject: [C++-sig] Boost.Python and C++ side garbage collection
Message-ID: <E1CUvXC-0002nq-Lu@lazarus.localnet.local>


Hello,

Straight to the question:

Is there a way to have Booth.Python or python keep a pointer to the C++
object in memory (as a pointer, not a string or other representation)
while the Python object wrapping it exists?

This may be a silly question. I don't know the internals of Boost.Python
or python well.

This is important because I am writting a program that uses libgc (the
boehm garbage collector) so pointers do not have owners in the standard
sense. The garbage collector will clean things up when noone references
them anymore. If a reference to the object exists in python land there
must be a plain pointer in memory somewhere or the object will be
deleted by libgc. And I don't python to delete it because C++ might
still have references.

Thanks a bunch

-Arthur


From Stefan.Quandt at bardenheuer.de  Fri Nov 19 08:47:30 2004
From: Stefan.Quandt at bardenheuer.de (Stefan Quandt)
Date: Fri, 19 Nov 2004 08:47:30 +0100
Subject: [C++-sig] Re: -sBUILD=debug-python
References: <cnd6dg$oqu$1@sea.gmane.org> <u7joliwwc.fsf@boost-consulting.com>
	<cneuii$peq$1@sea.gmane.org> <u3bz89t2t.fsf@boost-consulting.com>
	<cnicto$s64$1@sea.gmane.org> <umzxezyrx.fsf@boost-consulting.com>
Message-ID: <cnk8ej$5u2$1@sea.gmane.org>

David Abrahams wrote:

> Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:
> 
>> David Abrahams wrote:
>>
>>> Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:
>>> 
>>>> David Abrahams wrote:
>>>>
>>>>> Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:
>>>>> 
>>>>>> I try to build an extension with python-debug on Win32.
>>>>>>
>>>>>> I compiled a debug version of the Python interpreter from source and
>>>>>> adjusted PYTHON_ROOT in my Jamrules to it's location.
>>>>>>
>>>>>> Then I try to build my extendions with
>>>>>>   bjam -sBUILD=python-debug
>>>>>> When bjam tries to link boost_python.dll it complains not finding
>>>>>> "python23.lib".
>>>>>>
>>>>>> As far as I know it should be "python23_d.lib".
>>>>>> Or am I missing something?
>>>>> 
>>>>>    bjam -sBUILD=debug-python
>>>>>                 ^^^^^^^^^^^^
>>>> Ah, yes, that's what I did.
>>>> It's only written wrong in my posting.
>>>>
>>>> Why dows bjam references the non-debug version python23.lib when
>>>> linking boost_python.dll?
>>> 
>>> Oh
>>> , that's why.
>>> 
>>> You have to link with boost_python_pydebug.lib/.dll (or the
>>> boost_python.lib/.dll that's created in a path including
>>> .../debug-python/...) if you want to use Python debug mode.
>> As I wrote bjam tries to create a normal "boost_python.dll"
> 
> That's fine, if you're using the method in
> libs/python/example/Jamfile as you say.  That boost_python.dll is
> actually compiled against the "_d" version of the Python library.
I suspect that this does not happen!
And my central question is, why not.
When bjam tries to link boost_python.dll I get an ERROR complaining that the
NON DEBUG VERSION "python23.lib" could not be found. 

>>> Of course, if you use the method in libs/python/example/Jamfile you
>>> won't have this problem
>>
>> In fact my Jamfile is derived from the given example and it's very
>> similar in structure.
>> My extension target rule differs only in
>>  1) specifying multiple files in the source argument and
>>  2) having a requirements argument with <library-path> and <find-library>
>> items).
> 
> What libraries are you looking for?  I hope not Python or boost_python!
No, just a propietary Bluetooth-Stack and additional Win32 dependencies.





From ndbecker2 at verizon.net  Fri Nov 19 15:52:32 2004
From: ndbecker2 at verizon.net (Neal D. Becker)
Date: Fri, 19 Nov 2004 09:52:32 -0500
Subject: [C++-sig] python2.4 / maintenance
Message-ID: <cnl1bg$cqo$1@sea.gmane.org>

Is boost::python being actively maintained?  Is anyone testing with
python2.4?



From haller at ableton.com  Fri Nov 19 16:48:15 2004
From: haller at ableton.com (Stefan Haller)
Date: Fri, 19 Nov 2004 16:48:15 +0100
Subject: [C++-sig] Re: Custom WeakPtrs and "None"
In-Reply-To: <usm76zz15.fsf@boost-consulting.com>
Message-ID: <1gniac6.e9vi5g1gle0wmM%haller@ableton.com>

David Abrahams <dave at boost-consulting.com> wrote:

> haller at ableton.com (Stefan Haller) writes:
> 
> > Yes, I don't want the Python variable itself to be set to None
> > literally, I know there's no way to do that.  I just want it to *look* a
> > bit more like None if the WeakPtr is NULL.  For example, I was hoping
> > that it is possible to play tricks with overloading __cmp__ (or some
> > such) to make the expression (weakptr == None) evaluate to True if
> > weakptr is not None, but its pointee is NULL.
> 
> You could play that trick manually.  I would definitely *not* want to
> do that automatically in the library.

Could you please give me a hint how I would go about that?  I'm afraid I
don't know what exactly has to be done.

> Note that if you wrap this function:
> 
>      X* identity(X* x) { return x; }
> 
> you actually *will* get None back if you pass it one of these
> released weak_ptrs.

Yes, this is to be expected, right?  But I'm not sure how I can make use
of this in solving my problem.


-- 
Stefan Haller
Ableton
http://www.ableton.com/


From hugh-m at moving-picture.com  Fri Nov 19 17:34:49 2004
From: hugh-m at moving-picture.com (Hugh Macdonald)
Date: Fri, 19 Nov 2004 16:34:49 +0000
Subject: [C++-sig] Embedding/Extending Python at the same time...
Message-ID: <20041119163449.48365fd6.hugh-m@moving-picture.com>

I posted this on comp.lang.python, but then found this mailing list, so I hope it's not considered bad etiquette to post in both places....

I've got a slight problem... and I'm stuck as to where to go with it...

I'm running on Redhat7.2, using Python 2.2.2

I've got a compiled module that I wrote almost a year ago - it works fine, and I've never had any problems with it...

I also did an extension to a plugin (Shake) using Python, so the plugin loads a python module and calls various functions in there....

all in all, fine - I thought I had my head around the system, and there were no problems...

However, I've just hit a problem...

If, in my plugin code, I import a module that is pure python, I'm fine. However, if I import my own compiled module, then I start having problems....

Initially, the module was compiled without linking to the actual python library. When I run the plugin, it fails, and PyErr_Print() prints out:

Traceback (most recent call last):
  File "/usr/people/hugh-m/dev/apps/hub/python/modules/noarch/hubdbremote.py", line 19, in ?
    import hub
ImportError: /usr/people/hugh-m/dev/apps/hub/python/modules/linux.redhat7.2.i386/hub.so: undefined symbol: PyFloat_Type


If I modify the SConstruct file so that the module links to the python library, it gives the following error:

Fatal Python error: Interpreter not initialized (version mismatch?)

Any idea what I could do to get around either of these?

Should python modules be compiled with the library linked in?

The plugin is compiled with gcc3.2.1 (needed because it's a Shake plugin), and the module is compiled with gcc2.96 (because that is what Python needs it to be). Is this my issue? If so, how can I get around it?


Any suggestions most welcome... and apologies for the large number of questions..

Cheers

-- 
Hugh Macdonald
The Moving Picture Company


From dave at boost-consulting.com  Fri Nov 19 18:10:12 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Fri, 19 Nov 2004 12:10:12 -0500
Subject: [C++-sig] Re: -sBUILD=debug-python
References: <cnd6dg$oqu$1@sea.gmane.org> <u7joliwwc.fsf@boost-consulting.com>
	<cneuii$peq$1@sea.gmane.org> <u3bz89t2t.fsf@boost-consulting.com>
	<cnicto$s64$1@sea.gmane.org> <umzxezyrx.fsf@boost-consulting.com>
	<cnk8ej$5u2$1@sea.gmane.org>
Message-ID: <usm75pxxn.fsf@boost-consulting.com>

Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:

>> That's fine, if you're using the method in
>> libs/python/example/Jamfile as you say.  That boost_python.dll is
>> actually compiled against the "_d" version of the Python library.
> I suspect that this does not happen!
> And my central question is, why not.
> When bjam tries to link boost_python.dll I get an ERROR complaining that the
> NON DEBUG VERSION "python23.lib" could not be found. 

Do a 

  bjam -n -a -sBUILD=debug-python

to see what it tries to do when building.  If you can't see any
problems with it, post the whole thing here.  Otherwise, please post
what you see going wrong.

>>>> Of course, if you use the method in libs/python/example/Jamfile you
>>>> won't have this problem
>>>
>>> In fact my Jamfile is derived from the given example and it's very
>>> similar in structure.
>>> My extension target rule differs only in
>>>  1) specifying multiple files in the source argument and
>>>  2) having a requirements argument with <library-path> and <find-library>
>>> items).
>> 
>> What libraries are you looking for?  I hope not Python or boost_python!
> No, just a propietary Bluetooth-Stack and additional Win32 dependencies.

Hmm.

I'm trying again to reproduce your problem here.  Maybe by comparing
our bjam output we can learn something.  FYI, I am going into
boost/libs/python/example and running

  bjam -d+2 -sTOOLS=vc-7_1 -sBUILD=debug-python test1


-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From dave at boost-consulting.com  Fri Nov 19 18:14:11 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Fri, 19 Nov 2004 12:14:11 -0500
Subject: [C++-sig] Re: python2.4 / maintenance
References: <cnl1bg$cqo$1@sea.gmane.org>
Message-ID: <uhdnlpxr0.fsf@boost-consulting.com>

"Neal D. Becker" <ndbecker2 at verizon.net> writes:

> Is boost::python being actively maintained?  

Maintenance is mostly all it's getting; new features don't get added
regularly.  But, yes, maintenance is active for some reasonable
definition of "active." ;-)

> Is anyone testing with python2.4?

I'm not.  I don't know about other people.

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From dave at boost-consulting.com  Fri Nov 19 18:12:20 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Fri, 19 Nov 2004 12:12:20 -0500
Subject: [C++-sig] Re: Boost.Python and C++ side garbage collection
References: <E1CUvXC-0002nq-Lu@lazarus.localnet.local>
Message-ID: <umzxdpxu3.fsf@boost-consulting.com>

"Arthur Peters" <amp at singingwizard.org> writes:

> Hello,
>
> Straight to the question:
>
> Is there a way to have Booth.Python or python keep a pointer to the C++
> object in memory (as a pointer, not a string or other representation)
> while the Python object wrapping it exists?

It already does that, except in the default case where the C++
object's storage is actually embedded in that of the Python object.
The pointer is kept in the memory occupied by the Python object.

> This may be a silly question. I don't know the internals of Boost.Python
> or python well.

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From rwgk at yahoo.com  Fri Nov 19 19:19:44 2004
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Fri, 19 Nov 2004 10:19:44 -0800 (PST)
Subject: [C++-sig] Re: python2.4 / maintenance
In-Reply-To: <uhdnlpxr0.fsf@boost-consulting.com>
Message-ID: <20041119181944.29747.qmail@web20225.mail.yahoo.com>

--- David Abrahams <dave at boost-consulting.com> wrote:

> "Neal D. Becker" <ndbecker2 at verizon.net> writes:
> 
> > Is boost::python being actively maintained?  
> 
> Maintenance is mostly all it's getting; new features don't get added
> regularly.  But, yes, maintenance is active for some reasonable
> definition of "active." ;-)
> 
> > Is anyone testing with python2.4?
> 
> I'm not.  I don't know about other people.

I am, all the time. Boost.Python (latest CVS) and Python 2.4b2 are fully
compatible. I'll check the just-released 2.4c1 later today.
Ralf



		
__________________________________ 
Do you Yahoo!? 
The all-new My Yahoo! - Get yours free! 
http://my.yahoo.com 
 



From ndbecker2 at verizon.net  Fri Nov 19 21:44:39 2004
From: ndbecker2 at verizon.net (Neal D. Becker)
Date: Fri, 19 Nov 2004 15:44:39 -0500
Subject: [C++-sig] Re: Re: python2.4 / maintenance
References: <uhdnlpxr0.fsf@boost-consulting.com>
	<20041119181944.29747.qmail@web20225.mail.yahoo.com>
Message-ID: <cnllvn$ee1$1@sea.gmane.org>

Ralf W. Grosse-Kunstleve wrote:

> --- David Abrahams <dave at boost-consulting.com> wrote:
> 
>> "Neal D. Becker" <ndbecker2 at verizon.net> writes:
>> 
>> > Is boost::python being actively maintained?
>> 
>> Maintenance is mostly all it's getting; new features don't get added
>> regularly.  But, yes, maintenance is active for some reasonable
>> definition of "active." ;-)
>> 
>> > Is anyone testing with python2.4?
>> 
>> I'm not.  I don't know about other people.
> 
> I am, all the time. Boost.Python (latest CVS) and Python 2.4b2 are fully
> compatible. I'll check the just-released 2.4c1 later today.
> Ralf
> 
> 

Thanks!

