From roman.yakovenko at gmail.com  Tue Nov  1 06:37:26 2005
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 1 Nov 2005 07:37:26 +0200
Subject: [C++-sig] void* questions\thoughts
In-Reply-To: <uzmope5p8.fsf@boost-consulting.com>
References: <uk6ftfp6s.fsf@boost-consulting.com>
	<436678A5.28245.56992A94@localhost>
	<uzmope5p8.fsf@boost-consulting.com>
Message-ID: <7465b6170510312137i4388ff9aj5cebcc156d70f449@mail.gmail.com>

On 10/31/05, David Abrahams <dave at boost-consulting.com> wrote:
"Niall Douglas" <s_sourceforge at nedprod.com> writes:
> > Would you integrate a patch adding void * as an opaque type if it
> > came with docs, a testcase and ran on vc6?
>
> I'll go you one better.  If you give me everything else -- docs,
> tests, and your best guess at a portable solution -- I'll make sure it
> works on vc6.  The thing is, I suspect 1.33.1 is going to probably
> freeze any day now, so the chance of getting it in is evaporating
> fast.  In fact, only by calling it a regression does it have a hope.
>

I asked my question in order to get better understanding of void*,
beffore I implements
support to it in pyplusplus. ( TnFOX makes an extensive use of it. )

Now it seems that nothing should be done in pyplusplus, am I right?

Thanks

Roman Yakovenko


From dave at boost-consulting.com  Tue Nov  1 11:54:33 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 01 Nov 2005 05:54:33 -0500
Subject: [C++-sig] void* questions\thoughts
References: <uk6ftfp6s.fsf@boost-consulting.com>
	<436678A5.28245.56992A94@localhost>
	<uzmope5p8.fsf@boost-consulting.com>
	<7465b6170510312137i4388ff9aj5cebcc156d70f449@mail.gmail.com>
Message-ID: <uacgobpue.fsf@boost-consulting.com>

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

> On 10/31/05, David Abrahams <dave at boost-consulting.com> wrote:
> "Niall Douglas" <s_sourceforge at nedprod.com> writes:
>> > Would you integrate a patch adding void * as an opaque type if it
>> > came with docs, a testcase and ran on vc6?
>>
>> I'll go you one better.  If you give me everything else -- docs,
>> tests, and your best guess at a portable solution -- I'll make sure it
>> works on vc6.  The thing is, I suspect 1.33.1 is going to probably
>> freeze any day now, so the chance of getting it in is evaporating
>> fast.  In fact, only by calling it a regression does it have a hope.
>>
>
> I asked my question in order to get better understanding of void*,
> beffore I implements
> support to it in pyplusplus. ( TnFOX makes an extensive use of it. )
>
> Now it seems that nothing should be done in pyplusplus, am I right?

We'll see if Niall comes up with the goods.

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



From s_sourceforge at nedprod.com  Tue Nov  1 12:56:51 2005
From: s_sourceforge at nedprod.com (Niall Douglas)
Date: Tue, 01 Nov 2005 11:56:51 -0000
Subject: [C++-sig] void* questions\thoughts
In-Reply-To: <uacgobpue.fsf@boost-consulting.com>
Message-ID: <43675803.787.5A01B0CA@localhost>

On 1 Nov 2005 at 5:54, David Abrahams wrote:

> > I asked my question in order to get better understanding of void*,
> > beffore I implements
> > support to it in pyplusplus. ( TnFOX makes an extensive use of it. )
> >
> > Now it seems that nothing should be done in pyplusplus, am I right?
> 
> We'll see if Niall comes up with the goods.

I will start work on Sunday morning or afternoon and will let you 
know of my progress.

Cheers,
Niall





From roman.yakovenko at gmail.com  Tue Nov  1 15:14:00 2005
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 1 Nov 2005 16:14:00 +0200
Subject: [C++-sig] Pyste & pyplusplus
Message-ID: <7465b6170511010614s3a19a572neeac5296e6d0d23c@mail.gmail.com>

Hi. I have few question to Pyste users.

I am going to create comparison table between Pyste and pyplusplus.
What information do you like to find within the table?

I can think about: user interface, dependencies, design, features
list, performance,
generated code. Something else?

Also an other question is about backward computability. Do you need
pyplusplus to be able
to run Pyste scripts?

Thanks for help.

Roman Yakovenko


From dave at boost-consulting.com  Tue Nov  1 15:58:48 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 01 Nov 2005 09:58:48 -0500
Subject: [C++-sig] c++ bool  to python bool
References: <20050902143531.16d4274b.Nicolas.Rougier@loria.fr>
Message-ID: <u4q6w76tz.fsf@boost-consulting.com>

"Nicolas P. Rougier" <Nicolas.Rougier at loria.fr> writes:

> Hello,
>
> This is certainly a newbie question but how do I get boost python to
> wrap a c++ bool type to a python bool type ?
>
> Here is what I tried:
>
> #include <boost/python.hpp>
> class Boolean {
> public:
>     Boolean (bool v) : value(v) {}
>     bool value;
> };
>
> BOOST_PYTHON_MODULE(boolean) {
>     using namespace boost::python;
>     class_<Boolean>("Boolean", init <bool>() ) 
>         .def_readwrite ("value", &Boolean::value);
> }
>
> But the value property is considered as being an 'int' in python (and
> it's important for me to get the right type). I saw that boost python
> changelog indicates support for python's bool type but I found no
> further documentation. Did I miss something ? 

It works for me; are you sure your Python actually includes bool
support?

        python -c "print bool"

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



From dave at boost-consulting.com  Tue Nov  1 16:17:41 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 01 Nov 2005 10:17:41 -0500
Subject: [C++-sig] how to handle bare references with a shared_ptr
	HeldType
References: <FDC9B82C8E33C346B223298B17C2D34E0E103E@emgmail1.ellington.com>
	<200509221932.33058.meine@kogs.informatik.uni-hamburg.de>
	<uhdcbl8ad.fsf@boost-consulting.com>
	<200509241329.15613.hans_meine@gmx.net>
Message-ID: <uvezc5re2.fsf@boost-consulting.com>

Hans Meine <hans_meine at gmx.net> writes:

> Finally, I propose the attached (not-too-related) patch to the documentation, 
> which basically reorders some HeldType explanations to make their overall 
> relation more clear.

I do appreciate your efforts, but I'm sorry to say that I don't see
how that patch clarifies anything.  If you can explain that, I might
be inclined to apply it.


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



From dave at boost-consulting.com  Tue Nov  1 16:23:14 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 01 Nov 2005 10:23:14 -0500
Subject: [C++-sig] GCC version compatibility
References: <42CDA654.2080106@v.loewis.de> <uu0j6p7z1.fsf@boost-consulting.com>
	<20050708072807.GC3581@lap200.cdc.informatik.tu-darmstadt.de>
	<u8y0hl45u.fsf@boost-consulting.com> <42CEF948.3010908@v.loewis.de>
	<20050709102010.GA3836@lap200.cdc.informatik.tu-darmstadt.de>
	<42D0D215.9000708@v.loewis.de>
	<20050710125458.GA3587@lap200.cdc.informatik.tu-darmstadt.de>
	<42D15DB2.3020300@v.loewis.de>
	<20050716101357.GC3607@lap200.cdc.informatik.tu-darmstadt.de>
	<20051012120917.GA11058@lap200.cdc.informatik.tu-darmstadt.de>
Message-ID: <ufyqg5r4t.fsf@boost-consulting.com>

Christoph Ludwig <cludwig at cdc.informatik.tu-darmstadt.de> writes:

> Hi,
>
> this is to continue a discussion started back in July by a posting by 
> Dave Abrahams <url:http://thread.gmane.org/gmane.comp.python.devel/69651>
> regarding the compiler (C vs. C++) used to compile python's main() and to link
> the executable.
>
>
> On Sat, Jul 16, 2005 at 12:13:58PM +0200, Christoph Ludwig wrote:
>> On Sun, Jul 10, 2005 at 07:41:06PM +0200, "Martin v. L?wis" wrote:
>> > Maybe. For Python 2.4, feel free to contribute a more complex test. For
>> > Python 2.5, I would prefer if the entire code around ccpython.cc was
>> > removed.
>> 
>> I submitted patch #1239112 that implements the test involving two TUs for
>> Python 2.4. I plan to work on a more comprehensive patch for Python 2.5 but
>> that will take some time.
>
>
> I finally had the spare time to look into this problem again and submitted
> patch #1324762. The proposed patch implements the following:

<snip awesome patch>

Christoph, thanks so much for doing this!  Is there anything we should
do in the build system to account for your changes?

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



From dave at boost-consulting.com  Tue Nov  1 16:21:24 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 01 Nov 2005 10:21:24 -0500
Subject: [C++-sig] how to handle bare references with ashared_ptrHeldType
References: <FDC9B82C8E33C346B223298B17C2D34E0E1045@emgmail1.ellington.com>
Message-ID: <umzko5r7v.fsf@boost-consulting.com>

"Gregory Price" <price at ELLINGTON.com> writes:

> Here's another version of that passage:
>
>   HeldType may be a smart pointer to T.  In this case
>   Boost.Python will allow Python T instances to hold
>   either a C++ T instance or a C++ HeldType instance.
>   If HeldType is shared_ptr<T>, a from_python converter
>   will be registered to accept a Python T instance holding 
>   a C++ T instance and return a shared_ptr<T> instance
>   with a custom deleter that decrements the Python refcount.
>
> Is this right?

Yes, although I don't think details about refcounts would be
appropriate at that point in the docs.

> In particular, have I correctly gathered the following
> about how Boost.Python works (and about its terminology):
>
>  When a Boost.Python function is called (in Python),
>  it finds a from_python converter to convert each
>  of the arguments to a C++ object of the required type.
>  Similarly, after invoking the underlying C++ function,
>  a to_python converter is found to convert its return value
>  into a Python object.

Right.

> Where in the documentation is this sort of orientation 
> material found?  (Or where should it be?)

I dunno.  Maybe in the tutorial somewhere?

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



From meine at kogs1.informatik.uni-hamburg.de  Tue Nov  1 17:02:23 2005
From: meine at kogs1.informatik.uni-hamburg.de (Hans Meine)
Date: Tue, 1 Nov 2005 17:02:23 +0100
Subject: [C++-sig] how to handle bare references with a shared_ptr
	HeldType
In-Reply-To: <uvezc5re2.fsf@boost-consulting.com>
References: <FDC9B82C8E33C346B223298B17C2D34E0E103E@emgmail1.ellington.com>
	<200509241329.15613.hans_meine@gmx.net>
	<uvezc5re2.fsf@boost-consulting.com>
Message-ID: <200511011702.23863.meine@kogs.informatik.uni-hamburg.de>

On Tuesday 01 November 2005 16:17, David Abrahams wrote:
> I do appreciate your efforts, but I'm sorry to say that I don't see
> how that patch clarifies anything.  If you can explain that, I might
> be inclined to apply it.

I guess that it does not make things clearer for you, since you know what it 
explains and you have no difficulties in following the explanations; for me 
it makes things clearer simply because all items begin with "if/when 
<condition>", so I get a faster impression of what cases there are and 
whether the whole table is of interest for me.

Ciao, /  /
     /--/
    /  / ANS


From cludwig at cdc.informatik.tu-darmstadt.de  Tue Nov  1 17:15:34 2005
From: cludwig at cdc.informatik.tu-darmstadt.de (Christoph Ludwig)
Date: Tue, 1 Nov 2005 17:15:34 +0100
Subject: [C++-sig] GCC version compatibility
In-Reply-To: <ufyqg5r4t.fsf@boost-consulting.com>
References: <20050708072807.GC3581@lap200.cdc.informatik.tu-darmstadt.de>
	<u8y0hl45u.fsf@boost-consulting.com> <42CEF948.3010908@v.loewis.de>
	<20050709102010.GA3836@lap200.cdc.informatik.tu-darmstadt.de>
	<42D0D215.9000708@v.loewis.de>
	<20050710125458.GA3587@lap200.cdc.informatik.tu-darmstadt.de>
	<42D15DB2.3020300@v.loewis.de>
	<20050716101357.GC3607@lap200.cdc.informatik.tu-darmstadt.de>
	<20051012120917.GA11058@lap200.cdc.informatik.tu-darmstadt.de>
	<ufyqg5r4t.fsf@boost-consulting.com>
Message-ID: <20051101161534.GC8657@lap200.cdc.informatik.tu-darmstadt.de>

On Tue, Nov 01, 2005 at 10:23:14AM -0500, David Abrahams wrote:
> Christoph Ludwig <cludwig at cdc.informatik.tu-darmstadt.de> writes:
> > I finally had the spare time to look into this problem again and submitted
> > patch #1324762. The proposed patch implements the following:
> 
> <snip awesome patch>
> 
> Christoph, thanks so much for doing this! 

You are welcome. It was pretty straight forward once I realized how the C++
compiler is communicated from configure to distutils. But the patch has not
been applied to Python's CVS yet; I don't know how the Python developers think
about it. 

> Is there anything we should
> do in the build system to account for your changes?
> 

You mean in Boost.Build? I don't think so. My patches affect how you configure
Python itself and, by default, Python's main is now compiled and linked with
the C compiler. But AFAICT that does not change how you need to build C++
extension modules.

Regards

Christoph

-- 
http://www.informatik.tu-darmstadt.de/TI/Mitarbeiter/cludwig.html
LiDIA: http://www.informatik.tu-darmstadt.de/TI/LiDIA/Welcome.html



From dave at boost-consulting.com  Tue Nov  1 17:23:20 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 01 Nov 2005 11:23:20 -0500
Subject: [C++-sig] GCC version compatibility
In-Reply-To: <20051101161534.GC8657@lap200.cdc.informatik.tu-darmstadt.de>
	(Christoph Ludwig's message of "Tue, 1 Nov 2005 17:15:34 +0100")
References: <20050708072807.GC3581@lap200.cdc.informatik.tu-darmstadt.de>
	<u8y0hl45u.fsf@boost-consulting.com> <42CEF948.3010908@v.loewis.de>
	<20050709102010.GA3836@lap200.cdc.informatik.tu-darmstadt.de>
	<42D0D215.9000708@v.loewis.de>
	<20050710125458.GA3587@lap200.cdc.informatik.tu-darmstadt.de>
	<42D15DB2.3020300@v.loewis.de>
	<20050716101357.GC3607@lap200.cdc.informatik.tu-darmstadt.de>
	<20051012120917.GA11058@lap200.cdc.informatik.tu-darmstadt.de>
	<ufyqg5r4t.fsf@boost-consulting.com>
	<20051101161534.GC8657@lap200.cdc.informatik.tu-darmstadt.de>
Message-ID: <uacgo49s7.fsf@boost-consulting.com>

Christoph Ludwig <cludwig at cdc.informatik.tu-darmstadt.de> writes:

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

Hmm, maybe I should post encouraging someone to look at this.

>> Is there anything we should
>> do in the build system to account for your changes?
>> 
>
> You mean in Boost.Build? I don't think so. My patches affect how you configure
> Python itself and, by default, Python's main is now compiled and linked with
> the C compiler. But AFAICT that does not change how you need to build C++
> extension modules.