Better, check that about-to-be-released boost-1.32 will work with
soon-to-be-released python2.4?



From rwgk at yahoo.com  Fri Nov 19 22:25:49 2004
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Fri, 19 Nov 2004 13:25:49 -0800 (PST)
Subject: [C++-sig] Re: Re: python2.4 / maintenance
In-Reply-To: <cnllvn$ee1$1@sea.gmane.org>
Message-ID: <20041119212549.44210.qmail@web20226.mail.yahoo.com>

--- "Neal D. Becker" <ndbecker2 at verizon.net> wrote:
> Better, check that about-to-be-released boost-1.32 will work with
> soon-to-be-released python2.4?

I lost all interest in boost releases. I prefer to invest my time in keeping
the CVS mainline healthy (fortunately not much time at all these days). Could
you try the boost release candidate + Python 2.4c1?
Ralf



		
__________________________________ 
Do you Yahoo!? 
The all-new My Yahoo! - Get yours free! 
http://my.yahoo.com 
 



From dave at boost-consulting.com  Fri Nov 19 23:17:26 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Fri, 19 Nov 2004 17:17:26 -0500
Subject: [C++-sig] Re: Custom WeakPtrs and "None"
References: <usm76zz15.fsf@boost-consulting.com>
	<1gniac6.e9vi5g1gle0wmM%haller@ableton.com>
Message-ID: <uhdnlpjpl.fsf@boost-consulting.com>

haller at ableton.com (Stefan Haller) writes:

> David Abrahams <dave at boost-consulting.com> wrote:
>
>> haller at ableton.com (Stefan Haller) writes:
>> 
>> > Yes, I don't want the Python variable itself to be set to None
>> > literally, I know there's no way to do that.  I just want it to
>> > *look* a bit more like None if the WeakPtr is NULL.  For example,
>> > I was hoping that it is possible to play tricks with overloading
>> > __cmp__ (or some such) to make the expression (weakptr == None)
>> > evaluate to True if weakptr is not None, but its pointee is NULL.
>> 
>> You could play that trick manually.  I would definitely *not* want
>> to do that automatically in the library.
>
> Could you please give me a hint how I would go about that?  I'm
> afraid I don't know what exactly has to be done.

Just define an __eq__ function that compares a weakptr<T> with T*:

bool eq(weakptr<X> x, X* other)
{
    return x.get() == other;
}

bool neq(weakptr<X> x, X* other)
{
    return x.get() != other;
}


...
        class_<X, ... >("X")
             .def("__eq__", eq)
             .def("__neq__", neq)
             ;


-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From dave at boost-consulting.com  Fri Nov 19 23:10:05 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Fri, 19 Nov 2004 17:10:05 -0500
Subject: [C++-sig] Re: Embedding/Extending Python at the same time...
References: <20041119163449.48365fd6.hugh-m@moving-picture.com>
Message-ID: <umzxdpk1u.fsf@boost-consulting.com>

Hugh Macdonald <hugh-m at moving-picture.com> writes:

> I posted this on comp.lang.python, but then found this mailing list,
> so I hope it's not considered bad etiquette to post in both places....

Etiquette aside, does this question have anything to do with C++?  It
doesn't seem to.

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From dave at boost-consulting.com  Sat Nov 20 02:45:09 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Fri, 19 Nov 2004 20:45:09 -0500
Subject: [C++-sig] Re: -sBUILD=debug-python
References: <cnd6dg$oqu$1@sea.gmane.org> <u7joliwwc.fsf@boost-consulting.com>
	<cneuii$peq$1@sea.gmane.org> <u3bz89t2t.fsf@boost-consulting.com>
	<cnicto$s64$1@sea.gmane.org> <umzxezyrx.fsf@boost-consulting.com>
	<cnk8ej$5u2$1@sea.gmane.org> <usm75pxxn.fsf@boost-consulting.com>
Message-ID: <uvfc1waxm.fsf@boost-consulting.com>

David Abrahams <dave at boost-consulting.com> writes:

> Hmm.
>
> I'm trying again to reproduce your problem here.  Maybe by comparing
> our bjam output we can learn something.  FYI, I am going into
> boost/libs/python/example and running
>
>   bjam -d+2 -sTOOLS=vc-7_1 -sBUILD=debug-python test1
>

Here's what I got.  Everything seems to have worked perfectly.

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: c.err
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041119/37f6987b/attachment.txt>
-------------- next part --------------

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com

From iluetkeb at gmail.com  Sat Nov 20 23:01:18 2004
From: iluetkeb at gmail.com (=?ISO-8859-1?Q?Ingo_L=FCtkebohle?=)
Date: Sat, 20 Nov 2004 23:01:18 +0100
Subject: [C++-sig] Embedding/Extending Python at the same time...
In-Reply-To: <20041119163449.48365fd6.hugh-m@moving-picture.com>
References: <20041119163449.48365fd6.hugh-m@moving-picture.com>
Message-ID: <fdd727f80411201401eacb6d0@mail.gmail.com>

> Fatal Python error: Interpreter not initialized (version mismatch?)

Call Py_Initialize() yourself.  It doesn't hurt if it gets called more
than once.

Ingo


From hugh-m at moving-picture.com  Mon Nov 22 10:17:00 2004
From: hugh-m at moving-picture.com (Hugh Macdonald)
Date: Mon, 22 Nov 2004 09:17:00 +0000
Subject: [C++-sig] Re: Embedding/Extending Python at the same time...
In-Reply-To: <umzxdpk1u.fsf@boost-consulting.com>
References: <20041119163449.48365fd6.hugh-m@moving-picture.com>
	<umzxdpk1u.fsf@boost-consulting.com>
Message-ID: <20041122091700.5a0ad616.hugh-m@moving-picture.com>

On Fri, 19 Nov 2004 17:10:05 -0500
David Abrahams <dave at boost-consulting.com> wrote:

> Hugh Macdonald <hugh-m at moving-picture.com> writes:
> 
> > I posted this on comp.lang.python, but then found this mailing list,
> > so I hope it's not considered bad etiquette to post in both
> > places....
> 
> Etiquette aside, does this question have anything to do with C++?  It
> doesn't seem to.

I assumed that this list was to do with extending/embedding Python using C++....  If I've missed something out here, then I apologise....

-- 
Hugh Macdonald
The Moving Picture Company


From Stefan.Quandt at bardenheuer.de  Mon Nov 22 12:25:36 2004
From: Stefan.Quandt at bardenheuer.de (Stefan Quandt)
Date: Mon, 22 Nov 2004 12:25:36 +0100
Subject: [C++-sig] Re: -sBUILD=debug-python, solved
References: <cnd6dg$oqu$1@sea.gmane.org> <u7joliwwc.fsf@boost-consulting.com>
	<cneuii$peq$1@sea.gmane.org> <u3bz89t2t.fsf@boost-consulting.com>
	<cnicto$s64$1@sea.gmane.org> <umzxezyrx.fsf@boost-consulting.com>
	<cnk8ej$5u2$1@sea.gmane.org> <usm75pxxn.fsf@boost-consulting.com>
	<uvfc1waxm.fsf@boost-consulting.com>
Message-ID: <cnsibg$lna$1@sea.gmane.org>

David Abrahams wrote:

>> Hmm.
>>
>> I'm trying again to reproduce your problem here.  Maybe by comparing
>> our bjam output we can learn something.  FYI, I am going into
>> boost/libs/python/example and running
>>
>>   bjam -d+2 -sTOOLS=vc-7_1 -sBUILD=debug-python test1
>>
> 
> Here's what I got.  Everything seems to have worked perfectly.

Ok, I too builded boost/libs/python/example (using 1.32.0).
Worked perfectly.
Then step by step I extended the example Jamfile/Jamrules towards my
extension.
Worked perfectly.

By comparing with my old Jamrules I found this additional line causing the
trouble:
PYTHON_LIB_PATH = $(PYTHON_ROOT)/libs ;

This line makes no problem with the default build target, so I did not
suspect it.

Thank you for helping
  Stefan



From dave at boost-consulting.com  Mon Nov 22 13:23:17 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Mon, 22 Nov 2004 07:23:17 -0500
Subject: [C++-sig] Re: Embedding/Extending Python at the same time...
References: <20041119163449.48365fd6.hugh-m@moving-picture.com>
	<umzxdpk1u.fsf@boost-consulting.com>
	<20041122091700.5a0ad616.hugh-m@moving-picture.com>
Message-ID: <uis7y2hu2.fsf@boost-consulting.com>

Hugh Macdonald <hugh-m at moving-picture.com> writes:

> On Fri, 19 Nov 2004 17:10:05 -0500
> David Abrahams <dave at boost-consulting.com> wrote:
>
>> Hugh Macdonald <hugh-m at moving-picture.com> writes:
>> 
>> > I posted this on comp.lang.python, but then found this mailing list,
>> > so I hope it's not considered bad etiquette to post in both
>> > places....
>> 
>> Etiquette aside, does this question have anything to do with C++?  It
>> doesn't seem to.
>
> I assumed that this list was to do with extending/embedding Python
> using C++....  If I've missed something out here, then I
> apologise....

It is, but this seemed to be a general question that could apply just
as well to anyone using Python's `C' API.  No big deal.

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From Stefan.Quandt at bardenheuer.de  Mon Nov 22 15:11:56 2004
From: Stefan.Quandt at bardenheuer.de (Stefan Quandt)
Date: Mon, 22 Nov 2004 15:11:56 +0100
Subject: [C++-sig] Re: -sBUILD=debug-python: not solved
References: <cnd6dg$oqu$1@sea.gmane.org> <u7joliwwc.fsf@boost-consulting.com>
	<cneuii$peq$1@sea.gmane.org> <u3bz89t2t.fsf@boost-consulting.com>
	<cnicto$s64$1@sea.gmane.org> <umzxezyrx.fsf@boost-consulting.com>
	<cnk8ej$5u2$1@sea.gmane.org> <usm75pxxn.fsf@boost-consulting.com>
	<uvfc1waxm.fsf@boost-consulting.com>
Message-ID: <cnss3c$juh$1@sea.gmane.org>

David Abrahams wrote:

>> Hmm.
>>
>> I'm trying again to reproduce your problem here.  Maybe by comparing
>> our bjam output we can learn something.  FYI, I am going into
>> boost/libs/python/example and running
>>
>>   bjam -d+2 -sTOOLS=vc-7_1 -sBUILD=debug-python test1
>>
I did the same.
The resulting boost_python.dll was linked against the NON DEBUG version
python23.dll.

I modified three lines in Jamrules like below:
PYTHON_VERSION = 2.3 ;
PYTHON_ROOT = C:/Python-2.3.4 ;
TOOLS = msvc ;

I send you bjam output as attachment.
You will see that the "-DBOOST_PYTHON_DEBUGGING" is missing in the compiler
command lines!

I used MS VC6, boost_1_32_0 and bjam built from the sources.

Kind regards
  Stefan


-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: build.log
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041122/0768e569/attachment.txt>

From dave at boost-consulting.com  Mon Nov 22 18:38:22 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Mon, 22 Nov 2004 12:38:22 -0500
Subject: [C++-sig] Re: -sBUILD=debug-python: not solved
References: <cnd6dg$oqu$1@sea.gmane.org> <u7joliwwc.fsf@boost-consulting.com>
	<cneuii$peq$1@sea.gmane.org> <u3bz89t2t.fsf@boost-consulting.com>
	<cnicto$s64$1@sea.gmane.org> <umzxezyrx.fsf@boost-consulting.com>
	<cnk8ej$5u2$1@sea.gmane.org> <usm75pxxn.fsf@boost-consulting.com>
	<uvfc1waxm.fsf@boost-consulting.com> <cnss3c$juh$1@sea.gmane.org>
Message-ID: <ullctzsvl.fsf@boost-consulting.com>

Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:

>>> Hmm.
>>>
>>> I'm trying again to reproduce your problem here.  Maybe by comparing
>>> our bjam output we can learn something.  FYI, I am going into
>>> boost/libs/python/example and running
>>>
>>>   bjam -d+2 -sTOOLS=vc-7_1 -sBUILD=debug-python test1
>>>
> I did the same.
> The resulting boost_python.dll was linked against the NON DEBUG version
> python23.dll.
>
> I modified three lines in Jamrules like below:
> PYTHON_VERSION = 2.3 ;
> PYTHON_ROOT = C:/Python-2.3.4 ;
> TOOLS = msvc ;
>
> I send you bjam output as attachment.
> You will see that the "-DBOOST_PYTHON_DEBUGGING" is missing in the compiler
> command lines!

-DBOOST_PYTHON_DEBUGGING is never supposed to be in a command line.
I think you mean -DBOOST_DEBUG_PYTHON.

> I used MS VC6, boost_1_32_0 and bjam built from the sources.

Sorry, but either you didn't include the complete output, or there's
no way you did the same, because no extension is being built.  In
fact, it never even tries to link boost_python.dll.  What directory of
your Boost tree were you in, and exactly what command line did you
issue to get this result?

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From jeff.holle at verizon.net  Mon Nov 22 22:20:44 2004
From: jeff.holle at verizon.net (Jeffrey Holle)
Date: Mon, 22 Nov 2004 13:20:44 -0800
Subject: [C++-sig] new warning using vector_indexing_suite
Message-ID: <cntl2d$2vu$1@sea.gmane.org>

I use linux with gcc v3.4.3.

I've discovered that compiling my boost.python using application with 
1.32.0, produces very long winded warnings.

The problem is comparing signed and unsigned longs in 
indexing_suite_detail.hpp.

I've altered this file to avoid these warning, the diff file is attached.

I don't know if this change is correct.

I'd appreciate feedback if deemed not so.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: indexing_suite_detail.hpp.diff
Type: text/x-patch
Size: 280 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041122/fad1a37b/attachment.bin>

From caustin at gmail.com  Tue Nov 23 00:59:33 2004
From: caustin at gmail.com (Chad Austin)
Date: Mon, 22 Nov 2004 17:59:33 -0600
Subject: [C++-sig] new warning using vector_indexing_suite
In-Reply-To: <cntl2d$2vu$1@sea.gmane.org>
References: <cntl2d$2vu$1@sea.gmane.org>
Message-ID: <2096514e041122155961c55fb5@mail.gmail.com>

I just upgraded to Boost 1.32 today and that warning is incredibly
annoying.  I'm surprised the release process didn't catch something
like that.  :(

Chad

On Mon, 22 Nov 2004 13:20:44 -0800, Jeffrey Holle
<jeff.holle at verizon.net> wrote:
> I use linux with gcc v3.4.3.
> 
> I've discovered that compiling my boost.python using application with
> 1.32.0, produces very long winded warnings.
> 
> The problem is comparing signed and unsigned longs in
> indexing_suite_detail.hpp.
> 
> I've altered this file to avoid these warning, the diff file is attached.
> 
> I don't know if this change is correct.
> 
> I'd appreciate feedback if deemed not so.
> 
> 
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
> 
> 
> 
> 


-- 
Chad Austin
http://aegisknight.org/


From ahrivera at yahoo.com  Tue Nov 23 07:56:04 2004
From: ahrivera at yahoo.com (Alexis H. Rivera-Rios)
Date: Mon, 22 Nov 2004 22:56:04 -0800 (PST)
Subject: [C++-sig] understanding pointer ownership with Boost.Python
Message-ID: <20041123065604.63904.qmail@web12825.mail.yahoo.com>

Hi, 

I'm writing some dummy classes to learn how to create
Python classes that are derived from C++ classes. I
have the following class:

class Strategy
{
public:
	virtual ~Strategy() {};
	virtual void Execute() = 0;

};

I want to derive classes from that class using Python
to then send them to this other class:

class MutableObject
{
	std::vector<Strategy *> strategyList;
	Strategy *current;

public:
	MutableObject() {strategyList.push_back(new
NullStrategy()); SelectStrategy(0); }
	~MutableObject() { for (int
i=0;i<strategyList.size();++i) delete strategyList[i];
}


	void AddStrategy(Strategy *s) {
strategyList.push_back(s); }
	size_t GetNumStrategies() { return
strategyList.size(); }
	void SelectStrategy(size_t ith) { current =
strategyList.at(ith).get(); }
	void ExecuteStrategy() { current->Execute(); }
};

I exposed the classes doing the following:
struct StrategyWrap : Strategy, wrapper<Strategy>
{
public:
    void Execute()
    {
        this->get_override("Execute")();
    }
};

BOOST_PYTHON_MODULE(libhello)
{

	class_<StrategyWrap, boost::noncopyable>("Strategy")
    .def("Execute", pure_virtual(&Strategy::Execute))
    ;

	class_<MutableObject>("MutableObject")
    .def("AddStrategy", &MutableObject::AddStrategy,
with_custodian_and_ward<1,2>())
    .def("GetNumStrategies",
&MutableObject::GetNumStrategies)
    .def("SelectStrategy",
&MutableObject::SelectStrategy)
    .def("ExecuteStrategy",
&MutableObject::ExecuteStrategy)
	;
}

I used the with_custodian_and_ward because my
intention is to tell boost.python that whatever
instance is created in python will be owned by the
MutableObject class.  This class will free the memory.
 At least, that is my goal. 

I tested the code with the following script:
from libhello import *

class NewStrat(Strategy):
	def Execute(self):
		print "Another message"

x = MutableObject()

for i in xrange(0,x.GetNumStrategies()):
	x.SelectStrategy(i)
	x.ExecuteStrategy()

y = NewStrat()
x.AddStrategy(y)

for i in xrange(0,x.GetNumStrategies()):
	x.SelectStrategy(i)
	x.ExecuteStrategy()

I get the following error when I execute the script
above:
nothing to do here
nothing to do here
Another message
free(): invalid pointer 0x807f420! <---

All I can deduce is that either the object is being
delete twice. Or the Python object was created using
malloc() and my C++ code tries to delete it using
delete and that's why I get this error.

Is my theory correct? How do I go about fixing it?

I will appreciate your help.
Alexis


=====
Programming Tutorial:
In Python: To do this, do this
In Perl: To do this, do this or this or this or this...
In C: To do this, do this, but be careful
In C++: To do this, do this, but don't do this, be careful of this, watch out for this, and whatever you do, don't do this


		
__________________________________ 
Do you Yahoo!? 
The all-new My Yahoo! - Get yours free! 
http://my.yahoo.com 
 



From Stefan.Quandt at bardenheuer.de  Tue Nov 23 08:49:02 2004
From: Stefan.Quandt at bardenheuer.de (Stefan Quandt)
Date: Tue, 23 Nov 2004 08:49:02 +0100
Subject: [C++-sig] Re: -sBUILD=debug-python
References: <cnd6dg$oqu$1@sea.gmane.org> <u7joliwwc.fsf@boost-consulting.com>
	<cneuii$peq$1@sea.gmane.org> <u3bz89t2t.fsf@boost-consulting.com>
	<cnicto$s64$1@sea.gmane.org> <umzxezyrx.fsf@boost-consulting.com>
	<cnk8ej$5u2$1@sea.gmane.org> <usm75pxxn.fsf@boost-consulting.com>
	<uvfc1waxm.fsf@boost-consulting.com> <cnss3c$juh$1@sea.gmane.org>
	<ullctzsvl.fsf@boost-consulting.com>
Message-ID: <cnuq1e$b4h$1@sea.gmane.org>

David Abrahams wrote:

> Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:
> 
>>>> Hmm.
>>>>
>>>> I'm trying again to reproduce your problem here.  Maybe by comparing
>>>> our bjam output we can learn something.  FYI, I am going into
>>>> boost/libs/python/example and running
>>>>
>>>>   bjam -d+2 -sTOOLS=vc-7_1 -sBUILD=debug-python test1
>>>>
>> I did the same.
>> The resulting boost_python.dll was linked against the NON DEBUG version
>> python23.dll.
>>
>> I modified three lines in Jamrules like below:
>> PYTHON_VERSION = 2.3 ;
>> PYTHON_ROOT = C:/Python-2.3.4 ;
>> TOOLS = msvc ;
>>
>> I send you bjam output as attachment.
>> You will see that the "-DBOOST_PYTHON_DEBUGGING" is missing in the
>> compiler command lines!
> 
> -DBOOST_PYTHON_DEBUGGING is never supposed to be in a command line.
> I think you mean -DBOOST_DEBUG_PYTHON.
Yes.
 
>> I used MS VC6, boost_1_32_0 and bjam built from the sources.
> Sorry, but either you didn't include the complete output, or there's
> no way you did the same, because no extension is being built.  In
> fact, it never even tries to link boost_python.dll.  What directory of
> your Boost tree were you in, and exactly what command line did you
> issue to get this result?
Sorry, I interrupted the build because of the missing
"-DBOOST_DEBUG_PYTHON".
Logs of a complete log encloesd.
Build directory: boost_1_32_0\libs\python\example
Build command: bjam -d+2 -sBUILD=debug-python test1

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: mybuild.log
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041123/08f4d2f7/attachment.txt>

From dave at boost-consulting.com  Tue Nov 23 23:13:41 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 23 Nov 2004 17:13:41 -0500
Subject: [C++-sig] Re: understanding pointer ownership with Boost.Python
References: <20041123065604.63904.qmail@web12825.mail.yahoo.com>
Message-ID: <uk6scnrhm.fsf@boost-consulting.com>

"Alexis H. Rivera-Rios" <ahrivera at yahoo.com> writes:

> I used the with_custodian_and_ward because my
> intention is to tell boost.python that whatever
> instance is created in python will be owned by the
> MutableObject class.  This class will free the memory.

I'm pretty sure the code you posted can't compile.

  class MutableObject
  {
      std::vector<Strategy *> strategyList;
                  ^^^^^^^^^^
      Strategy *current;

  public:
      MutableObject() {strategyList.push_back(new
  NullStrategy()); SelectStrategy(0); }
      ~MutableObject() { for (int
  i=0;i<strategyList.size();++i) delete strategyList[i];
  }


      void AddStrategy(Strategy *s) {
  strategyList.push_back(s); }
      size_t GetNumStrategies() { return
  strategyList.size(); }
      void SelectStrategy(size_t ith) { current =
  strategyList.at(ith).get(); }
                      ^^^^^^
      void ExecuteStrategy() { current->Execute(); }
  };

So I don't believe you actually tested it.  

Your problem is that the result of NewStrat is a Python object that
owns its C++ Strategy subobject.  You need AddStrategy to assume
ownership, because the MutableObject is going to delete the object
later.  So wrap Strategy using

  class_<Strategy, std::auto_ptr<Strategy>...>  

and build a thin wrapper function for AddStrategy that accepts an
auto_ptr:

  void AddStrategy(MutableObject& m, std::auto_ptr<Strategy> s)
  {
       m.AddStrategy(s.get());
       s.release();
  }

      ...

     .def("AddStrategy", &AddStrategy)
      ...


Incidentally, your AddStrategy is probably not very exception-safe to
begin with.

HTH,

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From dave at boost-consulting.com  Tue Nov 23 23:15:26 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 23 Nov 2004 17:15:26 -0500
Subject: [C++-sig] Re: -sBUILD=debug-python
References: <cnd6dg$oqu$1@sea.gmane.org> <u7joliwwc.fsf@boost-consulting.com>
	<cneuii$peq$1@sea.gmane.org> <u3bz89t2t.fsf@boost-consulting.com>
	<cnicto$s64$1@sea.gmane.org> <umzxezyrx.fsf@boost-consulting.com>
	<cnk8ej$5u2$1@sea.gmane.org> <usm75pxxn.fsf@boost-consulting.com>
	<uvfc1waxm.fsf@boost-consulting.com> <cnss3c$juh$1@sea.gmane.org>
	<ullctzsvl.fsf@boost-consulting.com> <cnuq1e$b4h$1@sea.gmane.org>
Message-ID: <uekiknrep.fsf@boost-consulting.com>

Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:

D> David Abrahams wrote:
>
>> Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:
>> 
>>>>> Hmm.
>>>>>
>>>>> I'm trying again to reproduce your problem here.  Maybe by comparing
>>>>> our bjam output we can learn something.  FYI, I am going into
>>>>> boost/libs/python/example and running
>>>>>
>>>>>   bjam -d+2 -sTOOLS=vc-7_1 -sBUILD=debug-python test1
>>>>>
>>> I did the same.
>>> The resulting boost_python.dll was linked against the NON DEBUG version
>>> python23.dll.
>>>
>>> I modified three lines in Jamrules like below:
>>> PYTHON_VERSION = 2.3 ;
>>> PYTHON_ROOT = C:/Python-2.3.4 ;
>>> TOOLS = msvc ;
>>>
>>> I send you bjam output as attachment.
>>> You will see that the "-DBOOST_PYTHON_DEBUGGING" is missing in the
>>> compiler command lines!
>> 
>> -DBOOST_PYTHON_DEBUGGING is never supposed to be in a command line.
>> I think you mean -DBOOST_DEBUG_PYTHON.
> Yes.
>  
>>> I used MS VC6, boost_1_32_0 and bjam built from the sources.
>> Sorry, but either you didn't include the complete output, or there's
>> no way you did the same, because no extension is being built.  In
>> fact, it never even tries to link boost_python.dll.  What directory of
>> your Boost tree were you in, and exactly what command line did you
>> issue to get this result?
> Sorry, I interrupted the build because of the missing
> "-DBOOST_DEBUG_PYTHON".
> Logs of a complete log encloesd.
> Build directory: boost_1_32_0\libs\python\example
> Build command: bjam -d+2 -sBUILD=debug-python test1

I'm without a clue. Do the results look the same with 

   -n -sTOOLS=vc-7_1

?