OK, thanks again.
-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com


From dave at boost-consulting.com  Tue Nov  1 17:28:23 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 01 Nov 2005 11:28:23 -0500
Subject: [C++-sig] how to handle bare references with a shared_ptr
	HeldType
References: <FDC9B82C8E33C346B223298B17C2D34E0E103E@emgmail1.ellington.com>
	<200509241329.15613.hans_meine@gmx.net>
	<uvezc5re2.fsf@boost-consulting.com>
	<200511011702.23863.meine@kogs.informatik.uni-hamburg.de>
Message-ID: <uzmoo2uzc.fsf@boost-consulting.com>

Hans Meine <meine at kogs1.informatik.uni-hamburg.de> writes:

> On Tuesday 01 November 2005 16:17, David Abrahams wrote:
>> I do appreciate your efforts, but I'm sorry to say that I don't see
>> how that patch clarifies anything.  If you can explain that, I might
>> be inclined to apply it.
>
> I guess that it does not make things clearer for you, since you know
> what it explains and you have no difficulties in following the
> explanations; for me it makes things clearer simply because all
> items begin with "if/when <condition>", so I get a faster impression
> of what cases there are and whether the whole table is of interest
> for me.

Hmm.  Can we get someone to agree with Hans?  The problem is that of
course Hans' version is clearer to him, just like mine is perfectly
fine for me.

For reference, his patch is attached to
http://article.gmane.org/gmane.comp.python.c++/8631


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



From dave at boost-consulting.com  Tue Nov  1 17:52:44 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 01 Nov 2005 11:52:44 -0500
Subject: [C++-sig] how to handle bare references with a shared_ptr
	HeldType
References: <FDC9B82C8E33C346B223298B17C2D34E0E103E@emgmail1.ellington.com>
	<200509221932.33058.meine@kogs.informatik.uni-hamburg.de>
	<uhdcbl8ad.fsf@boost-consulting.com>
	<200509241329.15613.hans_meine@gmx.net>
Message-ID: <uslug2tur.fsf@boost-consulting.com>

Hans Meine <hans_meine at gmx.net> writes:

> On Saturday 24 September 2005 00:38, David Abrahams wrote:
>> Hans Meine <meine at kogs1.informatik.uni-hamburg.de> writes:
>> > OTOH, if you use return_internal_reference, BP should know that creating
>> > a shared_ptr with a refcount of 1 is evil..  is that really what it does?
>>
>> No, that's not what it does.  When you use return_internal_reference
>> it creates a Python object holding the C++ object with a raw pointer.
>> Isn't that in the documentation?
>
> That's how I understood it in the past, but I am not experienced with 
> customized HeldTypes, and Gregory seemed to be concerned that it does so:
>
>> > On Thursday 22 September 2005 16:37, Gregory Price wrote:
>> >> Is there a way to tell Boost.Python not to make a shared_ptr
>> >> for this return value -- to override the HeldType just for this
>> >> function's return value?
>
> So the answer really is: Use return_internal_reference.
>
> Concerning the documentation question:  The class_ documentation reads 
> "Heldtype [...] Specifies the type which is actually embedded in a Python 
> object wrapping a T instance. More details below."
> So I would've assumed that every T wrapper actually contains a smart_ptr if 
> that's the HeldType.  One could then make more clear that it's not always the 
> case.

Clarified, thanks.

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



From rwgk at yahoo.com  Tue Nov  1 23:00:25 2005
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Tue, 1 Nov 2005 14:00:25 -0800 (PST)
Subject: [C++-sig] new stl_iterator doesn't work with EDGs
Message-ID: <20051101220025.20032.qmail@web31505.mail.mud.yahoo.com>

The output below is an error message produced by Tru64 cxx 7.1, which is based
on EDG 304. I got a similar error message with EDG 238 under IRIX. I guess it
probably is the same for all EDGs in between.

I'd be happy to try out ideas for fixes. We have EDG 238, 245 and 304
available.

Cheers,
        Ralf

BTW: I just checked in a patch to reorder the includes. Due to Python
requirements, boost/python includes have to appear before any system includes.


cxx -o boost/libs/python/test/stl_iterator.o -c -std strict_ansi
-msg_display_number -msg_disable 186,450,1115,1133 -tlocal
-DBOOST_DISABLE_THREADS -DNDEBUG -O2 -fast -DBOOST_PYTHON_MAX_BASES=2
-I/net/legless/scratch1/rwgk/hot/boost
-I/usr/local_cci/Python-2.4.2_ticker_simple/include/python2.4
/net/legless/scratch1/rwgk/hot/boost/libs/python/test/stl_iterator.cpp
cxx: Error:
/net/legless/scratch1/rwgk/hot/boost/boost/iterator/iterator_facade.hpp, line
535: #265-D 
          function
          "boost::python::stl_input_iterator<ValueT>::equal [with ValueT=int]"
          is inaccessible
          detected during:
            instantiation of "bool boost::iterator_core_access::equal(const
                      Facade1 &, const Facade2 &, mpl_::true_) [with
                      Facade1=boost::python::stl_input_iterator<int>,
                      Facade2=boost::python::stl_input_iterator<int>]" at line
                      838
            instantiation of "boost::detail::enable_if_interoperable<Derived1,
                      Derived2,
                      boost::mpl::apply2<boost::detail::always_bool2,
                      Derived1, Derived2>::type>::type boost::operator!=(const
                      boost::iterator_facade<Derived1, V1, TC1, R1, D1> &,
                      const boost::iterator_facade<Derived2, V2, TC2, R2, D2>
                      &) [with
                      Derived1=boost::python::stl_input_iterator<int>, V1=int,
                      TC1=std::input_iterator_tag, R1=int,
                      D1=std::ptrdiff_t={signed long},
                      Derived2=boost::python::stl_input_iterator<int>, V2=int,
                      TC2=std::input_iterator_tag, R2=int,
                      D2=std::ptrdiff_t={signed long}]" at line 845 of
                      "/usr/lib/cmplrs/cxx/V7.1-006/include/cxx/list"
            instantiation of "void std::list<T,
                      Allocator>::_RWinsert_interval_dispatch(std::list<T,
                      Allocator>::iterator, InputIterator, InputIterator,
                      std::input_iterator_tag) [with T=int,
                      Allocator=std::allocator<int>,
                      InputIterator=boost::python::stl_input_iterator<int>]"
                      at line 156 of
                      "/usr/lib/cmplrs/cxx/V7.1-006/include/cxx/list.cc"
            instantiation of "void std::list<T,
                      Allocator>::insert(std::list<T, Allocator>::iterator,
                      InputIterator, InputIterator) [with T=int,
                      Allocator=std::allocator<int>,
                      InputIterator=boost::python::stl_input_iterator<int>]" 
          return f1.equal(f2);
--------------------^
cxx: Error:
/net/legless/scratch1/rwgk/hot/boost/boost/iterator/iterator_facade.hpp, line
517: #265-D 
          function "boost::python::stl_input_iterator<ValueT>::dereference
          [with ValueT=int]" is inaccessible
          detected during:
            instantiation of "Facade::reference
                      boost::iterator_core_access::dereference(const Facade &)
                      [with Facade=boost::python::stl_input_iterator<int>]" at
                      line 635
            instantiation of "boost::iterator_facade<Derived, Value,
                      CategoryOrTraversal, Reference, Difference>::reference
                      boost::iterator_facade<Derived, Value,
                      CategoryOrTraversal, Reference, Difference>::operator*()
                      const [with
                      Derived=boost::python::stl_input_iterator<int>,
                      Value=int, CategoryOrTraversal=std::input_iterator_tag,
                      Reference=int, Difference=std::ptrdiff_t={signed long}]"
                      at line 846 of
                      "/usr/lib/cmplrs/cxx/V7.1-006/include/cxx/list"
            instantiation of "void std::list<T,
                      Allocator>::_RWinsert_interval_dispatch(std::list<T,
                      Allocator>::iterator, InputIterator, InputIterator,
                      std::input_iterator_tag) [with T=int,
                      Allocator=std::allocator<int>,
                      InputIterator=boost::python::stl_input_iterator<int>]"
                      at line 156 of
                      "/usr/lib/cmplrs/cxx/V7.1-006/include/cxx/list.cc"
            instantiation of "void std::list<T,
                      Allocator>::insert(std::list<T, Allocator>::iterator,
                      InputIterator, InputIterator) [with T=int,
                      Allocator=std::allocator<int>,
                      InputIterator=boost::python::stl_input_iterator<int>]" 
          return f.dereference();
-------------------^
cxx: Error:
/net/legless/scratch1/rwgk/hot/boost/boost/iterator/iterator_facade.hpp, line
523: #265-D 
          function "boost::python::stl_input_iterator<ValueT>::increment [with
          ValueT=int]" is inaccessible
          detected during:
            instantiation of "void
                      boost::iterator_core_access::increment(Facade &) [with
                      Facade=boost::python::stl_input_iterator<int>]" at line
                      665
            instantiation of "Derived &boost::iterator_facade<Derived, Value,
                      CategoryOrTraversal, Reference,
                      Difference>::operator++() [with
                      Derived=boost::python::stl_input_iterator<int>,
                      Value=int, CategoryOrTraversal=std::input_iterator_tag,
                      Reference=int, Difference=std::ptrdiff_t={signed long}]"
                      at line 848 of
                      "/usr/lib/cmplrs/cxx/V7.1-006/include/cxx/list"
            instantiation of "void std::list<T,
                      Allocator>::_RWinsert_interval_dispatch(std::list<T,
                      Allocator>::iterator, InputIterator, InputIterator,
                      std::input_iterator_tag) [with T=int,
                      Allocator=std::allocator<int>,
                      InputIterator=boost::python::stl_input_iterator<int>]"
                      at line 156 of
                      "/usr/lib/cmplrs/cxx/V7.1-006/include/cxx/list.cc"
            instantiation of "void std::list<T,
                      Allocator>::insert(std::list<T, Allocator>::iterator,
                      InputIterator, InputIterator) [with T=int,
                      Allocator=std::allocator<int>,
                      InputIterator=boost::python::stl_input_iterator<int>]" 
          f.increment();
------------^
cxx: Info: 3 errors detected in the compilation of
"/net/legless/scratch1/rwgk/hot/boost/libs/python/test/stl_iterator.cpp".
scons: *** [boost/libs/python/test/stl_iterator.o] Error 1



	
		
__________________________________ 
Yahoo! Mail - PC Magazine Editors' Choice 2005 
http://mail.yahoo.com


From eric at boost-consulting.com  Wed Nov  2 19:03:51 2005
From: eric at boost-consulting.com (Eric Niebler)
Date: Wed, 02 Nov 2005 10:03:51 -0800
Subject: [C++-sig] new stl_iterator doesn't work with EDGs
In-Reply-To: <20051101220025.20032.qmail@web31505.mail.mud.yahoo.com>
References: <20051101220025.20032.qmail@web31505.mail.mud.yahoo.com>
Message-ID: <4368FF87.8060409@boost-consulting.com>


Ralf W. Grosse-Kunstleve wrote:
> cxx: Error:
> /net/legless/scratch1/rwgk/hot/boost/boost/iterator/iterator_facade.hpp, line
> 535: #265-D 
>           function
>           "boost::python::stl_input_iterator<ValueT>::equal [with ValueT=int]"
>           is inaccessible
>           detected during:
>             instantiation of "bool boost::iterator_core_access::equal


Odd. EDG doesn't seem to be picking up the friend declaration. The 
attached version works around the problem. Can you give it a shot?


> BTW: I just checked in a patch to reorder the includes. Due to Python
> requirements, boost/python includes have to appear before any system includes.

Thanks.


-- 
Eric Niebler
Boost Consulting
www.boost-consulting.com
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: stl_iterator.hpp
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051102/cf635693/attachment.txt>

From rwgk at yahoo.com  Wed Nov  2 23:21:07 2005
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Wed, 2 Nov 2005 14:21:07 -0800 (PST)
Subject: [C++-sig] new stl_iterator doesn't work with EDGs
In-Reply-To: <4368FF87.8060409@boost-consulting.com>
Message-ID: <20051102222107.33128.qmail@web31502.mail.mud.yahoo.com>

--- Eric Niebler <eric at boost-consulting.com> wrote:
> Odd. EDG doesn't seem to be picking up the friend declaration. The 
> attached version works around the problem. Can you give it a shot?
> ...
> #if !BOOST_WORKAROUND(__EDG_VERSION__, BOOST_TESTED_AT(304))
> private:
>     friend class iterator_core_access;
> #endif

Your patch works (tested with EDG 304) but this one works (tested with EDG 238
and 304), too:

Index: stl_iterator.hpp
===================================================================
RCS file: /cvsroot/boost/boost/boost/python/stl_iterator.hpp,v
retrieving revision 1.1
diff -r1.1 stl_iterator.hpp
39c39
<     friend class iterator_core_access;
---
>     friend class boost::iterator_core_access;


With the original version of stl_iterator.hpp EDG seems to "think" the friend
is boost::python::iterator_core_access (which doesn't exist). Is this a bug in
the EDG name lookup? Or is it a non-standard feature that it works with g++?

Let me know which patch you prefer and I'll test it on more platforms.

Cheers,
        Ralf



		
__________________________________ 
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com


From amohr at pixar.com  Wed Nov  2 23:22:23 2005
From: amohr at pixar.com (Alex Mohr)
Date: Wed, 02 Nov 2005 14:22:23 -0800
Subject: [C++-sig] Something like boost::any ... custom conversions?
Message-ID: <43693C1F.9090101@pixar.com>

Suppose I have a type that is much like boost::any.  If I have an 
instance of this in python (say holding an int), then I want to be able 
to call wrapped c++ code which takes arguments of held types (say 
foo(int)) by passing my any-like object directly.

That is, I want my any-like object to try to convert to the desired c++ 
type by checking to see if it's holding the desired c++ type.

Can anyone point me in a general direction of where to look to do this 
sort of thing?  I've tried looking at the conversion information in the 
reference manual, and I've looked at container_conversions.h, but I have 
to admit that things are still muddy for me.

The lvalue_from_pytype stuff seems like it might be in the right 
direction, but I don't really want lvalue conversion.  I only care about 
passing these things as rvalues.

Thanks a lot,

Alex


From eric at boost-consulting.com  Wed Nov  2 23:48:27 2005
From: eric at boost-consulting.com (Eric Niebler)
Date: Wed, 02 Nov 2005 14:48:27 -0800
Subject: [C++-sig] new stl_iterator doesn't work with EDGs
In-Reply-To: <20051102222107.33128.qmail@web31502.mail.mud.yahoo.com>
References: <4368FF87.8060409@boost-consulting.com>
	<20051102222107.33128.qmail@web31502.mail.mud.yahoo.com>
Message-ID: <4369423B.8000906@boost-consulting.com>


Ralf W. Grosse-Kunstleve wrote:
> 
> Your patch works (tested with EDG 304) but this one works (tested with EDG 238
> and 304), too:
> 
> Index: stl_iterator.hpp
> ===================================================================
> RCS file: /cvsroot/boost/boost/boost/python/stl_iterator.hpp,v
> retrieving revision 1.1
> diff -r1.1 stl_iterator.hpp
> 39c39
> <     friend class iterator_core_access;
> ---
> 
> >     friend class boost::iterator_core_access;
> 

Better. This is the right fix, so I went ahead and committed it. I 
tested it with VC7.1 and gcc, and the regression tests will tell us if 
any other compilers have a problem with it.

Thanks!

-- 
Eric Niebler
Boost Consulting
www.boost-consulting.com


From seefeld at sympatico.ca  Thu Nov  3 02:38:23 2005
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Wed, 02 Nov 2005 20:38:23 -0500
Subject: [C++-sig] Something like boost::any ... custom conversions?
In-Reply-To: <43693C1F.9090101@pixar.com>
References: <43693C1F.9090101@pixar.com>
Message-ID: <43696A0F.6060102@sympatico.ca>

Alex Mohr wrote:
> Suppose I have a type that is much like boost::any.  If I have an 
> instance of this in python (say holding an int), then I want to be able 
> to call wrapped c++ code which takes arguments of held types (say 
> foo(int)) by passing my any-like object directly.