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From joel at boost-consulting.com  Wed Nov 24 02:23:43 2004
From: joel at boost-consulting.com (Joel de Guzman)
Date: Wed, 24 Nov 2004 09:23:43 +0800
Subject: [C++-sig] Re: new warning using vector_indexing_suite
Message-ID: <41A3E29F.9080606@boost-consulting.com>

Hi,

I'm sorry about that. I'll definitely look into this. I'm not sure
if it is the correct fix. I do not have access to my workstation right
now (working at home), but I'll definitely look into it as soon as I
get the chance.

Regards,
--Joel

------------------------------------------------------------------------
 >
 > Subject:
 > new warning using vector_indexing_suite
 > From:
 > Jeffrey Holle <jeff.holle at verizon.net>
 > Date:
 > Mon, 22 Nov 2004 13:20:44 -0800
 >
 >
 > I use linux with gcc v3.4.3.
 >
 > I've discovered that compiling my boost.python using application with
 > 1.32.0, produces very long winded warnings.
 >
 > The problem is comparing signed and unsigned longs in
 > indexing_suite_detail.hpp.
 >
 > I've altered this file to avoid these warning, the diff file is attached.
 >
 > I don't know if this change is correct.
 >
 > I'd appreciate feedback if deemed not so.
 >
 >
 > ------------------------------------------------------------------------
 >
 > 593c593
 > <                 if (from > max_index) // Clip upper bounds to 
max_index.
 > ---
 >
 >>                if ((unsigned long)from > max_index) // Clip upper 
bounds to max_index.
 >
 > 615c615
 > <                 if (to > max_index)
 > ---
 >
 >>                if ((unsigned long)to > max_index)
 >>
 >>
 >>------------------------------------------------------------------------
 >>
 >>_______________________________________________
 >>C++-sig mailing list
 >>C++-sig at python.org
 >>http://mail.python.org/mailman/listinfo/c++-sig
 >>
 >>
 >> ------------------------------------------------------------------------
 >>
 >> Subject:
 >> Re: new warning using vector_indexing_suite
 >> From:
 >> Chad Austin <caustin at gmail.com>
 >> Date:
 >> Mon, 22 Nov 2004 17:59:33 -0600
 >>
 >>
 >>I just upgraded to Boost 1.32 today and that warning is incredibly
 >>annoying.  I'm surprised the release process didn't catch something
 >>like that.  :(
 >>
 >>Chad
 >>
 >>On Mon, 22 Nov 2004 13:20:44 -0800, Jeffrey Holle
 >><jeff.holle at verizon.net> wrote:
 >>
 >>>I use linux with gcc v3.4.3.
 >>>
 >>>I've discovered that compiling my boost.python using application with
 >>>1.32.0, produces very long winded warnings.
 >>>
 >>>The problem is comparing signed and unsigned longs in
 >>>indexing_suite_detail.hpp.
 >>>
 >>>I've altered this file to avoid these warning, the diff file is 
attached.
 >>>
 >>>I don't know if this change is correct.
 >>>
 >>>I'd appreciate feedback if deemed not so.





From Stefan.Quandt at bardenheuer.de  Wed Nov 24 09:28:27 2004
From: Stefan.Quandt at bardenheuer.de (Stefan Quandt)
Date: Wed, 24 Nov 2004 09:28:27 +0100
Subject: [C++-sig] Re: -sBUILD=debug-python
References: <cnd6dg$oqu$1@sea.gmane.org> <u7joliwwc.fsf@boost-consulting.com>
	<cneuii$peq$1@sea.gmane.org> <u3bz89t2t.fsf@boost-consulting.com>
	<cnicto$s64$1@sea.gmane.org> <umzxezyrx.fsf@boost-consulting.com>
	<cnk8ej$5u2$1@sea.gmane.org> <usm75pxxn.fsf@boost-consulting.com>
	<uvfc1waxm.fsf@boost-consulting.com> <cnss3c$juh$1@sea.gmane.org>
	<ullctzsvl.fsf@boost-consulting.com> <cnuq1e$b4h$1@sea.gmane.org>
	<uekiknrep.fsf@boost-consulting.com>
Message-ID: <co1gnc$3ji$1@sea.gmane.org>

David Abrahams wrote:

> Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:
> 
> D> David Abrahams wrote:
>>
>>> Stefan Quandt <Stefan.Quandt at bardenheuer.de> writes:
>>> 
>>>>>> Hmm.
>>>>>>
>>>>>> I'm trying again to reproduce your problem here.  Maybe by comparing
>>>>>> our bjam output we can learn something.  FYI, I am going into
>>>>>> boost/libs/python/example and running
>>>>>>
>>>>>>   bjam -d+2 -sTOOLS=vc-7_1 -sBUILD=debug-python test1
>>>>>>
>>>> I did the same.
>>>> The resulting boost_python.dll was linked against the NON DEBUG version
>>>> python23.dll.
>>>>
>>>> I modified three lines in Jamrules like below:
>>>> PYTHON_VERSION = 2.3 ;
>>>> PYTHON_ROOT = C:/Python-2.3.4 ;
>>>> TOOLS = msvc ;
>>>>
>>>> I send you bjam output as attachment.
>>>> You will see that the "-DBOOST_PYTHON_DEBUGGING" is missing in the
>>>> compiler command lines!
>>> 
>>> -DBOOST_PYTHON_DEBUGGING is never supposed to be in a command line.
>>> I think you mean -DBOOST_DEBUG_PYTHON.
>> Yes.
>>  
>>>> I used MS VC6, boost_1_32_0 and bjam built from the sources.
>>> Sorry, but either you didn't include the complete output, or there's
>>> no way you did the same, because no extension is being built.  In
>>> fact, it never even tries to link boost_python.dll.  What directory of
>>> your Boost tree were you in, and exactly what command line did you
>>> issue to get this result?
>> Sorry, I interrupted the build because of the missing
>> "-DBOOST_DEBUG_PYTHON".
>> Logs of a complete log encloesd.
>> Build directory: boost_1_32_0\libs\python\example
>> Build command: bjam -d+2 -sBUILD=debug-python test1
> 
> I'm without a clue. Do the results look the same with
> 
>    -n -sTOOLS=vc-7_1
> ?
Specifying TOOLS on the command line has no effect because of the line
  TOOLS=msvc
in my Jamrules file.

After deleting the line all compiler calls fail with
  Command line error D2016 : '/Ze' and '/Za' command-line options are
incompatible
Obviously bjam's tool detecting heuristics fail for MSVC6 (that's why I
originally added the TOOLS definition to my Jamrules).

Nevertheless with specifying "-sTOOLS=msvc" on the command line the
-DBOOST_DEBUG_PYTHON is now present in the compiler command lines!
And the "debug-python" build variant can be build successfully.

Thank you
  Stefan





From beyer at imb-jena.de  Wed Nov 24 09:54:00 2004
From: beyer at imb-jena.de (Andreas Beyer)
Date: Wed, 24 Nov 2004 09:54:00 +0100
Subject: [C++-sig] Re: new warning using vector_indexing_suite
In-Reply-To: <41A3E29F.9080606@boost-consulting.com>
References: <41A3E29F.9080606@boost-consulting.com>
Message-ID: <41A44C28.6080605@imb-jena.de>

Maybe the fix at
http://mail.python.org/pipermail/c++-sig/2004-September/008107.html
helps.

Andreas

Joel de Guzman wrote:
> Hi,
> 
> I'm sorry about that. I'll definitely look into this. I'm not sure
> if it is the correct fix. I do not have access to my workstation right
> now (working at home), but I'll definitely look into it as soon as I
> get the chance.
> 
> Regards,
> --Joel
> 
> ------------------------------------------------------------------------
>  >
>  > Subject:
>  > new warning using vector_indexing_suite
>  > From:
>  > Jeffrey Holle <jeff.holle at verizon.net>
>  > Date:
>  > Mon, 22 Nov 2004 13:20:44 -0800
>  >
>  >
>  > I use linux with gcc v3.4.3.
>  >
>  > I've discovered that compiling my boost.python using application with
>  > 1.32.0, produces very long winded warnings.
>  >
>  > The problem is comparing signed and unsigned longs in
>  > indexing_suite_detail.hpp.
>  >
>  > I've altered this file to avoid these warning, the diff file is 
> attached.
>  >
>  > I don't know if this change is correct.
>  >
>  > I'd appreciate feedback if deemed not so.
>  >
>  >
>  > ------------------------------------------------------------------------
>  >
>  > 593c593
>  > <                 if (from > max_index) // Clip upper bounds to 
> max_index.
>  > ---
>  >
>  >>                if ((unsigned long)from > max_index) // Clip upper 
> bounds to max_index.
>  >
>  > 615c615
>  > <                 if (to > max_index)
>  > ---
>  >
>  >>                if ((unsigned long)to > max_index)
>  >>
>  >>
>  >>------------------------------------------------------------------------
>  >>


-- 
--------------------------------------
   Dr. Andreas Beyer - beyer at imb-jena.de
   http://www.imb-jena.de/~beyer
--------------------------------------


From xavier.warin at der.edfgdf.fr  Wed Nov 24 10:49:52 2004
From: xavier.warin at der.edfgdf.fr (Xavier WARIN (OSIRIS-EDF))
Date: Wed, 24 Nov 2004 10:49:52 +0100
Subject: [C++-sig] VC7 internal compiler error with Boost 1 32 (python)
In-Reply-To: <cnl1bg$cqo$1@sea.gmane.org>
References: <cnl1bg$cqo$1@sea.gmane.org>
Message-ID: <41A45940.3030303@der.edfgdf.fr>

Hi,

I get an internal compiler error with .net2003 while compiling

inheritance.cpp

:\BOOST\boost_1_32_0\boost\mpl\aux_\preprocessed\plain\apply_wrap.hpp(48) 
: fatal error C1001: ERREUR INTERNE DU COMPILATEUR
         (fichier du compilateur 'msc1.cpp', ligne 2701)

I compiles in the debug mode but not in the release mode whatever the 
optimization option i use.
Did anybody get this error (it compiles OK with intel V 8.1). Do you 
know if there is a .net2003 correction available  or a workaround ?
Thank you

Xavier Warin


From vignatov at colorpilot.com  Wed Nov 24 11:09:37 2004
From: vignatov at colorpilot.com (Vladimir Ignatov)
Date: Wed, 24 Nov 2004 13:09:37 +0300
Subject: [C++-sig] VC7 internal compiler error with Boost 1 32 (python)
References: <cnl1bg$cqo$1@sea.gmane.org> <41A45940.3030303@der.edfgdf.fr>
Message-ID: <019d01c4d20d$b6f42d40$8119fea9@100xcd>

Hi!

If memory serve me well, this is a problem in VC 7.0  If boost::python
module code grows beyond some (very small) limit, this error is a result.
You need at least VC 7.1 to work productively.  Don't know about VC 6.

     Vladimir Ignatov


----- Original Message ----- 
From: "Xavier WARIN (OSIRIS-EDF)" <xavier.warin at der.edfgdf.fr>
To: "Development of Python/C++ integration" <c++-sig at python.org>
Sent: Wednesday, November 24, 2004 12:49 PM
Subject: [C++-sig] VC7 internal compiler error with Boost 1 32 (python)


> Hi,
>
> I get an internal compiler error with .net2003 while compiling
>
> inheritance.cpp
>
> :\BOOST\boost_1_32_0\boost\mpl\aux_\preprocessed\plain\apply_wrap.hpp(48)
> : fatal error C1001: ERREUR INTERNE DU COMPILATEUR
>          (fichier du compilateur 'msc1.cpp', ligne 2701)
>
> I compiles in the debug mode but not in the release mode whatever the
> optimization option i use.
> Did anybody get this error (it compiles OK with intel V 8.1). Do you
> know if there is a .net2003 correction available  or a workaround ?
> Thank you
>
> Xavier Warin



From iluetkeb at gmail.com  Wed Nov 24 11:43:54 2004
From: iluetkeb at gmail.com (=?ISO-8859-1?Q?Ingo_L=FCtkebohle?=)
Date: Wed, 24 Nov 2004 11:43:54 +0100
Subject: [C++-sig] embedding: API and signals
Message-ID: <fdd727f804112402432b23b9b3@mail.gmail.com>

Hi,

I'm migrating some code from a custom
not-much-more-than-refcountin-c++ wrapper around the Python API to
boost::python, for embedding.  This works great so far, I'm especially
happy about the automatic type-conversion for stuff I already have
boost-wrappers for.  Thanks a lot for a very usefull library :)

However, a couple of things I noticed:

 * boost::python seems to keep around the Python SIGINT handler by
default.  That means, SIGINT doesn't work while in C++ code.

Q: Would it be possible to have a chaining signal handler or a simple
method to swap Pythons signal handler to SIG_DFL at some chosen
juncture points?

 * boost::python requires the user to have Python.h in its include
path.  I have one use case where this is inconvenient:  All thats
exposed in my class header is a 'handle' (PyObject*), all manipulation
is done in the .cpp.  Ideally, I would only want to include Python.h
in the .cpp so that the user of *my* class doesn't need to have
Python.h around.

Q: Could one forward declare struct PyObject in handle_fwd.hpp or the
like to work around that?

 * the boost::python::object API (and list, tuple, dict, etc) lacks
some fairly basic methods, such as Size on sequences.  I know I can do
   extract<long>(list.attr('__len__')())
but that seems a little awkward and also, I wonder what would happen
if some smarty tries to define __len__ as an attribute?

Q: The Python/C API has convenience methods, could we use them?

 * thread-safety.  I understand this is an old topic but, as far as I
understood it, current solutions are all (with patch) or nothing
(current default).  Given that the user usually knows which parts of
his code could block or are time-consuming...

Q: could this be made configurable, ala return_value_policies?

regards

Ingo


From xin at xlevus.net  Wed Nov 24 12:14:19 2004
From: xin at xlevus.net (Chris Targett)
Date: Wed, 24 Nov 2004 11:14:19 +0000 (UTC)
Subject: [C++-sig] C++ Python Scope Problems
Message-ID: <loom.20041124T121127-94@post.gmane.org>

The Code:
http://orcon.xlevus.net/PyTest.tar.gz
I have included CMake Files and Make Makefiles. Requires boost and python
libraries to compile.

My Problem.
I have my Test C++ module (called 'Test' and exposed it through boost.)
In main.py I have created a pure python equlavient to my `Test` module called
pyTest.
In testScript.py I have a little function (testFunction) that Prints the three
varibles (of the object thats passed to it), increments the values by one, then
prints them again.

Back to main.py. Now when the functions are called to the C++ module, the values
are changed by `testFunction` while inside the scope of the function, but when
returned back to the default scope (what im assuming its called) the Values get
reverted back to their original.

When the same stuff is done on the Pure Python equlaivient, the values remain
changed when reverted back to the default scope.

What I want is, my C++ module to be more like my Python Module, and have the
varibles changed by the function.

Uh, I hope i make sense. 



From dave at boost-consulting.com  Wed Nov 24 14:32:36 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 24 Nov 2004 08:32:36 -0500
Subject: [C++-sig] Re: VC7 internal compiler error with Boost 1 32 (python)
References: <cnl1bg$cqo$1@sea.gmane.org> <41A45940.3030303@der.edfgdf.fr>
	<019d01c4d20d$b6f42d40$8119fea9@100xcd>
Message-ID: <ur7mjl6dn.fsf@boost-consulting.com>

"Vladimir Ignatov" <vignatov at colorpilot.com> writes:

> Hi!
>
> If memory serve me well, this is a problem in VC 7.0  If boost::python
> module code grows beyond some (very small) limit, this error is a result.
> You need at least VC 7.1 to work productively.  Don't know about VC 6.

Please, no wild theories: it builds just fine with both vc7 and vc6 in
release mode, as proven below.  Most likely the OP is using the IDE
and has some broken compiler option set.

See
http://www.boost.org/libs/python/doc/tutorial/doc/html/python/hello.html#id446728
and file:///C:/boost/libs/python/doc/building.html#building_ext


C:\boost\libs\python\build>myjam -sTOOLS=vc7.1 -sBUILD=release inheritance.obj
...patience...
...found 488 targets...
...updating 16 targets...
MkDir1 c:\build
MkDir1 c:\build\bin
MkDir1 c:\build\bin\boost
MkDir1 c:\build\bin\boost\libs
MkDir1 c:\build\bin\boost\libs\python
MkDir1 c:\build\bin\boost\libs\python\build
MkDir1 c:\build\bin\boost\libs\python\build\boost_python.dll
MkDir1 c:\build\bin\boost\libs\python\build\boost_python.dll\vc7.1
MkDir1 c:\build\bin\boost\libs\python\build\boost_python.dll\vc7.1\release
MkDir1 c:\build\bin\boost\libs\python\build\boost_python.dll\vc7.1\release\threading-multi
vc-C++ c:\build\bin\boost\libs\python\build\boost_python.dll\vc7.1\release\threading-multi\inheritance.obj
inheritance.cpp
MkDir1 c:\build\bin\boost\libs\python\build\libboost_python.lib
MkDir1 c:\build\bin\boost\libs\python\build\libboost_python.lib\vc7.1
MkDir1 c:\build\bin\boost\libs\python\build\libboost_python.lib\vc7.1\release
MkDir1 c:\build\bin\boost\libs\python\build\libboost_python.lib\vc7.1\release\threading-multi
vc-C++ c:\build\bin\boost\libs\python\build\libboost_python.lib\vc7.1\release\threading-multi\inheritance.obj
inheritance.cpp
...updated 16 targets...
C:\boost\libs\python\build>myjam -sTOOLS=vc7 -sBUILD=release inheritance.obj
...patience...
...found 488 targets...
...updating 8 targets...
MkDir1 c:\build\bin\boost\libs\python\build\boost_python.dll\vc7
MkDir1 c:\build\bin\boost\libs\python\build\boost_python.dll\vc7\release
MkDir1 c:\build\bin\boost\libs\python\build\boost_python.dll\vc7\release\threading-multi
vc-C++ c:\build\bin\boost\libs\python\build\boost_python.dll\vc7\release\threading-multi\inheritance.obj
inheritance.cpp
MkDir1 c:\build\bin\boost\libs\python\build\libboost_python.lib\vc7
MkDir1 c:\build\bin\boost\libs\python\build\libboost_python.lib\vc7\release
MkDir1 c:\build\bin\boost\libs\python\build\libboost_python.lib\vc7\release\threading-multi
vc-C++ c:\build\bin\boost\libs\python\build\libboost_python.lib\vc7\release\threading-multi\inheritance.obj
inheritance.cpp
...updated 8 targets...
C:\boost\libs\python\build>myjam -sTOOLS=msvc -sBUILD=release inheritance.obj
...found 488 targets...
...updating 8 targets...
MkDir1 c:\build\bin\boost\libs\python\build\boost_python.dll\msvc
MkDir1 c:\build\bin\boost\libs\python\build\boost_python.dll\msvc\release
MkDir1 c:\build\bin\boost\libs\python\build\boost_python.dll\msvc\release\threading-multi
vc-C++ c:\build\bin\boost\libs\python\build\boost_python.dll\msvc\release\threading-multi\inheritance.obj
inheritance.cpp
MkDir1 c:\build\bin\boost\libs\python\build\libboost_python.lib\msvc
MkDir1 c:\build\bin\boost\libs\python\build\libboost_python.lib\msvc\release
MkDir1 c:\build\bin\boost\libs\python\build\libboost_python.lib\msvc\release\threading-multi
vc-C++ c:\build\bin\boost\libs\python\build\libboost_python.lib\msvc\release\threading-multi\inheritance.obj
inheritance.cpp
...updated 8 targets...

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From dave at boost-consulting.com  Wed Nov 24 14:39:55 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 24 Nov 2004 08:39:55 -0500
Subject: [C++-sig] Re: C++ Python Scope Problems
References: <loom.20041124T121127-94@post.gmane.org>
Message-ID: <ufz2zl61g.fsf@boost-consulting.com>

Chris Targett <xin at xlevus.net> writes:

> The Code:
> http://orcon.xlevus.net/PyTest.tar.gz
> I have included CMake Files and Make Makefiles. Requires boost and python
> libraries to compile.
>
> My Problem.
> I have my Test C++ module (called 'Test' and exposed it through boost.)
> In main.py I have created a pure python equlavient to my `Test` module called
> pyTest.
> In testScript.py I have a little function (testFunction) that Prints the three
> varibles (of the object thats passed to it), increments the values by one, then
> prints them again.
>
> Back to main.py. Now when the functions are called to the C++ module, the values
> are changed by `testFunction` while inside the scope of the function, but when
> returned back to the default scope (what im assuming its called) the Values get
> reverted back to their original.
>
> When the same stuff is done on the Pure Python equlaivient, the values remain
> changed when reverted back to the default scope.
>
> What I want is, my C++ module to be more like my Python Module, and have the
> varibles changed by the function.
>
> Uh, I hope i make sense. 

I haven't looked at your archive, but I'm guessing that your problem
is unfixable.  In Python variables don't have their values changed,
they are rebound.  Try:

     python -c "x = 1; print(id(x)); x += 1; print(id(x))"

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



From dave at boost-consulting.com  Wed Nov 24 14:36:17 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 24 Nov 2004 08:36:17 -0500
Subject: [C++-sig] Re: embedding: API and signals
References: <fdd727f804112402432b23b9b3@mail.gmail.com>
Message-ID: <ullcrl67i.fsf@boost-consulting.com>

Ingo L?tkebohle <iluetkeb at gmail.com> writes:

> Hi,
>
> I'm migrating some code from a custom
> not-much-more-than-refcountin-c++ wrapper around the Python API to
> boost::python, for embedding.  This works great so far, I'm especially
> happy about the automatic type-conversion for stuff I already have
> boost-wrappers for.  Thanks a lot for a very usefull library :)
>
> However, a couple of things I noticed:
>
>  * boost::python seems to keep around the Python SIGINT handler by
> default.  That means, SIGINT doesn't work while in C++ code.
>
> Q: Would it be possible to have a chaining signal handler or a simple
> method to swap Pythons signal handler to SIG_DFL at some chosen
> juncture points?

Which chosen juncture points?
Can you supply a patch?

>  * boost::python requires the user to have Python.h in its include
> path.  I have one use case where this is inconvenient:  All thats
> exposed in my class header is a 'handle' (PyObject*), all manipulation
> is done in the .cpp.  Ideally, I would only want to include Python.h
> in the .cpp so that the user of *my* class doesn't need to have
> Python.h around.

Understood.

> Q: Could one forward declare struct PyObject in handle_fwd.hpp or the
> like to work around that?

I seriously doubt it, because of the many inline functions in
Boost.Python.  A rewrite of Boost.Python / Luabind called
Boost.Langbinding is in the works (but on the back burner) that would
allow you to do all wrapping without exposure to Python.h, but
unfortunately it's still nascent.

>  * the boost::python::object API (and list, tuple, dict, etc) lacks
> some fairly basic methods, such as Size on sequences.  I know I can do
>    extract<long>(list.attr('__len__')())
> but that seems a little awkward and also, I wonder what would happen
> if some smarty tries to define __len__ as an attribute?
>
> Q: The Python/C API has convenience methods, could we use them?

Sure, please submit a patch :^).

>  * thread-safety.  I understand this is an old topic but, as far as I
> understood it, current solutions are all (with patch) or nothing
> (current default).  Given that the user usually knows which parts of
> his code could block or are time-consuming...
>
> Q: could this be made configurable, ala return_value_policies?

Ditto.

Sorry, I wish I had more time to work on this project...

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From vignatov at colorpilot.com  Wed Nov 24 15:17:27 2004
From: vignatov at colorpilot.com (Vladimir Ignatov)
Date: Wed, 24 Nov 2004 17:17:27 +0300
Subject: [C++-sig] Re: VC7 internal compiler error with Boost 1 32 (python)
References: <cnl1bg$cqo$1@sea.gmane.org>
	<41A45940.3030303@der.edfgdf.fr><019d01c4d20d$b6f42d40$8119fea9@100xcd>
	<ur7mjl6dn.fsf@boost-consulting.com>
Message-ID: <02e301c4d230$558eedb0$8119fea9@100xcd>

David,

It is not "wild theory", it is "sad practice".  I am talking about building
Python modules using boost::python, not building boost::python library
itself.  Library builds fine but together with VC 7.0 it kind of "unusable"
for me since I can't build any "serious" module with it.   Say, if I have
class with 10 methods, then python module builds fine. But if I add one more
method, then compiler brokes with "internal error" message.  The only
"workaround" is split this module in two or more files. Tested on two
different computers. Repeated with 100% probability.