Out of curiosity, why don't you just let python do the type dispatching
for you ? It seems strange to work in a hybrid python/C++ world and then
use a tool that adds dynamic (i.e. runtime-) type dispatching to C++.

Thanks,
		Stefan


From tmaisnam at yahoo.com  Thu Nov  3 03:40:26 2005
From: tmaisnam at yahoo.com (T Maisnam)
Date: Wed, 2 Nov 2005 18:40:26 -0800 (PST)
Subject: [C++-sig] help on quant finance code
Message-ID: <20051103024027.75182.qmail@web30809.mail.mud.yahoo.com>

Hi Holger,
I am currently studying a financial engineering course
in sydney and am from an IT background. I came across
your name in one of the bulletin boards. Can you
please send me any links to financial program codes
that I can look into. I am never worked in finance and
i want to see some code implementation besides the
abstract math concepts that I am studying now.

cheers,
Tony Maisnam




	
		
__________________________________ 
Yahoo! Mail - PC Magazine Editors' Choice 2005 
http://mail.yahoo.com


From raghavendra_hpt at yahoo.co.in  Thu Nov  3 14:39:12 2005
From: raghavendra_hpt at yahoo.co.in (Raghavendra K . B)
Date: Thu, 3 Nov 2005 13:39:12 +0000 (GMT)
Subject: [C++-sig] Warning: non-POD class type passed through
Message-ID: <20051103133912.89821.qmail@web8401.mail.in.yahoo.com>

Hi All,
 
I am getting following Warning : non-POD class type passed through
 
  buf.Format(KKeyCreateTime, rate, noKeyPairCreations, keyCreateTime);
                                                                                    ^
 
 If any one has faced this problem , please let me know the solution .I badly need solution for this 
 
Thanks in Advance.
 
Raghavendra K.B

		
---------------------------------
 Enjoy this Diwali with Y! India Click here
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051103/b92ccf3f/attachment.htm>

From dave at boost-consulting.com  Thu Nov  3 15:13:48 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Thu, 03 Nov 2005 09:13:48 -0500
Subject: [C++-sig] Warning: non-POD class type passed through
References: <20051103133912.89821.qmail@web8401.mail.in.yahoo.com>
Message-ID: <ufyqdrf8j.fsf@boost-consulting.com>

"Raghavendra K . B" <raghavendra_hpt at yahoo.co.in> writes:

> Hi All,
>  
> I am getting following Warning : non-POD class type passed through
>  
>   buf.Format(KKeyCreateTime, rate, noKeyPairCreations, keyCreateTime);
>                                                                                     ^
>  
>  If any one has faced this problem , please let me know the solution .I badly need solution for this 

You are asking in the wrong list.  Followups should go to
comp.lang.c++ or comp.lang.c++.moderated.

To save you some trouble, Format almost certainly has a ... in its
signature.  If you pass a non-POD
(http://www.fnal.gov/docs/working-groups/fpcltf/Pkg/ISOcxx/doc/POD.html)
through ... you get undefined behavior.  Don't do that.

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



From amohr at pixar.com  Thu Nov  3 18:45:44 2005
From: amohr at pixar.com (Alex Mohr)
Date: Thu, 03 Nov 2005 09:45:44 -0800
Subject: [C++-sig] Something like boost::any ... custom conversions?
In-Reply-To: <43696A0F.6060102@sympatico.ca>
References: <43693C1F.9090101@pixar.com> <43696A0F.6060102@sympatico.ca>
Message-ID: <436A4CC8.40805@pixar.com>

Thanks for your reply.

> Out of curiosity, why don't you just let python do the type dispatching
> for you ? It seems strange to work in a hybrid python/C++ world and then
> use a tool that adds dynamic (i.e. runtime-) type dispatching to C++.

I can see where you're coming from, but unfortunately this is in a large 
existing C++ system which was not originally written to be used with 
Python and already makes use this any-like object.

Alex


From Oystein.Lekang at idi.ntnu.no  Thu Nov  3 19:17:44 2005
From: Oystein.Lekang at idi.ntnu.no (=?ISO-8859-1?Q?=D8ystein_Lekang?=)
Date: Thu, 3 Nov 2005 19:17:44 +0100 (MET)
Subject: [C++-sig] using boost.python to have access to boost::dynamic_bitset
Message-ID: <Pine.GSO.4.60.0511031910520.24624@selje.idi.ntnu.no>


I am in need of somthing like dynamic_bitset in python, and
am currently searching for the best way to achieve this.

So Im wondering how much work it would be to use boost.python
to make dynamic_bitset accessable to python, and whether 
it can be done by me. (myself not beeing to versed in c++)

I've had a brief look at swig, pyrex and pyinline, but
thought that since boost.python also is a part of boost,
it might be the way to go.


-- 
?ystein Lekang

From dave at boost-consulting.com  Thu Nov  3 23:35:00 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Thu, 03 Nov 2005 17:35:00 -0500
Subject: [C++-sig] Something like boost::any ... custom conversions?
References: <43693C1F.9090101@pixar.com>
Message-ID: <u64r9l5rf.fsf@boost-consulting.com>

Alex Mohr <amohr at pixar.com> writes:

> Suppose I have a type that is much like boost::any.  If I have an 
> instance of this in python (say holding an int), then I want to be able 
> to call wrapped c++ code which takes arguments of held types (say 
> foo(int)) by passing my any-like object directly.
>
> That is, I want my any-like object to try to convert to the desired c++ 
> type by checking to see if it's holding the desired c++ type.

You can do it for specific held types just by registering the
appropriate from_python converters:

> Can anyone point me in a general direction of where to look to do this 
> sort of thing?  I've tried looking at the conversion information in the 
> reference manual, and I've looked at container_conversions.h, but I have 
> to admit that things are still muddy for me.
>
> The lvalue_from_pytype stuff seems like it might be in the right 
> direction, but I don't really want lvalue conversion.  I only care about 
> passing these things as rvalues.

In general, if you can get ahold of an lvalue, it's usually a good
idea to do it anyway.  The bigger issue is that you don't have the
Python type object associated with your any-like class at compile
time.  But yeah, it's in the right direction.

I suggest you look at
http://news.gmane.org/find-root.php?message_id=%3c1026918209.17167.2.camel%40hunsmac.mit.edu%3e
and follow the links there.  It's probably about time to document this
stuff for real and pull it out of the detail:: namespaces.

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



From cramissor at yahoo.com  Thu Nov  3 23:39:19 2005
From: cramissor at yahoo.com (Marc Rossi)
Date: Thu, 3 Nov 2005 14:39:19 -0800 (PST)
Subject: [C++-sig] boost package import error
Message-ID: <20051103223919.49677.qmail@web34109.mail.mud.yahoo.com>

I'm running into import problems trying to create a
package that is similiar in layout to the one in the
docs described here:

http://www.boost.org/libs/python/doc/tutorial/doc/html/python/techniques.html#python.creating_packages

So my package layout is as follows:

    xyzzy/
        __init__.py
	research/
            __init__.py
            _research.so
	    stats.pyc

My __init__.py file contains 2 lines:

    from stats import *
    from _research import *

With the directory containing 'xyzzy' on my
PYTHONPATH, the following occurs:

    Python 2.4.2 (#2, Nov  3 2005, 11:35:03)
    [GCC 3.2.3 20030502 (Red Hat Linux 3.2.3-52)] on
linux2
    Type "help", "copyright", "credits" or "license"
for more       information.
    >>> import frfr.research
    Traceback (most recent call last):
      File "<stdin>", line 1, in ?
      File "xyzzy/research/__init__.py", line 2, in ?
        from _research import *
    ImportError: No module named _research

Running python with the -vv switch shows python does
indeed look at _research.so but does not load it.

However if I put the directory containing _research.so
on my PYTHONPATH the following command is successful:

    >>> import _research

Thanks for any help you can offer,

Marc



		
__________________________________ 
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com


From seefeld at sympatico.ca  Fri Nov  4 00:07:24 2005
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Thu, 03 Nov 2005 18:07:24 -0500
Subject: [C++-sig] boost package import error
In-Reply-To: <20051103223919.49677.qmail@web34109.mail.mud.yahoo.com>
References: <20051103223919.49677.qmail@web34109.mail.mud.yahoo.com>
Message-ID: <436A982C.1030905@sympatico.ca>

Marc Rossi wrote:

> Running python with the -vv switch shows python does
> indeed look at _research.so but does not load it.
> 
> However if I put the directory containing _research.so
> on my PYTHONPATH the following command is successful:
> 
>     >>> import _research

This is not related to C++ (or boost): As described
in http://docs.python.org/ref/import.html, the '*' wildcard
only applies to public names, if not specified otherwise
via __all__. Rename '_research' to 'research_impl' (for example)
and you will be fine.

Regards,
		Stefan


From ricardokirkner at gmail.com  Fri Nov  4 18:39:07 2005
From: ricardokirkner at gmail.com (Ricardo Kirkner)
Date: Fri, 4 Nov 2005 14:39:07 -0300
Subject: [C++-sig] boost.python error
Message-ID: <5e44108d0511040939s504be27bv2ac8b98feecfe4c8@mail.gmail.com>

Hi:

I just wrote a small wrapper for a class, using pyste (boost.python). I am
just trying to figure out how this works, before I get on wrapping the whole
real library.

I did run pyste succesfully (which generated a .cpp file) , and I also did
run bjam successfully, which generated the .so library. However, when I try
to import the newly generated library in python, I get the following error:

ImportError: ./opal.so: undefined symbol:
_ZTVN5boost6python7objects21py_function_impl_baseE

Can anyone tell me why is this error being produced (and how to fix it)?
Does this mean that the .so didnt linked to the boost library?

Thank you,

Ricardo Kirkner
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051104/021337ac/attachment.htm>

From rwgk at yahoo.com  Fri Nov  4 20:15:55 2005
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Fri, 4 Nov 2005 11:15:55 -0800 (PST)
Subject: [C++-sig] boost.python error
In-Reply-To: <5e44108d0511040939s504be27bv2ac8b98feecfe4c8@mail.gmail.com>
Message-ID: <20051104191555.47536.qmail@web31513.mail.mud.yahoo.com>

--- Ricardo Kirkner <ricardokirkner at gmail.com> wrote:
> ImportError: ./opal.so: undefined symbol:
> _ZTVN5boost6python7objects21py_function_impl_baseE
> 
> Can anyone tell me why is this error being produced (and how to fix it)?
> Does this mean that the .so didnt linked to the boost library?

I suspect that you have to point LD_LIBRARY_PATH to the directory with with
libboost_python.so. I don't use bjam, therefore I cannot help with the details,
but maybe you can get it going via "find . -name libboost_python.so" and then
"setenv LD_LIBRARY_PATH <path-you-found>:$LD_LIBRARY_PATH".

HTH,
        Ralf



		
__________________________________ 
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com


From ricardokirkner at gmail.com  Fri Nov  4 21:50:20 2005
From: ricardokirkner at gmail.com (Ricardo Kirkner)
Date: Fri, 4 Nov 2005 17:50:20 -0300
Subject: [C++-sig] boost.python error
In-Reply-To: <20051104191555.47536.qmail@web31513.mail.mud.yahoo.com>
References: <5e44108d0511040939s504be27bv2ac8b98feecfe4c8@mail.gmail.com>
	<20051104191555.47536.qmail@web31513.mail.mud.yahoo.com>
Message-ID: <5e44108d0511041250j1dd19632j687ddebc91a35ca1@mail.gmail.com>

Hi:

could you please share with me some example of how you compile/link your
libraries, not using bjam?

Maybe that will help me out.

Thank you,

Ricardo Kirkner

On 11/4/05, Ralf W. Grosse-Kunstleve <rwgk at yahoo.com> wrote:
>
> --- Ricardo Kirkner <ricardokirkner at gmail.com> wrote:
> > ImportError: ./opal.so: undefined symbol:
> > _ZTVN5boost6python7objects21py_function_impl_baseE
> >
> > Can anyone tell me why is this error being produced (and how to fix it)?
> > Does this mean that the .so didnt linked to the boost library?
>
> I suspect that you have to point LD_LIBRARY_PATH to the directory with
> with
> libboost_python.so. I don't use bjam, therefore I cannot help with the
> details,
> but maybe you can get it going via "find . -name libboost_python.so" and
> then
> "setenv LD_LIBRARY_PATH <path-you-found>:$LD_LIBRARY_PATH".
>
> HTH,
> Ralf
>
>
>
>
> __________________________________
> Yahoo! FareChase: Search multiple travel sites in one click.
> http://farechase.yahoo.com
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051104/f293e223/attachment.htm>

From rwgk at yahoo.com  Sat Nov  5 00:11:37 2005
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Fri, 4 Nov 2005 15:11:37 -0800 (PST)
Subject: [C++-sig] boost.python error
In-Reply-To: <5e44108d0511041250j1dd19632j687ddebc91a35ca1@mail.gmail.com>
Message-ID: <20051104231137.86525.qmail@web31512.mail.mud.yahoo.com>

--- Ricardo Kirkner <ricardokirkner at gmail.com> wrote:
> could you please share with me some example of how you compile/link your
> libraries, not using bjam?

Look at any of the "build_log" pages here to see the actual commands:

http://cci.lbl.gov/cctbx_build/show_error_summary.cgi?build_tag=2005_10_21_1046

Everything is open source:

http://cci.lbl.gov/cctbx_build/

To get the commands for your particular machine, get a source bundle, run "perl
cctbx_bundle.selfx", copy and paste the commands for compiling and linking
libboost_python.so (near the beginning) and throw away the rest (rm -rf).

Note that you will always have to worry about LD_LIBRARY_PATH; this is how you
tell dlopen (used by Python) where to find your custom shared libraries.

Cheers,
        Ralf



		
__________________________________ 
Start your day with Yahoo! - Make it your home page! 
http://www.yahoo.com/r/hs


From dave at boost-consulting.com  Sat Nov  5 13:33:28 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Sat, 05 Nov 2005 07:33:28 -0500
Subject: [C++-sig] boost.python error
References: <5e44108d0511041250j1dd19632j687ddebc91a35ca1@mail.gmail.com>
	<20051104231137.86525.qmail@web31512.mail.mud.yahoo.com>
Message-ID: <umzkjb7fr.fsf@boost-consulting.com>

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

> Note that you will always have to worry about LD_LIBRARY_PATH; this is how you
> tell dlopen (used by Python) where to find your custom shared libraries.

Unless you use bjam to run Python as a test case.  It sets up
LD_LIBRARY_PATH for you ;-)

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



From seefeld at sympatico.ca  Sat Nov  5 15:30:49 2005
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Sat, 05 Nov 2005 09:30:49 -0500
Subject: [C++-sig] boost.python error
In-Reply-To: <20051104231137.86525.qmail@web31512.mail.mud.yahoo.com>
References: <20051104231137.86525.qmail@web31512.mail.mud.yahoo.com>
Message-ID: <436CC219.6020304@sympatico.ca>

Ralf W. Grosse-Kunstleve wrote:

> Note that you will always have to worry about LD_LIBRARY_PATH; this is how you
> tell dlopen (used by Python) where to find your custom shared libraries.

just a little nit-picking:

The library being searched for here (libboost_python.so) isn't dlopened, only
the depending python module is. Python has its own way to figure out from
where to load its modules, and that doesn't involve LD_LIBRARY_PATH.
However, if those modules depend on other DSOs, the dynamic loader (ld.so
on my system) kicks in, and *then* you may need LD_LIBRARY_PATH to add
non-default places to look in.