    Vladimir Ignatov




----- Original Message ----- 
From: "David Abrahams" <dave at boost-consulting.com>
To: <c++-sig at python.org>
Sent: Wednesday, November 24, 2004 4:32 PM
Subject: [C++-sig] Re: VC7 internal compiler error with Boost 1 32 (python)


> "Vladimir Ignatov" <vignatov at colorpilot.com> writes:
>
> > Hi!
> >
> > If memory serve me well, this is a problem in VC 7.0  If boost::python
> > module code grows beyond some (very small) limit, this error is a
result.
> > You need at least VC 7.1 to work productively.  Don't know about VC 6.
>
> Please, no wild theories: it builds just fine with both vc7 and vc6 in
> release mode, as proven below.  Most likely the OP is using the IDE
> and has some broken compiler option set.
>
> See
>
http://www.boost.org/libs/python/doc/tutorial/doc/html/python/hello.html#id446728
> and file:///C:/boost/libs/python/doc/building.html#building_ext
>
>
> C:\boost\libs\python\build>myjam -sTOOLS=vc7.1 -sBUILD=release
inheritance.obj
> ...patience...
> ...found 488 targets...
> ...updating 16 targets...
> MkDir1 c:\build
> MkDir1 c:\build\bin
...
> inheritance.cpp
> ...updated 8 targets...
>
> -- 
> Dave Abrahams
> Boost Consulting
> http://www.boost-consulting.com



From ahrivera at yahoo.com  Wed Nov 24 16:25:56 2004
From: ahrivera at yahoo.com (Alexis H. Rivera-Rios)
Date: Wed, 24 Nov 2004 07:25:56 -0800 (PST)
Subject: [C++-sig] Re: Re: understanding pointer ownership with Boost.Python
Message-ID: <20041124152556.25565.qmail@web12825.mail.yahoo.com>

Hi,

Sorry about the sloppiness I was actually playing with
the shared_ptr inside the std::vector and since it
didn't work I decided to post the message.

You mentioned the AddStrategy is not exception safe. 
Can you explain me why or point me to a resource that
explains the problem and possible solution
alternatives?

Thanks for your help,
Alexis


> I used the with_custodian_and_ward because my
> intention is to tell boost.python that whatever
> instance is created in python will be owned by the
> MutableObject class.  This class will free the
memory.

I'm pretty sure the code you posted can't compile.

  class MutableObject
  {
      std::vector<Strategy *> strategyList;
                  ^^^^^^^^^^
      Strategy *current;

  public:
      MutableObject() {strategyList.push_back(new
  NullStrategy()); SelectStrategy(0); }
      ~MutableObject() { for (int
  i=0;i<strategyList.size();++i) delete
strategyList[i];
  }


      void AddStrategy(Strategy *s) {
  strategyList.push_back(s); }
      size_t GetNumStrategies() { return
  strategyList.size(); }
      void SelectStrategy(size_t ith) { current =
  strategyList.at(ith).get(); }
                      ^^^^^^
      void ExecuteStrategy() { current->Execute(); }
  };

So I don't believe you actually tested it.  

Your problem is that the result of NewStrat is a
Python object that
owns its C++ Strategy subobject.  You need AddStrategy
to assume
ownership, because the MutableObject is going to
delete the object
later.  So wrap Strategy using

  class_<Strategy, std::auto_ptr<Strategy>...>  

and build a thin wrapper function for AddStrategy that
accepts an
auto_ptr:

  void AddStrategy(MutableObject& m,
std::auto_ptr<Strategy> s)
  {
       m.AddStrategy(s.get());
       s.release();
  }

      ...

     .def("AddStrategy", &AddStrategy)
      ...


Incidentally, your AddStrategy is probably not very
exception-safe to
begin with.

HTH,

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com

=====
Programming Tutorial:
In Python: To do this, do this
In Perl: To do this, do this or this or this or this...
In C: To do this, do this, but be careful
In C++: To do this, do this, but don't do this, be careful of this, watch out for this, and whatever you do, don't do this


		
__________________________________ 
Do you Yahoo!? 
Meet the all-new My Yahoo! - Try it today! 
http://my.yahoo.com 
 



From grafik.list at redshift-software.com  Wed Nov 24 17:09:25 2004
From: grafik.list at redshift-software.com (Rene Rivera)
Date: Wed, 24 Nov 2004 10:09:25 -0600
Subject: [C++-sig] Re: -sBUILD=debug-python
In-Reply-To: <co1gnc$3ji$1@sea.gmane.org>
References: <cnd6dg$oqu$1@sea.gmane.org>
	<u7joliwwc.fsf@boost-consulting.com>	<cneuii$peq$1@sea.gmane.org>
	<u3bz89t2t.fsf@boost-consulting.com>	<cnicto$s64$1@sea.gmane.org>
	<umzxezyrx.fsf@boost-consulting.com>	<cnk8ej$5u2$1@sea.gmane.org>
	<usm75pxxn.fsf@boost-consulting.com>	<uvfc1waxm.fsf@boost-consulting.com>
	<cnss3c$juh$1@sea.gmane.org>	<ullctzsvl.fsf@boost-consulting.com>
	<cnuq1e$b4h$1@sea.gmane.org>	<uekiknrep.fsf@boost-consulting.com>
	<co1gnc$3ji$1@sea.gmane.org>
Message-ID: <41A4B235.1040605@redshift-software.com>

Stefan Quandt wrote:

> Specifying TOOLS on the command line has no effect because of the line
>   TOOLS=msvc
> in my Jamrules file.
> 
> After deleting the line all compiler calls fail with
>   Command line error D2016 : '/Ze' and '/Za' command-line options are
> incompatible
> Obviously bjam's tool detecting heuristics fail for MSVC6 (that's why I
> originally added the TOOLS definition to my Jamrules).

There are *no* tool detecting heuristics. You need to specify TOOLS 
somehow, or it will default to vc-7_1 on Windows, gcc otherwise.

> Nevertheless with specifying "-sTOOLS=msvc" on the command line the
> -DBOOST_DEBUG_PYTHON is now present in the compiler command lines!
> And the "debug-python" build variant can be build successfully.

And the reason is that since you specified the TOOLS in Jamrules that is 
loaded after python.jam so it never got a chance to react to the TOOLS 
setting of "msvc".

If you need to set the TOOLS someplace other than the command line, set 
it in boost-build.jam that is the first loaded file.


-- 
-- Grafik - Don't Assume Anything
-- Redshift Software, Inc. - http://redshift-software.com
-- rrivera/acm.org - grafik/redshift-software.com - 102708583/icq


From joel at boost-consulting.com  Wed Nov 24 17:18:30 2004
From: joel at boost-consulting.com (Joel de Guzman)
Date: Thu, 25 Nov 2004 00:18:30 +0800
Subject: [C++-sig] Re: new warning using vector_indexing_suite
In-Reply-To: <41A44C28.6080605@imb-jena.de>
References: <41A3E29F.9080606@boost-consulting.com>
	<41A44C28.6080605@imb-jena.de>
Message-ID: <41A4B456.1030908@boost-consulting.com>

Andreas Beyer wrote:
> Maybe the fix at
> http://mail.python.org/pipermail/c++-sig/2004-September/008107.html
> helps.

Looks good! Thanks. I'll look into this as soon as I can.

--Joel


From dave at boost-consulting.com  Wed Nov 24 17:59:28 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 24 Nov 2004 11:59:28 -0500
Subject: [C++-sig] Re: VC7 internal compiler error with Boost 1 32 (python)
References: <cnl1bg$cqo$1@sea.gmane.org> <41A45940.3030303@der.edfgdf.fr>
	<019d01c4d20d$b6f42d40$8119fea9@100xcd>
	<ur7mjl6dn.fsf@boost-consulting.com>
	<02e301c4d230$558eedb0$8119fea9@100xcd>
Message-ID: <uhdnfgp3j.fsf@boost-consulting.com>

"Vladimir Ignatov" <vignatov at colorpilot.com> writes:

> David,
>
> It is not "wild theory", it is "sad practice".  I am talking about building
> Python modules using boost::python, not building boost::python library
> itself.  

But the OP was trying to build the library.  I appreciate that you
were trying to help, but your theory about the cause of his problem
was highly misleading.  When we say we support VC6 and VC7, we mean
it, and it doesn't do anyone a service to suggest that the library
can't be built or used with those compilers.

> Library builds fine but together with VC 7.0 it kind of "unusable"
> for me since I can't build any "serious" module with it.   Say, if I have
> class with 10 methods, then python module builds fine. But if I add one more
> method, then compiler brokes with "internal error" message.  

Yes, I'm familiar with the issue.

> The only "workaround" is split this module in two or more
> files. Tested on two different computers. Repeated with 100%
> probability.

Right.  It's a completely different problem, though.

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From dave at boost-consulting.com  Wed Nov 24 18:02:40 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 24 Nov 2004 12:02:40 -0500
Subject: [C++-sig] Re: understanding pointer ownership with Boost.Python
References: <20041124152556.25565.qmail@web12825.mail.yahoo.com>
Message-ID: <ubrdngoy7.fsf@boost-consulting.com>

"Alexis H. Rivera-Rios" <ahrivera at yahoo.com> writes:

> You mentioned the AddStrategy is not exception safe. 
> Can you explain me why or point me to a resource that
> explains the problem

Well, *probably* not.  It depends on how you've documented it.

Before it is called, the caller of AddStrategy has responsibility for
deleting the pointer.  After it is called, the target object has
responsibility for deleting the pointer.  Who has responsibility if
the push_back throws an exception?

When answering consider this code:

  x.AddStrategy(new Foo);

where Foo is an appropriate typedef.

> and possible solution alternatives?

Use std::auto_ptr or boost::shared_ptr to manage ownership instead of
passing raw pointers.

HTH,

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From xin at xlevus.net  Wed Nov 24 23:42:39 2004
From: xin at xlevus.net (Chris Targett)
Date: Thu, 25 Nov 2004 11:42:39 +1300
Subject: [C++-sig] Re: C++ Python Scope Problems
References: <loom.20041124T121127-94@post.gmane.org>
	<ufz2zl61g.fsf@boost-consulting.com>
Message-ID: <pan.2004.11.24.22.42.38.108013@xlevus.net>

On Wed, 24 Nov 2004 08:39:55 -0500, David Abrahams wrote:

> Chris Targett <xin at xlevus.net> writes:
> 
>> The Code:
>> http://orcon.xlevus.net/PyTest.tar.gz I have included CMake Files and
>> Make Makefiles. Requires boost and python libraries to compile.
>>
>> My Problem.
>> I have my Test C++ module (called 'Test' and exposed it through boost.)
>> In main.py I have created a pure python equlavient to my `Test` module
>> called pyTest.
>> In testScript.py I have a little function (testFunction) that Prints
>> the three varibles (of the object thats passed to it), increments the
>> values by one, then prints them again.
>>
>> Back to main.py. Now when the functions are called to the C++ module,
>> the values are changed by `testFunction` while inside the scope of the
>> function, but when returned back to the default scope (what im assuming
>> its called) the Values get reverted back to their original.
>>
>> When the same stuff is done on the Pure Python equlaivient, the values
>> remain changed when reverted back to the default scope.
>>
>> What I want is, my C++ module to be more like my Python Module, and
>> have the varibles changed by the function.
>>
>> Uh, I hope i make sense.
> 
> I haven't looked at your archive, but I'm guessing that your problem is
> unfixable.  In Python variables don't have their values changed, they
> are rebound.  Try:
> 
>      python -c "x = 1; print(id(x)); x += 1; print(id(x))"
> 
> HTH,

I see. But if i do:

	import libHg 	# name of C++ library
	foo = libHg.Test()
	print id(foo.var1), foo.var1
	foo.var1 += 1
	print id(foo.var1), foo.var1

I get
	135523772 0
	135523760 1

Which clearley shows the varible is 'rebound'



From rwgk at yahoo.com  Thu Nov 25 00:57:54 2004
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Wed, 24 Nov 2004 15:57:54 -0800 (PST)
Subject: [C++-sig] Re: C++ Python Scope Problems
In-Reply-To: <pan.2004.11.24.22.42.38.108013@xlevus.net>
Message-ID: <20041124235754.62827.qmail@web20227.mail.yahoo.com>

--- Chris Targett <xin at xlevus.net> wrote:

> I see. But if i do:
> 
> 	import libHg 	# name of C++ library
> 	foo = libHg.Test()
> 	print id(foo.var1), foo.var1
> 	foo.var1 += 1
> 	print id(foo.var1), foo.var1
> 
> I get
> 	135523772 0
> 	135523760 1
> 
> Which clearley shows the varible is 'rebound'

I got me worried..., but what you show above works correctly as one would
expect. However, it has not much to do with your original problem. When you
don't get the expected result you are doing something far more complex:

        PyObject *pFunc = getPyFunc("testScript","testFunction");
        call<void>(pFunc, this);

David, please correct me if I am wrong: I think the call<>() must create a
temporary Python object with a copy of "this". The temporary disappears at the
end of the function call.

David, is there a way to create a temporary with a pointer to "this", e.g.
call<void>(pFunc, *this);
?

To the OP: as in real life, there are good and bad neighborhoods. In my
judgement you are in a bad neighborhood. If you call Python from C++ I am
wondering why you are in C++ in the first place. There will not be much of a
speed gain compared to a pure Python solution.
Here is a simple rule that works very well for me:

  Never call Python from C++.

This side-steps a lot of potential surprises like your original problem.
In addition, I'd bet your code will be better organized.

Cheers,
Ralf



	
		
__________________________________ 
Do you Yahoo!? 
Yahoo! Mail - You care about security. So do we. 
http://promotions.yahoo.com/new_mail


From dave at boost-consulting.com  Thu Nov 25 01:19:58 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 24 Nov 2004 19:19:58 -0500
Subject: [C++-sig] Re: C++ Python Scope Problems
References: <loom.20041124T121127-94@post.gmane.org>
	<ufz2zl61g.fsf@boost-consulting.com>
	<pan.2004.11.24.22.42.38.108013@xlevus.net>
Message-ID: <ur7mieq4x.fsf@boost-consulting.com>

Chris Targett <xin at xlevus.net> writes:

>> I haven't looked at your archive, but I'm guessing that your problem is
>> unfixable.  In Python variables don't have their values changed, they
>> are rebound.  Try:
>> 
>>      python -c "x = 1; print(id(x)); x += 1; print(id(x))"
>> 
>> HTH,
>
> I see. But if i do:
>
> 	import libHg 	# name of C++ library
> 	foo = libHg.Test()
> 	print id(foo.var1), foo.var1
> 	foo.var1 += 1
> 	print id(foo.var1), foo.var1
>
> I get
> 	135523772 0
> 	135523760 1
>
> Which clearley shows the varible is 'rebound'

...which was exactly my point.  Not sure what you're illustrating
   here.

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From dave at boost-consulting.com  Thu Nov 25 01:26:26 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 24 Nov 2004 19:26:26 -0500
Subject: [C++-sig] Re: C++ Python Scope Problems
References: <pan.2004.11.24.22.42.38.108013@xlevus.net>
	<20041124235754.62827.qmail@web20227.mail.yahoo.com>
Message-ID: <ullcqepu5.fsf@boost-consulting.com>

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

> --- Chris Targett <xin at xlevus.net> wrote:
>
>> I see. But if i do:
>> 
>> 	import libHg 	# name of C++ library
>> 	foo = libHg.Test()
>> 	print id(foo.var1), foo.var1
>> 	foo.var1 += 1
>> 	print id(foo.var1), foo.var1
>> 
>> I get
>> 	135523772 0
>> 	135523760 1
>> 
>> Which clearley shows the varible is 'rebound'
>
> I got me worried..., but what you show above works correctly as one would
> expect. However, it has not much to do with your original problem. When you
> don't get the expected result you are doing something far more complex:
>
>         PyObject *pFunc = getPyFunc("testScript","testFunction");
>         call<void>(pFunc, this);
>
> David, please correct me if I am wrong: I think the call<>() must create a
> temporary Python object with a copy of "this". The temporary disappears at the
> end of the function call.

Correct.

> David, is there a way to create a temporary with a pointer to "this", e.g.
> call<void>(pFunc, *this);
> ?

Of course:

         call<void>(pFunc, boost::python::ptr(this));

> To the OP: as in real life, there are good and bad neighborhoods. In
> my judgement you are in a bad neighborhood. If you call Python from
> C++ I am wondering why you are in C++ in the first place. There will
> not be much of a speed gain compared to a pure Python solution.

Not everybody does this for speed, though.

> Here is a simple rule that works very well for me:
>
>   Never call Python from C++.

It doesn't work for anyone who has a C++ library with virtual
functions that they want to override from Python.  It is often
acceptable to have your Python subclasses of some base be slow, since
your C++ derived classes will be fast.  If speed matters you can start
moving those subclasses into C++.

Even if you're not writing subclasses, similar arguments apply to any
scenario where you call Python from C++ (there's an isomorphism in
there ;->).

> This side-steps a lot of potential surprises like your original
> problem.  In addition, I'd bet your code will be better organized.

It probably does help to keep things organized, and it's a good idea
when applicable, especially if performance is very important to you.
But it doesn't apply to every valid use case.

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From xin at xlevus.net  Thu Nov 25 05:57:33 2004
From: xin at xlevus.net (Chris Targett)
Date: Thu, 25 Nov 2004 17:57:33 +1300
Subject: [C++-sig] Re: C++ Python Scope Problems
References: <pan.2004.11.24.22.42.38.108013@xlevus.net>
	<20041124235754.62827.qmail@web20227.mail.yahoo.com>
	<ullcqepu5.fsf@boost-consulting.com>
Message-ID: <pan.2004.11.25.04.57.29.297792@xlevus.net>

On Wed, 24 Nov 2004 19:26:26 -0500, David Abrahams wrote:

> "Ralf W. Grosse-Kunstleve" <rwgk at yahoo.com> writes:
> 
>> --- Chris Targett <xin at xlevus.net> wrote:
>>
>>> I see. But if i do:
>>> 
>>> 	import libHg 	# name of C++ library
>>> 	foo = libHg.Test()
>>> 	print id(foo.var1), foo.var1
>>> 	foo.var1 += 1
>>> 	print id(foo.var1), foo.var1
>>> 
>>> I get
>>> 	135523772 0
>>> 	135523760 1
>>> 
>>> Which clearley shows the varible is 'rebound'
>>
>> I got me worried..., but what you show above works correctly as one would
>> expect. However, it has not much to do with your original problem. When you
>> don't get the expected result you are doing something far more complex:
>>
>>         PyObject *pFunc = getPyFunc("testScript","testFunction");
>>         call<void>(pFunc, this);
>>
>> David, please correct me if I am wrong: I think the call<>() must create a
>> temporary Python object with a copy of "this". The temporary disappears at the
>> end of the function call.
> 
> Correct.
> 
>> David, is there a way to create a temporary with a pointer to "this", e.g.
>> call<void>(pFunc, *this);
>> ?
> 
> Of course:
> 
>          call<void>(pFunc, boost::python::ptr(this));

Ah, Exactly what I wanted to do, works a charm.

> 
>> To the OP: as in real life, there are good and bad neighborhoods. In
>> my judgement you are in a bad neighborhood. If you call Python from
>> C++ I am wondering why you are in C++ in the first place. There will
>> not be much of a speed gain compared to a pure Python solution.
> 
> Not everybody does this for speed, though.
> 
>> Here is a simple rule that works very well for me:
>>
>>   Never call Python from C++.
> 
> It doesn't work for anyone who has a C++ library with virtual
> functions that they want to override from Python.  It is often
> acceptable to have your Python subclasses of some base be slow, since
> your C++ derived classes will be fast.  If speed matters you can start
> moving those subclasses into C++.
> 
> Even if you're not writing subclasses, similar arguments apply to any
> scenario where you call Python from C++ (there's an isomorphism in
> there ;->).
> 
>> This side-steps a lot of potential surprises like your original
>> problem.  In addition, I'd bet your code will be better organized.
> 
> It probably does help to keep things organized, and it's a good idea
> when applicable, especially if performance is very important to you.
> But it doesn't apply to every valid use case.

Thankyou, Even if somebody had to ask the question with the right terms
for me, my question was answered. I'm sorry about my badly worded
questions with the wrong term.. But this fixes things. Cheers.



From Stefan.Quandt at bardenheuer.de  Thu Nov 25 09:38:23 2004
From: Stefan.Quandt at bardenheuer.de (Stefan Quandt)
Date: Thu, 25 Nov 2004 09:38:23 +0100
Subject: [C++-sig] RE: -sBUILD=debug-python
References: <cnd6dg$oqu$1@sea.gmane.org>
	<u7joliwwc.fsf@boost-consulting.com>	<cneuii$peq$1@sea.gmane.org>
	<u3bz89t2t.fsf@boost-consulting.com>	<cnicto$s64$1@sea.gmane.org>
	<umzxezyrx.fsf@boost-consulting.com>	<cnk8ej$5u2$1@sea.gmane.org>
	<usm75pxxn.fsf@boost-consulting.com>	<uvfc1waxm.fsf@boost-consulting.com>
	<cnss3c$juh$1@sea.gmane.org>	<ullctzsvl.fsf@boost-consulting.com>
	<cnuq1e$b4h$1@sea.gmane.org>	<uekiknrep.fsf@boost-consulting.com>
	<co1gnc$3ji$1@sea.gmane.org>
	<41A4B235.1040605@redshift-software.com>
Message-ID: <co45lh$em3$1@sea.gmane.org>

Rene Rivera wrote:

> Stefan Quandt wrote:
> 
>> Specifying TOOLS on the command line has no effect because of the line
>>   TOOLS=msvc
>> in my Jamrules file.
>> 
>> After deleting the line all compiler calls fail with
>>   Command line error D2016 : '/Ze' and '/Za' command-line options are
>> incompatible
>> Obviously bjam's tool detecting heuristics fail for MSVC6 (that's why I
>> originally added the TOOLS definition to my Jamrules).
> 
> There are *no* tool detecting heuristics. You need to specify TOOLS
> somehow, or it will default to vc-7_1 on Windows, gcc otherwise.
> 
>> Nevertheless with specifying "-sTOOLS=msvc" on the command line the
>> -DBOOST_DEBUG_PYTHON is now present in the compiler command lines!
>> And the "debug-python" build variant can be build successfully.
> 
> And the reason is that since you specified the TOOLS in Jamrules that is
> loaded after python.jam so it never got a chance to react to the TOOLS
> setting of "msvc".
I suppose that many boost users don't use the default tools.
So specifying TOOLS in a project file seems quite natural to me.

But placing it in the wrong file has bad consequences as my example shows.

> If you need to set the TOOLS someplace other than the command line, set
> it in boost-build.jam that is the first loaded file.
Including a statement like this in the boost documentation would help.

Many users of the boost libraries will not know much about the bjam
mechanics.
The tools/build/v1/build_system.htm describes Jam as a "open-source make
replacement".
I have a strong background with 'make' but in fact bjam seems to work very
different then make.
E.g. in a makefile the order of statements does not matter at all.
The mechanism of overriding a definition by setting it on the command line
seems at least similar.

Kind regards
  Stefan





From xin at xlevus.net  Thu Nov 25 10:15:52 2004
From: xin at xlevus.net (Chris Targett)
Date: Thu, 25 Nov 2004 09:15:52 +0000 (UTC)
Subject: [C++-sig] C++ new/delete with python
Message-ID: <loom.20041125T100808-369@post.gmane.org>

I've got some code.

   class test {
      test(*int);
      int *objects;
   }; 
 
   objects = new int[10];

I want to expose my class `test` to python via Boost, Is there anything special
I have to do to get this to expose properly?

I'm not too fussed about the constructor but if its workable then thats cool.

Is it possible to have a `new` defined array work within python?


Thanks



From joel at boost-consulting.com  Thu Nov 25 10:40:53 2004
From: joel at boost-consulting.com (Joel de Guzman)
Date: Thu, 25 Nov 2004 17:40:53 +0800
Subject: [C++-sig] Re: new warning using vector_indexing_suite
In-Reply-To: <41A4B456.1030908@boost-consulting.com>
References: <41A3E29F.9080606@boost-consulting.com>	<41A44C28.6080605@imb-jena.de>
	<41A4B456.1030908@boost-consulting.com>
Message-ID: <41A5A8A5.9060700@boost-consulting.com>

Joel de Guzman wrote:
> Andreas Beyer wrote:
> 
>> Maybe the fix at
>> http://mail.python.org/pipermail/c++-sig/2004-September/008107.html
>> helps.
> 
> 
> Looks good! Thanks. I'll look into this as soon as I can.

Done. Committed to CVS. Thanks!

-- 
Joel de Guzman
http://www.boost-consulting.com
http://spirit.sf.net


From caustin at gmail.com  Thu Nov 25 11:36:27 2004
From: caustin at gmail.com (Chad Austin)
Date: Thu, 25 Nov 2004 04:36:27 -0600
Subject: [C++-sig] Re: new warning using vector_indexing_suite
In-Reply-To: <41A5A8A5.9060700@boost-consulting.com>
References: <41A3E29F.9080606@boost-consulting.com>
	<41A44C28.6080605@imb-jena.de> <41A4B456.1030908@boost-consulting.com>
	<41A5A8A5.9060700@boost-consulting.com>
Message-ID: <2096514e04112502363f69526@mail.gmail.com>

Awesome.  Is there any chance there could be a point release, boost
1.32.1 or some such?  :)

Chad


On Thu, 25 Nov 2004 17:40:53 +0800, Joel de Guzman
<joel at boost-consulting.com> wrote:
> Joel de Guzman wrote:
> > Andreas Beyer wrote:
> >
> >> Maybe the fix at
> >> http://mail.python.org/pipermail/c++-sig/2004-September/008107.html
> >> helps.
> >
> >
> > Looks good! Thanks. I'll look into this as soon as I can.
> 
> Done. Committed to CVS. Thanks!
> 
> --
> Joel de Guzman
> http://www.boost-consulting.com
> http://spirit.sf.net
> 
> 
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
> 


-- 
Chad Austin
http://aegisknight.org/


From qhf at oce.nl  Fri Nov 26 16:01:42 2004
From: qhf at oce.nl (Fred Houben)
Date: Fri, 26 Nov 2004 16:01:42 +0100
Subject: [C++-sig] problem with overloaded address operator
Message-ID: <002e01c4d3c8$d6a89970$6bf6bc86@oce.nl>

Hello,
I'm using boost.python 1.30.2 with Python 2.3 and msvc6.
I want to wrap a struct that has an object as data member.
The problem is that the address operator has been overloaded and returns a data member address instead of the object address.
Personally I think its' bad practise to overload the address operator (especially in this manner) , but it didn't write the code and removing the operator isn't a option.