Regards,
		Stefan


From ricardokirkner at gmail.com  Sun Nov  6 03:44:09 2005
From: ricardokirkner at gmail.com (Ricardo Kirkner)
Date: Sat, 5 Nov 2005 23:44:09 -0300
Subject: [C++-sig] boost.python error
In-Reply-To: <436CC219.6020304@sympatico.ca>
References: <20051104231137.86525.qmail@web31512.mail.mud.yahoo.com>
	<436CC219.6020304@sympatico.ca>
Message-ID: <5e44108d0511051844rdaafb4fxe4e6fd971056afb8@mail.gmail.com>

Hi:

I tried to compile my module several ways. I managed to get pass the
undefined symbol error, by compiling the module directly (not using
bjam). However, now I am getting the following error:

ImportError: dynamic module does not define init function (initopal)

I believe that the problem with bjam might be a Jamfile/Jamrules
configuration error. (I could correctly compile the hello world
example)

Please, if anyone knows a little more about this, give me a hand :-)

Regards,

Ricardo Kirkner

On 11/5/05, Stefan Seefeld <seefeld at sympatico.ca> wrote:
> Ralf W. Grosse-Kunstleve wrote:
>
> > Note that you will always have to worry about LD_LIBRARY_PATH; this is how you
> > tell dlopen (used by Python) where to find your custom shared libraries.
>
> just a little nit-picking:
>
> The library being searched for here (libboost_python.so) isn't dlopened, only
> the depending python module is. Python has its own way to figure out from
> where to load its modules, and that doesn't involve LD_LIBRARY_PATH.
> However, if those modules depend on other DSOs, the dynamic loader (ld.so
> on my system) kicks in, and *then* you may need LD_LIBRARY_PATH to add
> non-default places to look in.
>
> Regards,
>                 Stefan
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>


From rwgk at yahoo.com  Sun Nov  6 16:50:49 2005
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Sun, 6 Nov 2005 07:50:49 -0800 (PST)
Subject: [C++-sig] boost.python error
In-Reply-To: <436CC219.6020304@sympatico.ca>
Message-ID: <20051106155049.18362.qmail@web31502.mail.mud.yahoo.com>

--- Stefan Seefeld <seefeld at sympatico.ca> wrote:

> Ralf W. Grosse-Kunstleve wrote:
> 
> > Note that you will always have to worry about LD_LIBRARY_PATH; this is how
> you
> > tell dlopen (used by Python) where to find your custom shared libraries.
> 
> just a little nit-picking:
> 
> The library being searched for here (libboost_python.so) isn't dlopened, only
> the depending python module is. Python has its own way to figure out from
> where to load its modules, and that doesn't involve LD_LIBRARY_PATH.
> However, if those modules depend on other DSOs, the dynamic loader (ld.so
> on my system) kicks in, and *then* you may need LD_LIBRARY_PATH to add
> non-default places to look in.

Thanks for the clarification; I agree.
I noticed that the Linux man page for dlopen mentions LD_LIBRARY_PATH right at
the beginning. Therefore I was thinking the dynamic loader uses dlopen
eventually, just like Python does. Or is there another mechanism?

Cheers,
        Ralf


	
		
__________________________________ 
Yahoo! Mail - PC Magazine Editors' Choice 2005 
http://mail.yahoo.com


From rwgk at yahoo.com  Sun Nov  6 16:54:21 2005
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Sun, 6 Nov 2005 07:54:21 -0800 (PST)
Subject: [C++-sig] boost.python error
In-Reply-To: <5e44108d0511051844rdaafb4fxe4e6fd971056afb8@mail.gmail.com>
Message-ID: <20051106155421.96407.qmail@web31509.mail.mud.yahoo.com>

--- Ricardo Kirkner <ricardokirkner at gmail.com> wrote:
> ImportError: dynamic module does not define init function (initopal)

If you want to enable "import opal" you need to write

BOOST_PYTHON_MODULE(opal)

in your Boost.Python extension.

Cheers,
        Ralf



		
__________________________________ 
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com


From ricardokirkner at gmail.com  Sun Nov  6 17:13:15 2005
From: ricardokirkner at gmail.com (Ricardo Kirkner)
Date: Sun, 6 Nov 2005 13:13:15 -0300
Subject: [C++-sig] boost.python error
In-Reply-To: <20051106155421.96407.qmail@web31509.mail.mud.yahoo.com>
References: <5e44108d0511051844rdaafb4fxe4e6fd971056afb8@mail.gmail.com>
	<20051106155421.96407.qmail@web31509.mail.mud.yahoo.com>
Message-ID: <5e44108d0511060813w1e42b226jab48e624bf79442a@mail.gmail.com>

I give you the code returned by pyste (which I am using to build the
wrapper). Maybe you can tell me whats wrong with it. I also include
the original source (OpalMath.h)

I also post my Jamfile and Jamrules files.

When I build with bjam, I get an error telling me

ImportError: ./OpalMath.so: undefined symbol:
_ZTVN5boost6python7objects21py_function_impl_baseE

when I build using g++ directly I get the error

ImportError: dynamic module does not define init function (initOpalMath)

To build using g++ I use

g++ -c -Wall -fPIC -I/usr/include/boost -I/usr/include/python2.3 -Isrc
-o OpalMath.o OpalMath.cpp
g++ -shared -fPIC -Wl --no-undefined --warn-once -L/usr/lib
-lpython2.3 -lboost_python -DBOOST_PYTHON_DYNAMIC_MODULE -o
OpalMath.so OpalMath.o

I hope this info can tell you something about why I am not succeeding
in building this very small wrapper...

Thanks in advance

Ricardo Kirkner

On 11/6/05, Ralf W. Grosse-Kunstleve <rwgk at yahoo.com> wrote:
> --- Ricardo Kirkner <ricardokirkner at gmail.com> wrote:
> > ImportError: dynamic module does not define init function (initopal)
>
> If you want to enable "import opal" you need to write
>
> BOOST_PYTHON_MODULE(opal)
>
> in your Boost.Python extension.
>
> Cheers,
>         Ralf
>
>
>
>
> __________________________________
> Yahoo! FareChase: Search multiple travel sites in one click.
> http://farechase.yahoo.com
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: OpalMath.cpp
Type: text/x-c++src
Size: 934 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051106/ed159a1c/attachment.cpp>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: OpalMath.h
Type: text/x-chdr
Size: 3640 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051106/ed159a1c/attachment.h>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Jamfile
Type: application/octet-stream
Size: 421 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051106/ed159a1c/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Jamrules
Type: application/octet-stream
Size: 1580 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051106/ed159a1c/attachment-0001.obj>

From seefeld at sympatico.ca  Sun Nov  6 17:12:10 2005
From: seefeld at sympatico.ca (Stefan Seefeld)
Date: Sun, 06 Nov 2005 11:12:10 -0500
Subject: [C++-sig] boost.python error
In-Reply-To: <20051106155049.18362.qmail@web31502.mail.mud.yahoo.com>
References: <20051106155049.18362.qmail@web31502.mail.mud.yahoo.com>
Message-ID: <436E2B5A.5080708@sympatico.ca>

Ralf W. Grosse-Kunstleve wrote:

> Thanks for the clarification; I agree.
> I noticed that the Linux man page for dlopen mentions LD_LIBRARY_PATH right at
> the beginning. Therefore I was thinking the dynamic loader uses dlopen
> eventually, just like Python does. Or is there another mechanism?

dlopen uses LD_LIBRARY_PATH only if the path doesn't contain '/', which is
not the case for python imports (i.e. python's import function makes sure
the path either starts with '/' or './').

As for ld.so using dlopen, I was under the impression that dlopen is merely
a means for users (i.e. developers) to access the dynamic loader, not the
other way around.
I'm not familiar enough with the details concerning symbol resolution and
initialization, which may be different for both cases.

Regards,
		Stefan


From s_sourceforge at nedprod.com  Sun Nov  6 18:55:28 2005
From: s_sourceforge at nedprod.com (Niall Douglas)
Date: Sun, 06 Nov 2005 17:55:28 -0000
Subject: [C++-sig] Patch implementing void * support
Message-ID: <436E4390.14657.548C6F9@localhost>

The enclosed patch adds void * support to Boost.Python by making void 
an opaque type. Code that uses void * "just works" with this patch 
with no further effort required.

Also enclosed is a testcase and the docs have been modified.

Ran full testsuite on msvc7.1 and msvc6 - everything works. I don't 
have GCC to hand ATM, but apart from maybe a missing typename 
specifier I can't see anything which would fault.

Cheers,
Niall




-------------- next part --------------
Only in boost: bin
diff -aur boost_1_33_0/boost/python/converter/registered.hpp boost/boost/python/converter/registered.hpp
--- boost_1_33_0/boost/python/converter/registered.hpp	2005-05-18 02:34:36.000000000 +0100
+++ boost/boost/python/converter/registered.hpp	2005-11-06 15:18:14.000000000 +0000
@@ -77,15 +77,35 @@
   }
   
   template <class T>
-  registration const& 
-  registry_lookup(T&(*)())
+  inline registration const& 
+  registry_lookup2(T&(*)())
   {
       detail::register_shared_ptr1((T*)0);
       return registry::lookup(type_id<T&>());
   }
 
   template <class T>
-  registration const& registered_base<T>::converters = detail::registry_lookup((T(*)())0);
+  inline registration const& 
+  registry_lookup1()
+  {
+      return registry_lookup2((T(*)())0);
+  }
+
+  template <>
+  inline registration const& 
+#ifdef BOOST_NO_CV_VOID_SPECIALIZATIONS
+  registry_lookup1<void>()
+#else
+  registry_lookup1<const volatile void>()
+#endif
+  {
+      detail::register_shared_ptr1((void*)0);
+      return registry::lookup(type_id<void>());
+  }
+
+  template <class T>
+	  registration const& registered_base<T>::converters = detail::registry_lookup1<T>();
+
 }
 
 }}} // namespace boost::python::converter
diff -aur boost_1_33_0/boost/python/lvalue_from_pytype.hpp boost/boost/python/lvalue_from_pytype.hpp
--- boost_1_33_0/boost/python/lvalue_from_pytype.hpp	2004-08-20 12:09:16.000000000 +0100
+++ boost/boost/python/lvalue_from_pytype.hpp	2005-11-06 16:53:14.000000000 +0000
@@ -15,12 +15,25 @@
 
 namespace detail
 {
+  struct voidptrholder;		// Used to indicate a void * is being extracted
+
+  template <class T>
+  inline type_info extractor_type_id2()
+  {
+      return type_id<T>();
+  }
+  template <>
+  inline type_info extractor_type_id2<voidptrholder>()
+  {
+      return type_id<void>();
+  }
+
   // Given a pointer-to-function of 1 parameter returning a reference
   // type, return the type_id of the function's return type.
   template <class T, class U>
   inline type_info extractor_type_id(T&(*)(U))
   {
-      return type_id<T>();
+      return extractor_type_id2<T>();
   }
 
   // A function generator whose static execute() function is an lvalue
diff -aur boost_1_33_0/boost/python/opaque_pointer_converter.hpp boost/boost/python/opaque_pointer_converter.hpp
--- boost_1_33_0/boost/python/opaque_pointer_converter.hpp	2004-08-20 12:09:16.000000000 +0100
+++ boost/boost/python/opaque_pointer_converter.hpp	2005-11-06 17:16:25.000000000 +0000
@@ -15,6 +15,7 @@
 # include <boost/python/detail/none.hpp>
 # include <boost/type_traits/remove_pointer.hpp>
 # include <boost/type_traits/is_pointer.hpp>
+# include <boost/type_traits/is_void.hpp>
 
 // opaque_pointer_converter --
 //
@@ -54,6 +55,15 @@
         , detail::opaque_pointer_converter_requires_a_pointer_type<Pointer>
     >::type ptr_type;
 
+	typedef typename ::boost::remove_pointer<ptr_type>::type noptr_type;
+	typedef typename mpl::if_c<
+		::boost::is_void<typename noptr_type>::value,
+		detail::voidptrholder,
+		typename noptr_type
+	>::type fixed_type;
+
+    typedef typename ::boost::add_reference<typename fixed_type>::type ref_type;
+
 private:
     struct instance;
 
@@ -84,10 +94,10 @@
         return (result);
     }
 
-    static typename ::boost::remove_pointer<ptr_type>::type&
+    static typename ref_type
     execute(instance &p_)
     {
-        return *p_.x;
+        return *(typename fixed_type *) p_.x;
     }
 
 private:
@@ -132,4 +142,8 @@
     }                                                                           \
 }}
 #  endif
+
+// Declare void as an opaque type for automatic void * usage
+BOOST_PYTHON_OPAQUE_SPECIALIZED_TYPE_ID(void)
+
 # endif    // OPAQUE_POINTER_CONVERTER_HPP_
Only in boost/libs/python/build/VisualStudio: Debug
Only in boost_1_33_0/libs/python/build/VisualStudio: boost_python.dsp
Only in boost_1_33_0/libs/python/build/VisualStudio: boost_python.dsw
Only in boost/libs/python/build/VisualStudio: boost_python.ncb
Only in boost/libs/python/build/VisualStudio: boost_python.sln
Only in boost/libs/python/build/VisualStudio: boost_python.suo
Only in boost/libs/python/build/VisualStudio: boost_python.vcproj
Only in boost/libs/python/build: bin-stage
diff -aur boost_1_33_0/libs/python/doc/v2/faq.html boost/libs/python/doc/v2/faq.html
--- boost_1_33_0/libs/python/doc/v2/faq.html	2004-11-28 03:54:58.000000000 +0000
+++ boost/libs/python/doc/v2/faq.html	2005-11-06 17:17:31.000000000 +0000
@@ -67,8 +67,6 @@
       >error C2064: term does not evaluate to a function taking 2 arguments</a>
       </dt>
 
-      <dt><a href="#voidptr">How do I handle <tt>void *</tt> conversion?</a></dt>
-
       <dt><a href="#custom_string"
       >How can I automatically convert my custom string type to
        and from a Python string?</a></dt>
@@ -694,29 +692,6 @@
     <p>(The bug has been reported to Microsoft.)</p>
 
     <hr>
-    <h2><a name="voidptr"></a>How do I handle <tt>void *</tt> conversion?</h2>
-    <font size="-1"><i>Niall Douglas provides these notes:</i></font><p>
-    For several reasons Boost.Python does not support <tt>void *</tt> as
-    an argument or as a return value. However, it is possible to wrap
-    functions with <tt>void *</tt> arguments or return values using
-    thin wrappers and the <i>opaque pointer</i> facility. E.g.:
-    <pre>// Declare the following in each translation unit
-struct void_ {};
-BOOST_PYTHON_OPAQUE_SPECIALIZED_TYPE_ID(void_);
-
-void *foo(int par1, void *par2);
-
-void_ *foo_wrapper(int par1, void_ *par2)
-{
-    return (void_ *) foo(par1, par2);
-}
-...
-BOOST_PYTHON_MODULE(bar)
-{
-    def("foo", &amp;foo_wrapper);
-}</pre>
-
-    <hr>
     <h2><a name="custom_string"></a>How can I automatically
     convert my custom string type to and from a Python string?</h2>
     <font size="-1"><i>Ralf W. Grosse-Kunstleve provides these
diff -aur boost_1_33_0/libs/python/test/Jamfile boost/libs/python/test/Jamfile
--- boost_1_33_0/libs/python/test/Jamfile	2005-07-07 15:00:30.000000000 +0100
+++ boost/libs/python/test/Jamfile	2005-11-06 15:28:29.000000000 +0000
@@ -151,6 +151,7 @@
 [ bpl-test extract ]
 
 [ bpl-test opaque ]
+[ bpl-test voidptr ]
 
 [ bpl-test pickle1 ]
 [ bpl-test pickle2 ]
Only in boost/libs/python/test: bjam.exe
Only in boost/libs/python/test: hello.txt
Only in boost/libs/python/test: voidptr
Only in boost/libs/python/test: voidptr.cpp
Only in boost/libs/python/test: voidptr.py
Only in boost/tools/build/jam_src: bin.ntx86
Only in boost/tools/build/jam_src: bootstrap
Only in boost/tools/build/jam_src: builtins.obj
Only in boost/tools/build/jam_src: class.obj
Only in boost/tools/build/jam_src: command.obj
Only in boost/tools/build/jam_src: compile.obj
Only in boost/tools/build/jam_src: execnt.obj
Only in boost/tools/build/jam_src: execunix.obj
Only in boost/tools/build/jam_src: execvms.obj
Only in boost/tools/build/jam_src: expand.obj
Only in boost/tools/build/jam_src: filent.obj
Only in boost/tools/build/jam_src: fileos2.obj
Only in boost/tools/build/jam_src: filesys.obj
Only in boost/tools/build/jam_src: fileunix.obj
Only in boost/tools/build/jam_src: filevms.obj
Only in boost/tools/build/jam_src: glob.obj
Only in boost/tools/build/jam_src: hash.obj
Only in boost/tools/build/jam_src: hcache.obj
Only in boost/tools/build/jam_src: hdrmacro.obj
Only in boost/tools/build/jam_src: headers.obj
Only in boost/tools/build/jam_src: jam.obj
Only in boost/tools/build/jam_src: jambase.obj
Only in boost/tools/build/jam_src: jamgram.obj
Only in boost/tools/build/jam_src: lists.obj
Only in boost/tools/build/jam_src: make.obj
Only in boost/tools/build/jam_src: make1.obj
Only in boost/tools/build/jam_src: modules.obj
Only in boost/tools/build/jam_src: native.obj
Only in boost/tools/build/jam_src: newstr.obj
Only in boost/tools/build/jam_src: option.obj
Only in boost/tools/build/jam_src: order.obj
Only in boost/tools/build/jam_src: parse.obj
Only in boost/tools/build/jam_src: path.obj
Only in boost/tools/build/jam_src: pathunix.obj
Only in boost/tools/build/jam_src: pathvms.obj
Only in boost/tools/build/jam_src: property-set.obj
Only in boost/tools/build/jam_src: pwd.obj
Only in boost/tools/build/jam_src: regex.obj
Only in boost/tools/build/jam_src: regexp.obj
Only in boost/tools/build/jam_src: rules.obj
Only in boost/tools/build/jam_src: scan.obj
Only in boost/tools/build/jam_src: search.obj
Only in boost/tools/build/jam_src: sequence.obj
Only in boost/tools/build/jam_src: set.obj
Only in boost/tools/build/jam_src: strings.obj
Only in boost/tools/build/jam_src: subst.obj
Only in boost/tools/build/jam_src: timestamp.obj
Only in boost/tools/build/jam_src: variable.obj
Only in boost/tools/build/jam_src: vc70.pdb
Only in boost/tools/build/jam_src: w32_getreg.obj
-------------- next part --------------
# Copyright Niall Douglas 2005.
# Distributed under the Boost Software License, Version 1.0.
# (See accompanying file LICENSE_1_0.txt or copy at
# http://www.boost.org/LICENSE_1_0.txt)


"""
>>> from voidptr_ext import *


    Check for correct conversion
    
>>> use(get())

    Check that None is converted to a NULL void pointer
    
>>> useany(get())
1
>>> useany(None)
0

    Check that we don't lose type information by converting NULL
    opaque pointers to None
    
>>> assert getnull() is None
>>> useany(getnull())
0

   Check that there is no conversion from integers ...
   
>>> try: use(0)
... except TypeError: pass
... else: print 'expected a TypeError'

   ... and from strings to opaque objects
   
>>> try: use("")
... except TypeError: pass
... else: print 'expected a TypeError'
"""
def run(args = None):
    import sys
    import doctest

    if args is not None:
        sys.argv = args
    return doctest.testmod(sys.modules.get(__name__))
    
if __name__ == '__main__':
    print "running..."
    import sys
    status = run()[0]
    if (status == 0): print "Done."
    sys.exit(status)
-------------- next part --------------
The following section of this message contains a file attachment
prepared for transmission using the Internet MIME message format.
If you are using Pegasus Mail, or any other MIME-compliant system,
you should be able to save it or view it from within your mailer.
If you cannot, please ask your system administrator for assistance.

   ---- File information -----------
     File:  voidptr.cpp
     Date:  6 Nov 2005, 17:38
     Size:  892 bytes.
     Type:  Program-source
-------------- next part --------------
A non-text attachment was scrubbed...
Name: voidptr.cpp
Type: application/octet-stream
Size: 892 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051106/6ddaeb9e/attachment.obj>

From roman.yakovenko at gmail.com  Mon Nov  7 06:11:19 2005
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Mon, 7 Nov 2005 07:11:19 +0200
Subject: [C++-sig] Patch implementing void * support
In-Reply-To: <436E4390.14657.548C6F9@localhost>
References: <436E4390.14657.548C6F9@localhost>
Message-ID: <7465b6170511062111v3efd7dd4v49d5bb3ff5c72b75@mail.gmail.com>

On 11/6/05, Niall Douglas <s_sourceforge at nedprod.com> wrote:
> The enclosed patch adds void * support to Boost.Python by making void
> an opaque type. Code that uses void * "just works" with this patch
> with no further effort required.

Cool. I will try your patch today on GCC.

Roman Yakovenko


From meine at kogs1.informatik.uni-hamburg.de  Mon Nov  7 14:06:30 2005
From: meine at kogs1.informatik.uni-hamburg.de (Hans Meine)
Date: Mon, 7 Nov 2005 14:06:30 +0100
Subject: [C++-sig] boost.python error
In-Reply-To: <5e44108d0511060813w1e42b226jab48e624bf79442a@mail.gmail.com>
References: <5e44108d0511051844rdaafb4fxe4e6fd971056afb8@mail.gmail.com>
	<20051106155421.96407.qmail@web31509.mail.mud.yahoo.com>
	<5e44108d0511060813w1e42b226jab48e624bf79442a@mail.gmail.com>
Message-ID: <200511071406.30853.meine@kogs.informatik.uni-hamburg.de>

On Sunday 06 November 2005 17:13, Ricardo Kirkner wrote:
> ImportError: dynamic module does not define init function (initOpalMath)
You could check for that in the source (remember the mentioned macro) and in 
the .o file with "nm".

> To build using g++ I use
>
> g++ -c -Wall -fPIC -I/usr/include/boost -I/usr/include/python2.3 -Isrc
> -o OpalMath.o OpalMath.cpp
> g++ -shared -fPIC -Wl --no-undefined --warn-once -L/usr/lib
> -lpython2.3 -lboost_python -DBOOST_PYTHON_DYNAMIC_MODULE -o
> OpalMath.so OpalMath.o
Note that -DBOOST_PYTHON_DYNAMIC_MODULE defines a preprocessor symbol.  That's 
a noop when linking.  (Probably you wanted to add that within the "g++ 
-c ..." line.)

-- 
Ciao, /  /
     /--/
    /  / ANS


From ricardokirkner at gmail.com  Mon Nov  7 14:28:00 2005
From: ricardokirkner at gmail.com (Ricardo Kirkner)
Date: Mon, 7 Nov 2005 10:28:00 -0300
Subject: [C++-sig] boost.python error
In-Reply-To: <200511071406.30853.meine@kogs.informatik.uni-hamburg.de>
References: <5e44108d0511051844rdaafb4fxe4e6fd971056afb8@mail.gmail.com>
	<20051106155421.96407.qmail@web31509.mail.mud.yahoo.com>
	<5e44108d0511060813w1e42b226jab48e624bf79442a@mail.gmail.com>
	<200511071406.30853.meine@kogs.informatik.uni-hamburg.de>
Message-ID: <5e44108d0511070528w31feecb2nbd0e34d79e14a0ca@mail.gmail.com>

Thank you all for your help.. I finally managed to get the library to
compile. I don't quite know what it was, but I am sure it had
something to do with Jamfile/Jamrules/boost-build.jam files
configuration.

Regards,

Ricardo Kirkner

On 11/7/05, Hans Meine <meine at kogs1.informatik.uni-hamburg.de> wrote:
> On Sunday 06 November 2005 17:13, Ricardo Kirkner wrote:
> > ImportError: dynamic module does not define init function (initOpalMath)
> You could check for that in the source (remember the mentioned macro) and in
> the .o file with "nm".
>
> > To build using g++ I use
> >
> > g++ -c -Wall -fPIC -I/usr/include/boost -I/usr/include/python2.3 -Isrc
> > -o OpalMath.o OpalMath.cpp
> > g++ -shared -fPIC -Wl --no-undefined --warn-once -L/usr/lib
> > -lpython2.3 -lboost_python -DBOOST_PYTHON_DYNAMIC_MODULE -o
> > OpalMath.so OpalMath.o
> Note that -DBOOST_PYTHON_DYNAMIC_MODULE defines a preprocessor symbol.  That's
> a noop when linking.  (Probably you wanted to add that within the "g++
> -c ..." line.)
>
> --
> Ciao, /  /
>      /--/
>     /  / ANS
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>


From ricardokirkner at gmail.com  Mon Nov  7 14:30:43 2005
From: ricardokirkner at gmail.com (Ricardo Kirkner)
Date: Mon, 7 Nov 2005 10:30:43 -0300
Subject: [C++-sig] call policies
Message-ID: <5e44108d0511070530r34b4b022wc25cb52f6b3dac3a@mail.gmail.com>

Hello again:

this time I have a question regarding call policies. I have a function
that returns a double * which actually is an array of doubles.

I am required to provide a policy for this functions, however I cannot
find out which one would apply here (from how well I understand each
posibility).

Can anyone please enlighten me?

Thank you

Ricardo Kirkner


From bray at sent.com  Mon Nov  7 15:54:25 2005
From: bray at sent.com (Brian Ray)
Date: Mon, 7 Nov 2005 08:54:25 -0600
Subject: [C++-sig] Patch implementing void * support
In-Reply-To: <436E4390.14657.548C6F9@localhost>
References: <436E4390.14657.548C6F9@localhost>
Message-ID: <2CCF1296-D7BC-4124-83CB-A9B652919C7F@sent.com>

I wish someone can tell me in what cases this is useful. I am sure it  
*is* helpful but I am trying to imagine another more extensive use  
case. Or, is this something more useful to Python Boost itself and  
not users of Python Boost.


On Nov 6, 2005, at 11:55 AM, Niall Douglas wrote:

> The enclosed patch adds void * support to Boost.Python by making void
> an opaque type. Code that uses void * "just works" with this patch
> with no further effort required.
>
> Also enclosed is a testcase and the docs have been modified.
>
> Ran full testsuite on msvc7.1 and msvc6 - everything works. I don't
> have GCC to hand ATM, but apart from maybe a missing typename
> specifier I can't see anything which would fault.
>
> Cheers,
> Niall
>


From s_sourceforge at nedprod.com  Mon Nov  7 16:29:37 2005
From: s_sourceforge at nedprod.com (Niall Douglas)
Date: Mon, 07 Nov 2005 15:29:37 -0000
Subject: [C++-sig] Patch implementing void * support
In-Reply-To: <2CCF1296-D7BC-4124-83CB-A9B652919C7F@sent.com>
References: <436E4390.14657.548C6F9@localhost>
Message-ID: <436F72E1.8296.9E99B51@localhost>

On 7 Nov 2005 at 8:54, Brian Ray wrote:

> I wish someone can tell me in what cases this is useful. I am sure it  
> *is* helpful but I am trying to imagine another more extensive use  
> case. Or, is this something more useful to Python Boost itself and  
> not users of Python Boost.

A lot of GUI toolkit C++ code has message handlers of the form:

virtual int Widget::handle(MsgType msg, void *data)

... where MsgType specifies what *data is. For almost all cases, the 
python bindings will call the C++ implementation and just pass the 
void * unchanged. Where an override is made, you can write separate 
little accessor functions to restore type info to data so python can 
access it.

This idiom is especially popular in C code too.

Cheers,
Niall





From etaurel at cells.es  Mon Nov  7 18:11:19 2005
From: etaurel at cells.es (Emmanuel Taurel)
Date: Mon, 07 Nov 2005 18:11:19 +0100
Subject: [C++-sig] Problem using the ptr() function
In-Reply-To: <u1x25kcih.fsf@boost-consulting.com>
Message-ID: <0IPL008KCH3D2Q@cells-s02.cells.es>

Hello everybody,

As requested by Dave, I have reduced my example as far as I can.
Sorry for the very, very long response time....

I recall my problem: I have a C++ class called Device. In the Python world,
I create a class inheriting from this C++ class (called MyDevice) and I
create one python method in this new class which is called "a_new_method".
This python method receives one parameter of a type for which I have
installed Boost converter. The type is called "MyStrArray"
The Python code creates one instance of the MyDevice class and call its C++
method called "execute_device_method". This C++ method callback the new
"a_new_method" python code with the help of the Boost call_method()
function.
If, when calling back the new Python method, I am using the Boost ptr()
Function, I have a run-time error:

Traceback (most recent call last):
  File "main.py", line 13, in ?
    d.execute_device_method()
TypeError: No Python class registered for C++ class std::vector<std::string,
std::allocator<std::string> >

If I don't use the Boost ptr() function, everything works fine and the
"a_new_method" Python method is executed correctly.

I have added to this file two files which are the Python and the C++ code.

Thank's for your help

Sincerely

Emmanuel Taurel (etaurel at cells.es)


-------------- next part --------------
A non-text attachment was scrubbed...
Name: main.py
Type: application/octet-stream
Size: 299 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051107/830d8d9a/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: py_ds.cpp
Type: application/octet-stream
Size: 3036 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051107/830d8d9a/attachment-0001.obj>

From rwgk at yahoo.com  Mon Nov  7 20:11:16 2005
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Mon, 7 Nov 2005 11:11:16 -0800 (PST)
Subject: [C++-sig] call policies
In-Reply-To: <5e44108d0511070530r34b4b022wc25cb52f6b3dac3a@mail.gmail.com>
Message-ID: <20051107191116.45638.qmail@web31515.mail.mud.yahoo.com>

--- Ricardo Kirkner <ricardokirkner at gmail.com> wrote:

> this time I have a question regarding call policies. I have a function
> that returns a double * which actually is an array of doubles.
> 
> I am required to provide a policy for this functions, however I cannot
> find out which one would apply here (from how well I understand each
> posibility).
> 
> Can anyone please enlighten me?

I don't think call policies will help you here. More fundamental
considerations are required.

The Python scripting level is supposed to provide a safe environment.
According to the Python spirit is should be impossible to provoke a
segmentation fault from the scripting level, or to access uninitialized
memory. Of course you can undermine this via C/C++, but Boost.Python
doesn't support grossly unsafe practices.