I've simplified the problem into a Hello world example:

#include <boost/python.hpp>
using namespace boost::python;

class Hello
{
public:
  char const* greet(){return "Hello, world";}
};

class World
{
public:
 char const* greet(){return hello.greet();}
 Hello* operator&() { return &hello; }
 Hello hello;
};

struct Data
{
 World world;
};

BOOST_PYTHON_MODULE(world)
{
class_<Hello>("Hello")
  .def("greet", &Hello::greet)
  ;
class_<World>("World")
  .def("greet", &World::greet)
  ;
class_<Data>("Data")
  .def_readonly("world", &Data::world)
  ;
}

this result in the following error:
pointer_holder.hpp(177) : error c2440: 'initializing" : cannot convert from 'class Hello *' to 'class World *'

I solved the problem by editing pointer_holder.hpp. Replacing Value* p = &*this->m_p; with boost::addressof(*this->m_p);.
This way I avoid using the address operator. This works fine, but maybe there's something I'm missing here, any input is welcome.
I'm forced to uses boost.python for now 1.30.2, but is this problem solved in other releases?

Thanks,
Fred Houben 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041126/1654db0c/attachment.htm>

From dave at boost-consulting.com  Fri Nov 26 17:11:01 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Fri, 26 Nov 2004 11:11:01 -0500
Subject: [C++-sig] Re: new warning using vector_indexing_suite
References: <41A3E29F.9080606@boost-consulting.com>
	<41A44C28.6080605@imb-jena.de> <41A4B456.1030908@boost-consulting.com>
	<41A5A8A5.9060700@boost-consulting.com>
	<2096514e04112502363f69526@mail.gmail.com>
Message-ID: <uvfbsbnfu.fsf@boost-consulting.com>

Chad Austin <caustin at gmail.com> writes:

> Awesome.  Is there any chance there could be a point release, boost
> 1.32.1 or some such?  :)

We're highly unlikely to do a point release just to kill off some
warnings, unless you're prepared to manage it. :)

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From dave at boost-consulting.com  Fri Nov 26 17:12:55 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Fri, 26 Nov 2004 11:12:55 -0500
Subject: [C++-sig] Re: problem with overloaded address operator
References: <002e01c4d3c8$d6a89970$6bf6bc86@oce.nl>
Message-ID: <upt20bnco.fsf@boost-consulting.com>

"Fred Houben" <qhf at oce.nl> writes:

> I solved the problem by editing pointer_holder.hpp. Replacing Value*
> p = &*this->m_p; with boost::addressof(*this->m_p);.

That's exactly the right fix.

> This way I avoid using the address operator. This works fine, but
> maybe there's something I'm missing here, any input is welcome.

You didn't miss anything.

> I'm forced to uses boost.python for now 1.30.2, but is this problem
> solved in other releases?

It is not.  If you'd like to do a survey of the current CVS for uses
of operator& and submit a patch that replaces all of them, I'd be
happy to apply it, though.

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From caustin at gmail.com  Sat Nov 27 03:17:50 2004
From: caustin at gmail.com (Chad Austin)
Date: Fri, 26 Nov 2004 20:17:50 -0600
Subject: [C++-sig] Re: new warning using vector_indexing_suite
In-Reply-To: <uvfbsbnfu.fsf@boost-consulting.com>
References: <41A3E29F.9080606@boost-consulting.com>
	<41A44C28.6080605@imb-jena.de> <41A4B456.1030908@boost-consulting.com>
	<41A5A8A5.9060700@boost-consulting.com>
	<2096514e04112502363f69526@mail.gmail.com>
	<uvfbsbnfu.fsf@boost-consulting.com>
Message-ID: <2096514e04112618172613c44f@mail.gmail.com>

On Fri, 26 Nov 2004 11:11:01 -0500, David Abrahams
<dave at boost-consulting.com> wrote:
> Chad Austin <caustin at gmail.com> writes:
> 
> > Awesome.  Is there any chance there could be a point release, boost
> > 1.32.1 or some such?  :)
> 
> We're highly unlikely to do a point release just to kill off some
> warnings, unless you're prepared to manage it. :)

Hehe, I'm certainly not prepared to do that...  I guess my real
question is, will the next release take ten months too?  "Release
early, release often" and such.

Ignorant about the boost release process and roadblocks,
Chad


From dave at boost-consulting.com  Sat Nov 27 05:03:45 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Fri, 26 Nov 2004 23:03:45 -0500
Subject: [C++-sig] Re: new warning using vector_indexing_suite
References: <41A3E29F.9080606@boost-consulting.com>
	<41A44C28.6080605@imb-jena.de> <41A4B456.1030908@boost-consulting.com>
	<41A5A8A5.9060700@boost-consulting.com>
	<2096514e04112502363f69526@mail.gmail.com>
	<uvfbsbnfu.fsf@boost-consulting.com>
	<2096514e04112618172613c44f@mail.gmail.com>
Message-ID: <uk6s89bvi.fsf@boost-consulting.com>

Chad Austin <caustin at gmail.com> writes:

> On Fri, 26 Nov 2004 11:11:01 -0500, David Abrahams
> <dave at boost-consulting.com> wrote:
>> Chad Austin <caustin at gmail.com> writes:
>> 
>> > Awesome.  Is there any chance there could be a point release, boost
>> > 1.32.1 or some such?  :)
>> 
>> We're highly unlikely to do a point release just to kill off some
>> warnings, unless you're prepared to manage it. :)
>
> Hehe, I'm certainly not prepared to do that...  I guess my real
> question is, will the next release take ten months too?  "Release
> early, release often" and such.

I don't know.

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From s_sourceforge at nedprod.com  Sat Nov 27 14:57:32 2004
From: s_sourceforge at nedprod.com (Niall Douglas)
Date: Sat, 27 Nov 2004 13:57:32 -0000
Subject: [C++-sig] Re: new warning using vector_indexing_suite
In-Reply-To: <uk6s89bvi.fsf@boost-consulting.com>
Message-ID: <41A887CC.15496.4775EDD9@localhost>

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 26 Nov 2004 at 23:03, David Abrahams wrote:

> >> > Awesome.  Is there any chance there could be a point release,
> >> > boost 1.32.1 or some such?  :)
> >> 
> >> We're highly unlikely to do a point release just to kill off some
> >> warnings, unless you're prepared to manage it. :)
> >
> > Hehe, I'm certainly not prepared to do that...  I guess my real
> > question is, will the next release take ten months too?  "Release
> > early, release often" and such.
> 
> I don't know.

My only thought is Dave, and I'm sure you'll also foresee this, is 
that if unaddressed you're going to be getting posts in here about 
the warnings continuously for the next ten months!

How about a small patch file fixing the warnings on the main page? 
Any volunteers?

Cheers,
Niall





-----BEGIN PGP SIGNATURE-----
Version: idw's PGP-Frontend 4.9.6.1 / 9-2003 + PGP 8.0.2

iQA/AwUBQaiHzMEcvDLFGKbPEQJodwCgqiKeVx3BtogkC3MfGh0Y/KBdIewAni9p
BC6nf6Fr4Qr8q1iLcaZGHE0t
=1zWN
-----END PGP SIGNATURE-----


From dave at boost-consulting.com  Sun Nov 28 01:59:24 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Sat, 27 Nov 2004 19:59:24 -0500
Subject: [C++-sig] Re: new warning using vector_indexing_suite
In-Reply-To: <41A887CC.15496.4775EDD9@localhost>
References: <uk6s89bvi.fsf@boost-consulting.com>
	<41A887CC.15496.4775EDD9@localhost>
Message-ID: <cob7to$f5c$1@sea.gmane.org>

Niall Douglas wrote:

> My only thought is Dave, and I'm sure you'll also foresee this, is 
> that if unaddressed you're going to be getting posts in here about 
> the warnings continuously for the next ten months!

Yeah :(

> How about a small patch file fixing the warnings on the main page? 

Sounds like a great idea.

> Any volunteers?

Ditto!

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com



From dave at boost-consulting.com  Sun Nov 28 02:02:49 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Sat, 27 Nov 2004 20:02:49 -0500
Subject: [C++-sig] Re: problem with overloaded address operator
In-Reply-To: <002e01c4d3c8$d6a89970$6bf6bc86@oce.nl>
References: <002e01c4d3c8$d6a89970$6bf6bc86@oce.nl>
Message-ID: <cob845$f5c$2@sea.gmane.org>

"Fred Houben" <qhf at oce.nl> writes:


 >> I solved the problem by editing pointer_holder.hpp. Replacing Value*
 >> p = &*this->m_p; with boost::addressof(*this->m_p);.


That's exactly the right fix.


 >> This way I avoid using the address operator. This works fine, but
 >> maybe there's something I'm missing here, any input is welcome.


You didn't miss anything.


 >> I'm forced to uses boost.python for now 1.30.2, but is this problem
 >> solved in other releases?


It is not.  If you'd like to do a survey of the current CVS for uses
of operator& and submit a patch that replaces all of them, I'd be
happy to apply it, though.



From pk_silicon2002 at rediffmail.com  Sun Nov 28 21:05:37 2004
From: pk_silicon2002 at rediffmail.com (Prashanta Mohanty)
Date: 28 Nov 2004 20:05:37 -0000
Subject: [C++-sig] Que. C++
Message-ID: <20041128200537.21368.qmail@webmail18.rediffmail.com>

 ?dear,
i want know the answer:
please write a copy constructor for the class given below

class one{
  char **name;
  int no_of_name;
........
.......
};

regards
prashanta
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041128/6b3adb10/attachment.htm>

From jeff.holle at verizon.net  Mon Nov 29 05:01:00 2004
From: jeff.holle at verizon.net (Jeffrey Holle)
Date: Sun, 28 Nov 2004 20:01:00 -0800
Subject: [C++-sig] Re: Que. C++
In-Reply-To: <20041128200537.21368.qmail@webmail18.rediffmail.com>
References: <20041128200537.21368.qmail@webmail18.rediffmail.com>
Message-ID: <41AA9EFC.9000103@verizon.net>

You should have asked your instructor/teacher if you should implement a 
deep or shallow copy!

Prashanta Mohanty wrote:
>   dear,
> i want know the answer:
> please write a copy constructor for the class given below
> 
> class one{
>   char **name;
>   int no_of_name;
> ........
> .......
> };
> 
> regards
> prashanta
> 
> 
> 
> <http://clients.rediff.com/signature/track_sig.asp>
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig


From chencc at csie.nctu.edu.tw  Tue Nov 30 05:39:07 2004
From: chencc at csie.nctu.edu.tw (=?big5?B?s6+s6L/L?=)
Date: Tue, 30 Nov 2004 12:39:07 +0800
Subject: [C++-sig] boost.python tests examples defaults.py
Message-ID: <000a01c4d696$8e2def30$f001a8c0@albertchen>

Hi ...
    I have tried to build the boost.python with VC6 on winXP.  I found there may be something wrong with defaults.py.  I copy the defaults.cpp and defaults.py on another direction. Then, I issued " python defaults.py". There are run-time erorr about memory. I run the command in debug mode, and found there may be error about std::string argument.  However, this run-time error didn't occur under linux.

                                                                                                                                                                                             chencc
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20041130/af96d420/attachment.htm>

From dave at boost-consulting.com  Tue Nov 30 13:04:06 2004
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 30 Nov 2004 07:04:06 -0500
Subject: [C++-sig] Re: boost.python tests examples defaults.py
In-Reply-To: <000a01c4d696$8e2def30$f001a8c0@albertchen>
References: <000a01c4d696$8e2def30$f001a8c0@albertchen>
Message-ID: <cohnk6$okh$1@sea.gmane.org>

?????? wrote:
> Hi ...
>     I have tried to build the boost.python with VC6 on winXP.  I found there may be something wrong with defaults.py.  I copy the defaults.cpp and defaults.py on another direction. Then, I issued " python defaults.py". There are run-time erorr about memory. I run the command in debug mode, and found there may be error about std::string argument.  However, this run-time error didn't occur under linux.

It's hard to say much about your problem without more specific 
information about what you did, what errors there are, etc.

That said, I can assure you that the test is passing just fine on VC6:
http://www.meta-comm.com/engineering/boost-regression/developer/python.html

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com