In your particular case, you have to tell Python the size of the
double* array, and you have to take care all elements are initialized.
There are several alternatives to achieve this. A simple alternative is
to write a "thin wrapper" which copies the double* array to a new
Python list of floats. It goes like this (all untested):

your.hpp:

  class yours
  {
    public:
      unsigned
      number_of_values() const;

      double*
      values() const;
  };

Boost.Python extension:

  #include <boost/python.hpp>

  boost::python::list
  values_as_list(yours const& self)
  {
    boost::python::list result;
    for(unsigned i=0;i<self.number_of_values();i++) {
      result.append(self.values()[i]);
    }
    return result;
  }

  BOOST_PYTHON_MODULE(your_choice)
  {
    using namespace boost::python;
    class_<yours>("yours")
      def("values", values_as_list)
    ;
  }

Cheers,
        Ralf



		
__________________________________ 
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com


From ricardokirkner at gmail.com  Tue Nov  8 00:06:19 2005
From: ricardokirkner at gmail.com (Ricardo Kirkner)
Date: Mon, 7 Nov 2005 20:06:19 -0300
Subject: [C++-sig] call policies
In-Reply-To: <20051107191116.45638.qmail@web31515.mail.mud.yahoo.com>
References: <5e44108d0511070530r34b4b022wc25cb52f6b3dac3a@mail.gmail.com>
	<20051107191116.45638.qmail@web31515.mail.mud.yahoo.com>
Message-ID: <5e44108d0511071506r55a4cdeav2aad8bf13f8e9a8d@mail.gmail.com>

Hi:

Thank you, this worked... however, pyste is still complaining about my
method (the one that returned a double*) returns a pointer, and has no
policy defined -- although this method now uses a wrapper which does
not return a pointer but a tuple...

On the other hand, I have read that the operator << should be replaced
by the str function. However,

1. pyste isn't doing it.. (might be a thing with pyste)
2. when I replace that operator with .def( str( self ) ), compilation fails with

no matching function for call to
'boost::python::api::object::visit(boost::python::class_<myclass,
boost::python::detail::not_specified,
boost::python::detail::not_specified,
boost::python::detail::not_specified>&) const'

any ideas?

thank you



On 11/7/05, Ralf W. Grosse-Kunstleve <rwgk at yahoo.com> wrote:
> --- Ricardo Kirkner <ricardokirkner at gmail.com> wrote:
>
> > this time I have a question regarding call policies. I have a function
> > that returns a double * which actually is an array of doubles.
> >
> > I am required to provide a policy for this functions, however I cannot
> > find out which one would apply here (from how well I understand each
> > posibility).
> >
> > Can anyone please enlighten me?
>
> I don't think call policies will help you here. More fundamental
> considerations are required.
>
> The Python scripting level is supposed to provide a safe environment.
> According to the Python spirit is should be impossible to provoke a
> segmentation fault from the scripting level, or to access uninitialized
> memory. Of course you can undermine this via C/C++, but Boost.Python
> doesn't support grossly unsafe practices.
>
> In your particular case, you have to tell Python the size of the
> double* array, and you have to take care all elements are initialized.
> There are several alternatives to achieve this. A simple alternative is
> to write a "thin wrapper" which copies the double* array to a new
> Python list of floats. It goes like this (all untested):
>
> your.hpp:
>
>   class yours
>   {
>     public:
>       unsigned
>       number_of_values() const;
>
>       double*
>       values() const;
>   };
>
> Boost.Python extension:
>
>   #include <boost/python.hpp>
>
>   boost::python::list
>   values_as_list(yours const& self)
>   {
>     boost::python::list result;
>     for(unsigned i=0;i<self.number_of_values();i++) {
>       result.append(self.values()[i]);
>     }
>     return result;
>   }
>
>   BOOST_PYTHON_MODULE(your_choice)
>   {
>     using namespace boost::python;
>     class_<yours>("yours")
>       def("values", values_as_list)
>     ;
>   }
>
> Cheers,
>         Ralf
>
>
>
>
> __________________________________
> Yahoo! FareChase: Search multiple travel sites in one click.
> http://farechase.yahoo.com
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>


From rwgk at yahoo.com  Tue Nov  8 00:37:36 2005
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Mon, 7 Nov 2005 15:37:36 -0800 (PST)
Subject: [C++-sig] call policies
In-Reply-To: <5e44108d0511071506r55a4cdeav2aad8bf13f8e9a8d@mail.gmail.com>
Message-ID: <20051107233736.37268.qmail@web31503.mail.mud.yahoo.com>

--- Ricardo Kirkner <ricardokirkner at gmail.com> wrote:
> Thank you, this worked... however, pyste is still complaining about my
> method (the one that returned a double*) returns a pointer, and has no
> policy defined -- although this method now uses a wrapper which does
> not return a pointer but a tuple...
> 
> On the other hand, I have read that the operator << should be replaced
> by the str function. However,
> 
> 1. pyste isn't doing it.. (might be a thing with pyste)

Sorry, I never used Pyste. I hope someone else can help.

> 2. when I replace that operator with .def( str( self ) ), compilation fails
> with

IIRC some compilers have issues with the name lookup. You could try

  .def(self_ns::str(self))

I believe.



		
__________________________________ 
Start your day with Yahoo! - Make it your home page! 
http://www.yahoo.com/r/hs


From ricardokirkner at gmail.com  Tue Nov  8 00:42:21 2005
From: ricardokirkner at gmail.com (Ricardo Kirkner)
Date: Mon, 7 Nov 2005 20:42:21 -0300
Subject: [C++-sig] call policies
In-Reply-To: <20051107233736.37268.qmail@web31503.mail.mud.yahoo.com>
References: <5e44108d0511071506r55a4cdeav2aad8bf13f8e9a8d@mail.gmail.com>
	<20051107233736.37268.qmail@web31503.mail.mud.yahoo.com>
Message-ID: <5e44108d0511071542o55ab6776pab8cb479b99cb84a@mail.gmail.com>

Ok... thanks for the reply...

How will boost.python figure out that the corresponding method for

.def(self_ns::str(self))

should be (for example)

std::ostream& operator<< (in the c++ code)

?

I will try to skip pyste, and code directly in boost, as to have less
code levels to deal with...

ricardo

On 11/7/05, Ralf W. Grosse-Kunstleve <rwgk at yahoo.com> wrote:
> --- Ricardo Kirkner <ricardokirkner at gmail.com> wrote:
> > Thank you, this worked... however, pyste is still complaining about my
> > method (the one that returned a double*) returns a pointer, and has no
> > policy defined -- although this method now uses a wrapper which does
> > not return a pointer but a tuple...
> >
> > On the other hand, I have read that the operator << should be replaced
> > by the str function. However,
> >
> > 1. pyste isn't doing it.. (might be a thing with pyste)
>
> Sorry, I never used Pyste. I hope someone else can help.
>
> > 2. when I replace that operator with .def( str( self ) ), compilation fails
> > with
>
> IIRC some compilers have issues with the name lookup. You could try
>
>   .def(self_ns::str(self))
>
> I believe.
>
>
>
>
> __________________________________
> Start your day with Yahoo! - Make it your home page!
> http://www.yahoo.com/r/hs
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>


From rwgk at yahoo.com  Tue Nov  8 03:45:51 2005
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Mon, 7 Nov 2005 18:45:51 -0800 (PST)
Subject: [C++-sig] call policies
In-Reply-To: <5e44108d0511071542o55ab6776pab8cb479b99cb84a@mail.gmail.com>
Message-ID: <20051108024551.62158.qmail@web31509.mail.mud.yahoo.com>

--- Ricardo Kirkner <ricardokirkner at gmail.com> wrote:

> Ok... thanks for the reply...
> 
> How will boost.python figure out that the corresponding method for
> 
> .def(self_ns::str(self))
> 
> should be (for example)
> 
> std::ostream& operator<< (in the c++ code)

Only David knows *how* Boost.Python does it, but this test proves that it does
exactly what you are looking for:

boost/libs/python/test/operators.cpp

Note that you can always provide your own custom string method via

.def("__str__", &yours::method)

or

.def("__str__", another_thin_wrapper)

BTW: I find the test source code extremely useful as a reference. It is worth
spending 30 minutes looking through all the .cpp files to get an overview.

Cheers,
        Ralf



	
		
__________________________________ 
Yahoo! Mail - PC Magazine Editors' Choice 2005 
http://mail.yahoo.com


From dave at boost-consulting.com  Tue Nov  8 15:53:37 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Tue, 08 Nov 2005 09:53:37 -0500
Subject: [C++-sig] Problem using the ptr() function
References: <u1x25kcih.fsf@boost-consulting.com>
	<0IPL008KCH3D2Q@cells-s02.cells.es>
Message-ID: <uirv342dq.fsf@boost-consulting.com>

Emmanuel Taurel <etaurel at cells.es> writes:

> As requested by Dave, I have reduced my example as far as I can.

Your example can still be reduced a *lot* further. You could start by
taking out all the use of std::cout.  I doubt you need a loop in
a_new_method.  Next time, please remove *all* unnecessary code and get
it down to the bare minimum required to exhibit the behavior you don't
expect.

Anyway, this time the problem is pretty straightforward.  Basically,
ptr(p) will only work when the type of *p has been exported with
class_.  In that case, Python ends up with an object of some Python
type corresponding to the C++ type of *p, containing a copy of p.

You are converting MyStrArray to and from a Python list instead of
wrapping it as a class.  There's no way that modifications to a
regular python list will be able to automatically affect *p.

In the past it has been requested that we support "write-back"
converters that let you convert, e.g., a MyStrArray to a Python list,
and then, when the call completes, convert the list back to a
MyStrArray.  We don't support that yet, but since you're actually
doing the call from Python to C++, it's easy for you to fake it:

    python::list send_msa_(send_msa); // convert to a Python list
    
	call_method<void>(dev_ptr->m_self.get(),"a_new_method", send_msa_);

    send_msa = extract<MyStrArray>(send_msa_);

BTW, something about

  class DeviceWrap : public Device, public wrapper<Device>
  {
  public:
      DeviceWrap(PyObject *self):Device(),m_self(borrowed(self)) {};
      DeviceWrap(PyObject *self, const string &na):Device(na),m_self(borrowed(self)) {};

      handle<> m_self;
  };

looks very wrong.  If you're using wrapper<...> there's almost never
cause for an initial self argument, and storing it in a handle<> is
almost guaranteed to result in a leak or a reference counting error.
I would unconditionally drop handle<>, and either try to make
wrapper<Device> work by itself, using get_override to get ahold of
a_new_method, or drop wrapper<Device> store a raw PyObject* in lieu of
handle<>, specializing has_back_reference for Device.  In this case
the latter approach might be better, by allowing you to get rid of the
artificial derived class and the dangerous downcast.
    
Good luck,
Dave

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



From skottmckay at gmail.com  Tue Nov  8 17:09:25 2005
From: skottmckay at gmail.com (Scott McKay)
Date: Tue, 8 Nov 2005 09:09:25 -0700
Subject: [C++-sig] call policies
In-Reply-To: <5e44108d0511071506r55a4cdeav2aad8bf13f8e9a8d@mail.gmail.com>
References: <5e44108d0511070530r34b4b022wc25cb52f6b3dac3a@mail.gmail.com>
	<20051107191116.45638.qmail@web31515.mail.mud.yahoo.com>
	<5e44108d0511071506r55a4cdeav2aad8bf13f8e9a8d@mail.gmail.com>
Message-ID: <51d017870511080809wbf485aj3f502200ac959ad8@mail.gmail.com>

Pyste doesn't appear to cancel warnings if wrappers are defined for a
method. It should be simple to add though.

S

On 11/7/05, Ricardo Kirkner <ricardokirkner at gmail.com> wrote:
>
> Hi:
>
> Thank you, this worked... however, pyste is still complaining about my
> method (the one that returned a double*) returns a pointer, and has no
> policy defined -- although this method now uses a wrapper which does
> not return a pointer but a tuple...
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051108/959a98e5/attachment.htm>

From ricardokirkner at gmail.com  Tue Nov  8 21:44:38 2005
From: ricardokirkner at gmail.com (Ricardo Kirkner)
Date: Tue, 8 Nov 2005 17:44:38 -0300
Subject: [C++-sig] call policies
In-Reply-To: <20051108024551.62158.qmail@web31509.mail.mud.yahoo.com>
References: <5e44108d0511071542o55ab6776pab8cb479b99cb84a@mail.gmail.com>
	<20051108024551.62158.qmail@web31509.mail.mud.yahoo.com>
Message-ID: <5e44108d0511081244w1e151676n1607d67c0239f353@mail.gmail.com>

Hi:

I am doing great now... I am coding right in boost.python, without
recurring to pyste.

However I have found a situation that makes me doubt. Being mData defined as

real mData[3];

and having these methods as member functions of a class

		inline real * getData()
		{
			return mData;
		}

		inline const real * getData()const
		{
			return mData;
		}

How should I wrap them? I was thinking of defining a wrapper that
would return a list built from the elements of mData, but... how would
using a wrapper cope with this overloaded function? Should I only
define it once, and ignore the second override? (for example the one
with const?)

I have looked at the tutorial, but the part of overriding only handles
the case where the function itself doesn't return a pointer or a
reference...

More or less the same thing is happening with the [] operator... the
code (c++) would return a real & while I could return a real, and
using a wrapper, but this would prevent the returned item from being
modified (as opposed to the c++ version, where a reference can be used
to modify the object).

Any thoughts are appreciated...

Thanks again...

Ricardo

pd: I promise that as I continue to learn this wrapping_up_a_library
thing I will stop making basic questions and start making really
interesting ones :-P

On 11/7/05, Ralf W. Grosse-Kunstleve <rwgk at yahoo.com> wrote:
> --- Ricardo Kirkner <ricardokirkner at gmail.com> wrote:
>
> > Ok... thanks for the reply...
> >
> > How will boost.python figure out that the corresponding method for
> >
> > .def(self_ns::str(self))
> >
> > should be (for example)
> >
> > std::ostream& operator<< (in the c++ code)
>
> Only David knows *how* Boost.Python does it, but this test proves that it does
> exactly what you are looking for:
>
> boost/libs/python/test/operators.cpp
>
> Note that you can always provide your own custom string method via
>
> .def("__str__", &yours::method)
>
> or
>
> .def("__str__", another_thin_wrapper)
>
> BTW: I find the test source code extremely useful as a reference. It is worth
> spending 30 minutes looking through all the .cpp files to get an overview.
>
> Cheers,
>         Ralf
>
>
>
>
>
> __________________________________
> Yahoo! Mail - PC Magazine Editors' Choice 2005
> http://mail.yahoo.com
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>


From rwgk at yahoo.com  Tue Nov  8 23:15:16 2005
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Tue, 8 Nov 2005 14:15:16 -0800 (PST)
Subject: [C++-sig] call policies
In-Reply-To: <5e44108d0511081244w1e151676n1607d67c0239f353@mail.gmail.com>
Message-ID: <20051108221516.38309.qmail@web31507.mail.mud.yahoo.com>

--- Ricardo Kirkner <ricardokirkner at gmail.com> wrote:
> real mData[3];
> 
> and having these methods as member functions of a class
> 
> 		inline real * getData()
> 		{
> 			return mData;
> 		}
> 
> 		inline const real * getData()const
> 		{
> 			return mData;
> 		}
> 
> How should I wrap them? I was thinking of defining a wrapper that
> would return a list built from the elements of mData, but... how would
> using a wrapper cope with this overloaded function? Should I only
> define it once, and ignore the second override? (for example the one
> with const?)

You can keep your C++ interface if you want. C++ const-ness doesn't map
directly to Python, but it isn't in the way.

Your idea with the wrapper returning a new list seems reasonable:

  boost::python::list
  getData_as_list(T const& self)
  {
    //...
  }

then 

  .def("getData", getData_as_list)

This will use the "const" overload.

The Python interface cannot map directly to the C++ interface when it comes
to resetting data. I'd try this:

  void
  setData_from_list(T& self, boost::python::list const& new_data)
  {
     double* d = self.getData(); // will use the non-const overload
     //... loop over new_data, d[i] =
boost::python::extract<double>(new_data[i])();
  }

I am not sure about the exact details; I hope you can work it out.
Finally:

  def("setData", setData_from_list)

BTW: your problems remind me of my early days with Boost.Python. I moved away
from using raw pointers entirely. Instead I wrote a library with a family of
useful array types, including types with and without dynamic memory allocation.
A general purpose "container_conversions.h" header provides mappings from C++
arrays to Python tuples and vice versa. The mappings for a C++ type can be
defined with just one line. See the Boost.Python FAQ for some pointers. If you
want to use Boost.Python at a larger scale you may want to know about this
approach. Sadly I never found the time to document my libraries, but I think
the code you'll find is still useful as a source of ideas, and I'll answer
questions here.

Cheers,
        Ralf



	
		
__________________________________ 
Yahoo! Mail - PC Magazine Editors' Choice 2005 
http://mail.yahoo.com


From ricardokirkner at gmail.com  Wed Nov  9 13:49:18 2005
From: ricardokirkner at gmail.com (Ricardo Kirkner)
Date: Wed, 9 Nov 2005 09:49:18 -0300
Subject: [C++-sig] call policies
In-Reply-To: <20051108221516.38309.qmail@web31507.mail.mud.yahoo.com>
References: <5e44108d0511081244w1e151676n1607d67c0239f353@mail.gmail.com>
	<20051108221516.38309.qmail@web31507.mail.mud.yahoo.com>
Message-ID: <5e44108d0511090449t4bcdbc24q35db09ece1d979e5@mail.gmail.com>

Thank you for the info... I will look further into it to see if I can
use it to solve my problems..

Unfortunately I cannot avoid using pointers, since I am trying to
write a wrapper for a library (which wasn't written by me) which uses
them....

I will have to manage to get around them somehow...

Thanks for the suggestions, though...

Regards,

Ricardo Kirkner

On 11/8/05, Ralf W. Grosse-Kunstleve <rwgk at yahoo.com> wrote:
> --- Ricardo Kirkner <ricardokirkner at gmail.com> wrote:
> > real mData[3];
> >
> > and having these methods as member functions of a class
> >
> >               inline real * getData()
> >               {
> >                       return mData;
> >               }
> >
> >               inline const real * getData()const
> >               {
> >                       return mData;
> >               }
> >
> > How should I wrap them? I was thinking of defining a wrapper that
> > would return a list built from the elements of mData, but... how would
> > using a wrapper cope with this overloaded function? Should I only
> > define it once, and ignore the second override? (for example the one
> > with const?)
>
> You can keep your C++ interface if you want. C++ const-ness doesn't map
> directly to Python, but it isn't in the way.
>
> Your idea with the wrapper returning a new list seems reasonable:
>
>   boost::python::list
>   getData_as_list(T const& self)
>   {
>     //...
>   }
>
> then
>
>   .def("getData", getData_as_list)
>
> This will use the "const" overload.
>
> The Python interface cannot map directly to the C++ interface when it comes
> to resetting data. I'd try this:
>
>   void
>   setData_from_list(T& self, boost::python::list const& new_data)
>   {
>      double* d = self.getData(); // will use the non-const overload
>      //... loop over new_data, d[i] =
> boost::python::extract<double>(new_data[i])();
>   }
>
> I am not sure about the exact details; I hope you can work it out.
> Finally:
>
>   def("setData", setData_from_list)
>
> BTW: your problems remind me of my early days with Boost.Python. I moved away
> from using raw pointers entirely. Instead I wrote a library with a family of
> useful array types, including types with and without dynamic memory allocation.
> A general purpose "container_conversions.h" header provides mappings from C++
> arrays to Python tuples and vice versa. The mappings for a C++ type can be
> defined with just one line. See the Boost.Python FAQ for some pointers. If you
> want to use Boost.Python at a larger scale you may want to know about this
> approach. Sadly I never found the time to document my libraries, but I think
> the code you'll find is still useful as a source of ideas, and I'll answer
> questions here.
>
> Cheers,
>         Ralf
>
>
>
>
>
> __________________________________
> Yahoo! Mail - PC Magazine Editors' Choice 2005
> http://mail.yahoo.com
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>


From rwgk at yahoo.com  Wed Nov  9 18:47:27 2005
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Wed, 9 Nov 2005 09:47:27 -0800 (PST)
Subject: [C++-sig] call policies
In-Reply-To: <5e44108d0511090449t4bcdbc24q35db09ece1d979e5@mail.gmail.com>
Message-ID: <20051109174727.43001.qmail@web31501.mail.mud.yahoo.com>

--- Ricardo Kirkner <ricardokirkner at gmail.com> wrote:

> Thank you for the info... I will look further into it to see if I can
> use it to solve my problems..
> 
> Unfortunately I cannot avoid using pointers, since I am trying to
> write a wrapper for a library (which wasn't written by me) which uses
> them....
> 
> I will have to manage to get around them somehow...

In case the library you are wrapping is large:

A few months ago I did some quick experiments (http://cci.lbl.gov/gltbx/)
wrapping the OpenGL library, which uses raw pointers extensively. I wrote a
simple Python script that generates the Boost.Python bindings. This is
supported by some hand-written wrapper support code. I found this combination
to be a very workable approach.

Cheers,
        Ralf



		
__________________________________ 
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com


From amohr at pixar.com  Thu Nov 10 03:12:21 2005
From: amohr at pixar.com (Alex Mohr)
Date: Wed, 09 Nov 2005 18:12:21 -0800
Subject: [C++-sig] enum strangeness...
Message-ID: <4372AC85.8070107@pixar.com>

I noticed something today about enums that I've wrapped with 
boost.python.  Here's a small example.

#include <boost/python.hpp>
using namespace boost::python;

struct Foo {
	enum Bar { Baz };
};

BOOST_PYTHON_MODULE(Mod) {
	scope fooScope = class_<Foo>("Foo");
	enum_<Foo::Bar>("Bar")
		.value("Baz", Foo::Baz)
		.export_values();
}

Now, in Python I can refer to the enum values by saying 'Mod.Foo.Baz' 
which is what I expect, but it seems to recur in a strange way.  I can 
also say:

Mod.Foo.Baz.Baz.Baz.Baz.Baz.Baz.Baz.Baz

for as long as I like.  This seems relatively benign, but strange.  Is 
it expected?

Thanks,

Alex


From korg at csse.unimelb.edu.au  Thu Nov 10 06:14:11 2005
From: korg at csse.unimelb.edu.au (Cameron Blackwood)
Date: Thu, 10 Nov 2005 16:14:11 +1100
Subject: [C++-sig] solaris-intel sem deadlock problem... :-/
Message-ID: <20051110051412.61AF96D617@kali.csse.unimelb.edu.au>



Any ideas welcome.... im out of ideas... :-/



Summary:
--------
Semaphore code hangs when call from python module but not from c/c++.

Kill me....


Background:
-----------
Im working on a project that is a massive c++ piece of work that
offers a single object with useful methods.

Its a database and it starts 3 other processes and uses shared memory
and semaphores to communicate.

The c++ interface is the native one (and its kinda big at ~30M exe's).

It has a C interface which creates a dll of c++ code and wraps
each method call in a c function.

There is a python wrapper that wraps the c interface.


Problem:
--------
It works fine on linux-x86, linux-ppc64 and we're just porting to
solaris-intel but we've hit a problem.

The startup code for the c++ object does some forking and sem work and
it works fine for the c++ interface and the c interface but the python
module just hangs in the semaphore semget loop.

The constructor for the c++ object is called with the same args via
python or c (so its not that). 

We're using boost for config passing, but the boost config values are
the same.

The c interface works (so its not the dll).

It could be some solaris issue, but the c and c++ interfaces work
fine... its just the python interface.

:-(

Has anyone got _any_ ideas? Or am I in for a big chunk of c++
semaphore debugging? :-/ (refer to kill me comment ;)



cheers,
cam

--
 / Cameron Blackwood                       korg  at  csse.unimelb.edu.au \
[ Research Programmer, Aditi Project                     +61 3  8344-1315 ]
 \ University of Melbourne                                               /


From ndbecker2 at gmail.com  Thu Nov 10 14:14:17 2005
From: ndbecker2 at gmail.com (Neal Becker)
Date: Thu, 10 Nov 2005 08:14:17 -0500
Subject: [C++-sig] swig?
Message-ID: <dkvh36$s8h$1@sea.gmane.org>

I noticed that recent versions of swig have quite significantly advanced the
support for c++.  I wonder if anyone has experience they can share,
regarding a comparison of boost::python with a recent swig.

I have been using boost::python for some time, but am interested in
considering swig, because in my present environment, it would be easier for
me to deploy code using swig.



From dave at boost-consulting.com  Thu Nov 10 17:08:49 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Thu, 10 Nov 2005 11:08:49 -0500
Subject: [C++-sig] enum strangeness...
References: <4372AC85.8070107@pixar.com>
Message-ID: <u7jbg4h9q.fsf@boost-consulting.com>

Alex Mohr <amohr at pixar.com> writes:

> I noticed something today about enums that I've wrapped with 
> boost.python.  Here's a small example.
>
> #include <boost/python.hpp>
> using namespace boost::python;
>
> struct Foo {
> 	enum Bar { Baz };
> };
>
> BOOST_PYTHON_MODULE(Mod) {
> 	scope fooScope = class_<Foo>("Foo");
> 	enum_<Foo::Bar>("Bar")
> 		.value("Baz", Foo::Baz)
> 		.export_values();
> }
>
> Now, in Python I can refer to the enum values by saying 'Mod.Foo.Baz' 
> which is what I expect, but it seems to recur in a strange way.  I can 
> also say:
>
> Mod.Foo.Baz.Baz.Baz.Baz.Baz.Baz.Baz.Baz
>
> for as long as I like.  This seems relatively benign, but strange.  Is 
> it expected?

Yeah, an enum's values are members of its type, and you can access
members of a type through its instances.  No surprise.

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



From petrucio at hoplon.com  Thu Nov 10 18:48:23 2005
From: petrucio at hoplon.com (Petrucio)
Date: Thu, 10 Nov 2005 14:48:23 -0300
Subject: [C++-sig] swig?
In-Reply-To: <dkvh36$s8h$1@sea.gmane.org>
References: <dkvh36$s8h$1@sea.gmane.org>
Message-ID: <6.0.2.0.1.20051110143038.01b44cf8@mail.hoplon.com>

In my experience, the compilation of the SWIG wrappers is much, much faster 
than compilation of the boost.python wrappers. The source of errors in 
compilation is also easier to track down, since there's little or no 
templates used in SWIG wrappers.

SWIG is also easier to set up and use, you usually need much less 
declarations of what you want to export, even less than when using Pyste.

It also has a lot of pre-built typemaps that save you a lot of time 
depending on what you want to export, for instance when dealing with 
pointer to basic types, SWIG has an easy solution for that in a typemap, in 
contrast to exporting those using boost.python, wich is not something you 
look forward to.

In the other hand, boost.python is more robust. I lack the memory to give 
you specific examples, but there's some stupid things that SWIG does that 
will blow your mind. Boost.python is also more 'purist', in some sense. But 
since practicality beats purity, I'm more of a SWIG guy.

I made a few performance tests, and expected boost.python to be a clear 
favorite here, but there were not much diference between the two, if I 
remember correctly.

I made a lot of other comparisons between the two, but the results are not 
in this computer. I'll send them some other day, if I remember to get them.

---
Fabio 'Petrucio' Stange - Game Programmer
Hoplon Infotainment - www.taikodom.com


At 10:14 10/11/2005, you wrote:
>I noticed that recent versions of swig have quite significantly advanced the
>support for c++.  I wonder if anyone has experience they can share,
>regarding a comparison of boost::python with a recent swig.
>
>I have been using boost::python for some time, but am interested in
>considering swig, because in my present environment, it would be easier for
>me to deploy code using swig.
>
>_______________________________________________
>C++-sig mailing list
>C++-sig at python.org
>http://mail.python.org/mailman/listinfo/c++-sig



From tomas at liquid.se  Thu Nov 10 18:05:39 2005
From: tomas at liquid.se (Tomas)
Date: Thu, 10 Nov 2005 18:05:39 +0100
Subject: [C++-sig] swig?
References: <dkvh36$s8h$1@sea.gmane.org>
	<6.0.2.0.1.20051110143038.01b44cf8@mail.hoplon.com>
Message-ID: <001301c5e619$00d7f110$0200a8c0@liquid.se>

Is it in SWIG possible to have smart pointers that holds objects of exported
classes?

/Tomas

----- Original Message ----- 
From: "Petrucio" <petrucio at hoplon.com>
To: "Development of Python/C++ integration" <c++-sig at python.org>
Sent: Thursday, November 10, 2005 6:48 PM
Subject: Re: [C++-sig] swig?


> In my experience, the compilation of the SWIG wrappers is much, much
faster
> than compilation of the boost.python wrappers. The source of errors in
> compilation is also easier to track down, since there's little or no
> templates used in SWIG wrappers.
>
> SWIG is also easier to set up and use, you usually need much less
> declarations of what you want to export, even less than when using Pyste.
>
> It also has a lot of pre-built typemaps that save you a lot of time
> depending on what you want to export, for instance when dealing with
> pointer to basic types, SWIG has an easy solution for that in a typemap,
in
> contrast to exporting those using boost.python, wich is not something you
> look forward to.
>
> In the other hand, boost.python is more robust. I lack the memory to give
> you specific examples, but there's some stupid things that SWIG does that
> will blow your mind. Boost.python is also more 'purist', in some sense.
But
> since practicality beats purity, I'm more of a SWIG guy.
>
> I made a few performance tests, and expected boost.python to be a clear
> favorite here, but there were not much diference between the two, if I
> remember correctly.
>
> I made a lot of other comparisons between the two, but the results are not
> in this computer. I'll send them some other day, if I remember to get
them.
>
> ---
> Fabio 'Petrucio' Stange - Game Programmer
> Hoplon Infotainment - www.taikodom.com
>
>
> At 10:14 10/11/2005, you wrote:
> >I noticed that recent versions of swig have quite significantly advanced
the
> >support for c++.  I wonder if anyone has experience they can share,
> >regarding a comparison of boost::python with a recent swig.
> >
> >I have been using boost::python for some time, but am interested in
> >considering swig, because in my present environment, it would be easier
for
> >me to deploy code using swig.
> >
> >_______________________________________________
> >C++-sig mailing list
> >C++-sig at python.org
> >http://mail.python.org/mailman/listinfo/c++-sig
>
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig



From petrucio at hoplon.com  Thu Nov 10 20:20:44 2005
From: petrucio at hoplon.com (Petrucio)
Date: Thu, 10 Nov 2005 16:20:44 -0300
Subject: [C++-sig] swig?
In-Reply-To: <001301c5e619$00d7f110$0200a8c0@liquid.se>
References: <dkvh36$s8h$1@sea.gmane.org>
	<6.0.2.0.1.20051110143038.01b44cf8@mail.hoplon.com>
	<001301c5e619$00d7f110$0200a8c0@liquid.se>
Message-ID: <6.0.2.0.1.20051110161854.01b5b560@mail.hoplon.com>

Not sure, since I don't usually use smart pointers. See
http://www.swig.org/Doc1.3/Python.html#Python_nn27

At 14:05 10/11/2005, you wrote:
>Is it in SWIG possible to have smart pointers that holds objects of exported
>classes?
>
>/Tomas
>
>----- Original Message -----
>From: "Petrucio" <petrucio at hoplon.com>
>To: "Development of Python/C++ integration" <c++-sig at python.org>
>Sent: Thursday, November 10, 2005 6:48 PM
>Subject: Re: [C++-sig] swig?
>
>
> > In my experience, the compilation of the SWIG wrappers is much, much
>faster
> > than compilation of the boost.python wrappers. The source of errors in
> > compilation is also easier to track down, since there's little or no
> > templates used in SWIG wrappers.
> >
> > SWIG is also easier to set up and use, you usually need much less
> > declarations of what you want to export, even less than when using Pyste.
> >
> > It also has a lot of pre-built typemaps that save you a lot of time
> > depending on what you want to export, for instance when dealing with
> > pointer to basic types, SWIG has an easy solution for that in a typemap,
>in
> > contrast to exporting those using boost.python, wich is not something you
> > look forward to.
> >
> > In the other hand, boost.python is more robust. I lack the memory to give
> > you specific examples, but there's some stupid things that SWIG does that
> > will blow your mind. Boost.python is also more 'purist', in some sense.
>But
> > since practicality beats purity, I'm more of a SWIG guy.
> >
> > I made a few performance tests, and expected boost.python to be a clear
> > favorite here, but there were not much diference between the two, if I
> > remember correctly.
> >
> > I made a lot of other comparisons between the two, but the results are not
> > in this computer. I'll send them some other day, if I remember to get
>them.
> >
> > ---
> > Fabio 'Petrucio' Stange - Game Programmer
> > Hoplon Infotainment - www.taikodom.com
> >
> >
> > At 10:14 10/11/2005, you wrote:
> > >I noticed that recent versions of swig have quite significantly advanced
>the
> > >support for c++.  I wonder if anyone has experience they can share,
> > >regarding a comparison of boost::python with a recent swig.
> > >
> > >I have been using boost::python for some time, but am interested in
> > >considering swig, because in my present environment, it would be easier
>for
> > >me to deploy code using swig.
> > >
> > >_______________________________________________
> > >C++-sig mailing list
> > >C++-sig at python.org
> > >http://mail.python.org/mailman/listinfo/c++-sig
> >
> > _______________________________________________
> > C++-sig mailing list
> > C++-sig at python.org
> > http://mail.python.org/mailman/listinfo/c++-sig
>
>_______________________________________________
>C++-sig mailing list
>C++-sig at python.org
>http://mail.python.org/mailman/listinfo/c++-sig



From grant at grantgoodyear.org  Tue Nov 15 00:10:25 2005
From: grant at grantgoodyear.org (Grant Goodyear)
Date: Mon, 14 Nov 2005 17:10:25 -0600
Subject: [C++-sig] pyste generation of getitem/setitem .def's
Message-ID: <20051114231025.GD7228@bmb24.uth.tmc.edu>

I recently hacked up my boost.python interface file to add __getitem__
and __setitem__ methods:

    class_< EMAN::EMData >("EMData", init<  >())
        .def(init< const EMAN::EMData& >())
        .def("__getitem__", &emdata_getitem)
        .def("__setitem__", &emdata_setitem)

I was generating the interface file using pyste, 

  Include("emdata_wrapitems.h") # *item wrappers declared here
  EMData = Class("EMAN::EMData", "emdata.h")

but I'm not sure how to get pyste to add the getitem/setitem .def's for
me.  I'm sure there's a simple trick that's more elegant than sedding
the file, but I'm not sure what it is.  There's "declaration_code" and
"module_code"; is there a "class_code" or something similar?

Thanks,
Grant
-- 
Grant Goodyear		
web: http://www.grantgoodyear.org	
e-mail: grant at grantgoodyear.org	


From jusylves at videotron.ca  Tue Nov 15 01:59:03 2005
From: jusylves at videotron.ca (Julien Sylvestre)
Date: Mon, 14 Nov 2005 19:59:03 -0500
Subject: [C++-sig] boost.python: modifying class getattr
Message-ID: <437932D7.7000402@videotron.ca>

Hello all,
    I'm trying to access the type structure for a c++ class exposed to
Python using boost.python, so that I can have ob_type->tp_getattro point
to a custom function instead of PyObject_GenericGetAttr (for those who
wonder why, it's a performance issue: with my current project,
PyObject_GenericGetAttr is always called first, fails to find the
requested attribute, and then __getattr__ is called on the wrapped
object, which returns the attribute. I get an extra PyString_FromFormatV
for each access to an attribute value, and this slows down my code by ~20%)

If I write:
object Foo_ = class_<Foo>("Foo", init< >())
      .def("__getattr__", &Foo::Foo_getattr)
      .def("__setattr__", &Foo::Foo_setattr)
...,

is there a way to modify the type object associated with Foo_ instances
so that instead of calling PyObject_GenericGetAttr when an attribute is
requested, Python calls a function
PyObject *
Foo_GetAttr(PyObject *obj, PyObject *name)
?

Thanks,
Julien.




From jan at langernetz.de  Tue Nov 15 12:22:33 2005
From: jan at langernetz.de (Jan Langer)
Date: Tue, 15 Nov 2005 12:22:33 +0100
Subject: [C++-sig] [Boost.Python] Solaris / gcc2.95.2 wchar.h problems
Message-ID: <dlcgdp$us7$1@sea.gmane.org>

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

Hello,
I read that this newsgroup is suitable for Boost.Python problems.

I wanted to compile Boost 1.33.0 and I ran into problems with
Boost.Python. My system is Solaris 5.7 with gcc 2.95.2. The Python
version is 2.4.1. When compling the file inheritance.cpp it includes
integer_traits.hpp which exits with an error in line 116 because
WCHAR_MIN and WCHAR_MAX are not defined. However, my wchar.h include
file defines those two macros. I couldn't find what was going wrong, but
moving the include of integer_traits.hpp in inheritance.cpp to the top
of the list of include files solves the problem. But this results in the
warning below.

- -----
gcc-C++-action
../../../bin/boost/libs/python/build/libboost_python.a/gcc/debug/inheritance.o
In file included from
/data/tmp/laja/boost_1_33_0/boost/python/detail/wrap_python.hpp:30,
                 from
/data/tmp/laja/boost_1_33_0/boost/python/detail/prefix.hpp:13,
                 from
/data/tmp/laja/boost_1_33_0/boost/python/type_id.hpp:8,
                 from
/data/tmp/laja/boost_1_33_0/boost/python/object/inheritance.hpp:8,
                 from
/data/tmp/laja/boost_1_33_0/libs/python/build/../src/object/inheritance.cpp:6:
/home/4all/packages/python/2.4.1/include/python2.4/pyconfig.h:820:
warning: `_FILE_OFFSET_BITS' redefined
/usr/include/sys/feature_tests.h:96: warning: this is the location of
the previous definition
- ------

If someone is interested in this problem, I could provide more details.
Jan
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (SunOS)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFDecT4yJOS3PG4wcwRAnRIAJ9LFKa6LWrmlDVrWL7GOu0po9bC9QCgoGG0
fWim8v7NG4Eg7uhoIVcyct4=
=OC/Y
-----END PGP SIGNATURE-----



From grant at grantgoodyear.org  Tue Nov 15 17:20:50 2005
From: grant at grantgoodyear.org (Grant Goodyear)
Date: Tue, 15 Nov 2005 10:20:50 -0600
Subject: [C++-sig] pyste generation of getitem/setitem .def's
In-Reply-To: <20051114231025.GD7228@bmb24.uth.tmc.edu>
References: <20051114231025.GD7228@bmb24.uth.tmc.edu>
Message-ID: <20051115162050.GE7228@bmb24.uth.tmc.edu>

Okay, more searching of the archives brought me an answer.  I'll attach
a complete, short, and silly example for the next person who runs into
this problem.  My thanks to whomever provided the "Vector2d" example
from which I liberally stole.

-g2boojum-
-- 
Grant Goodyear		
web: http://www.grantgoodyear.org	
e-mail: grant at grantgoodyear.org	
-------------- next part --------------
class Vector2d {
    public:
        Vector2d():x(0.0),y(0.0) {}
        Vector2d(const double xval, const double yval): x(xval),y(yval) {}
        ~Vector2d() {}
        double& operator[](const int i);
    protected:
        double x, y;
};

-------------- next part --------------
#include "foo.h"

double& Vector2d::operator[](const int i) { return i==0? x : y; }

-------------- next part --------------
#include "foo.h"
#include <boost/python.hpp>

#ifndef FOO_WRAP_H_
#define FOO_WRAP_H_

using boost::python::object;

object vec2dget(object self, object key);
void vec2dset(object self, object key, object val);

#endif // FOO_WRAP_H_
-------------- next part --------------
#include "foo_wrap.h"
#include "foo.h"

#ifndef FOO_H_
#define FOO_H_

using namespace boost::python;

object vec2dget(object self, object key) {
    int i = extract<int>(key);
    Vector2d& s = extract<Vector2d&>(self);
    return object(s[i]);
}

void vec2dset(object self, object key, object val) {
    Vector2d& s = extract<Vector2d&>(self);
    int i = extract<int>(key);
    double v = extract<double>(val);
    s[i] = v;
}

#endif // FOO_H
-------------- next part --------------
Include("foo_wrap.h")
Vector2d = Class("Vector2d", "foo_wrap.h")
add_method(Vector2d, "vec2dget")
rename(Vector2d.vec2dget, "__getitem__")
add_method(Vector2d, "vec2dset")
rename(Vector2d.vec2dset, "__setitem__")
-------------- next part --------------

// Boost Includes ==============================================================
#include <boost/python.hpp>
#include <boost/cstdint.hpp>

// Includes ====================================================================
#include <foo_wrap.h>

// Using =======================================================================
using namespace boost::python;

// Module ======================================================================
BOOST_PYTHON_MODULE(libpyfoo)
{
    class_< Vector2d >("Vector2d", init<  >())
        .def(init< const Vector2d& >())
        .def(init< const double, const double >())
        .def("__getitem__", &vec2dget)
        .def("__setitem__", &vec2dset)
    ;

}


From hans_meine at gmx.net  Tue Nov 15 18:33:28 2005
From: hans_meine at gmx.net (Hans Meine)
Date: Tue, 15 Nov 2005 18:33:28 +0100
Subject: [C++-sig] boost.python: modifying class getattr
In-Reply-To: <437932D7.7000402@videotron.ca>
References: <437932D7.7000402@videotron.ca>
Message-ID: <200511151833.29352.hans_meine@gmx.net>

On Tuesday 15 November 2005 01:59, Julien Sylvestre wrote:
> If I write:
> object Foo_ = class_<Foo>("Foo", init< >())
>       .def("__getattr__", &Foo::Foo_getattr)
>       .def("__setattr__", &Foo::Foo_setattr)
> ...,
>
> is there a way to modify the type object associated with Foo_ instances
> so that instead of calling PyObject_GenericGetAttr when an attribute is
> requested, Python calls a function
> PyObject *Foo_GetAttr(PyObject *obj, PyObject *name)

Ah, that would be cool; <dreaming>I would be looking for an analogous way to 
speed up __getitem__ making it comparable to that of a Python 
list..</dreaming>

Ciao, /  /
     /--/
    /  / ANS


From etaurel at cells.es  Wed Nov 16 16:34:37 2005
From: etaurel at cells.es (Emmanuel Taurel)
Date: Wed, 16 Nov 2005 16:34:37 +0100
Subject: [C++-sig] Specific exception classes
Message-ID: <0IQ2000QD0MDLE@cells-s02.cells.es>

Hello everybody.

 

I am using BPL to wrap a C++ library and I would like to define my own
exception. Therefore, I have written one exception translator that I have
registered with the "register_exception_translator". My exception translator
looks like this (DevFailed is the name of the C++ exception class) 

 

void translator_dev_failed(const DevFailed &the_ex)

{

            PyObject *py_exc =
PyErr_NewException("PyTangoDs.DevFailed",NULL,NULL);

            PyErr_SetString(py_exc,the_ex.errors[0].desc.in());

}

 

In Python, I have:

 

from PyTangoDs import *

try:

    ...

 

except DevFailed,e:

    print 'Received a DevFailed exception:',e

except:

    print 'An unforeseen exception occured....'

 

The python interpreter complains that it does not know the DevFailed name
with this error :

 

Traceback (most recent call last):

  File "main.py", line 19, in ?

    except DevFailed,e:

NameError: name 'DevFailed' is not defined

 

My question is : How is it possible to have your own exception ?

 

Thank's for your answer

 

Sincerely

 

Emmanuel Taurel (etaurel at cells.es)

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051116/c57aaa2f/attachment.htm>

From dave at boost-consulting.com  Wed Nov 16 18:37:30 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 16 Nov 2005 12:37:30 -0500
Subject: [C++-sig] boost.python: modifying class getattr
References: <437932D7.7000402@videotron.ca>
Message-ID: <uwtj8fq91.fsf@boost-consulting.com>

Julien Sylvestre <jusylves at videotron.ca> writes:

> Hello all,
>     I'm trying to access the type structure for a c++ class exposed to
> Python using boost.python, so that I can have ob_type->tp_getattro point
> to a custom function instead of PyObject_GenericGetAttr (for those who
> wonder why, it's a performance issue: with my current project,
> PyObject_GenericGetAttr is always called first, fails to find the
> requested attribute, and then __getattr__ is called on the wrapped
> object, which returns the attribute. I get an extra PyString_FromFormatV
> for each access to an attribute value, and this slows down my code by ~20%)
>
> If I write:
> object Foo_ = class_<Foo>("Foo", init< >())
>       .def("__getattr__", &Foo::Foo_getattr)
>       .def("__setattr__", &Foo::Foo_setattr)
> ...,
>
> is there a way to modify the type object associated with Foo_ instances
> so that instead of calling PyObject_GenericGetAttr when an attribute is
> requested, Python calls a function
> PyObject *
> Foo_GetAttr(PyObject *obj, PyObject *name)
> ?

  PyTypeObject* Foo_t = (PyTypeObject*)Foo_.ptr();
  Foo_t->tp_getattro = Foo_GetAttr;

HTH,

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



From dave at boost-consulting.com  Wed Nov 16 18:42:35 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 16 Nov 2005 12:42:35 -0500
Subject: [C++-sig] [Boost.Python] Solaris / gcc2.95.2 wchar.h problems
References: <dlcgdp$us7$1@sea.gmane.org>
Message-ID: <ur79gfq0k.fsf@boost-consulting.com>

Jan Langer <jan at langernetz.de> writes:

> Hello,
> I read that this newsgroup is suitable for Boost.Python problems.
>
> I wanted to compile Boost 1.33.0 and I ran into problems with
> Boost.Python. My system is Solaris 5.7 with gcc 2.95.2. The Python
> version is 2.4.1. When compling the file inheritance.cpp it includes
> integer_traits.hpp which exits with an error in line 116 because
> WCHAR_MIN and WCHAR_MAX are not defined. 

You should take this up with the maintainer of the relevant part of
integer_traits.hpp.  cvs annotate shows it to be John Maddock.

1.20         (johnmadd 21-Oct-02): // These are an implementation detail and not part of the interface
1.20         (johnmadd 21-Oct-02): #include <limits.h>
1.26         (johnmadd 17-Dec-04): // we need wchar.h for WCHAR_MAX/MIN but not all platforms provide it, 
1.26         (johnmadd 17-Dec-04): // and some may have <wchar.h> but not <cwchar> ...
1.26         (johnmadd 17-Dec-04): #if !defined(BOOST_NO_INTRINSIC_WCHAR_T) && (!defined(BOOST_NO_CWCHAR) || defined(sun) || defined(__sun))
1.20         (johnmadd 21-Oct-02): #include <wchar.h>
1.20         (johnmadd 21-Oct-02): #endif

> However, my wchar.h include
> file defines those two macros. I couldn't find what was going wrong, but
> moving the include of integer_traits.hpp in inheritance.cpp to the top
> of the list of include files solves the problem. 

You can't do that. Translation units that use Python.h have to start
with a Python header, and as a consequence translation units that use
Boost.Python have to start with a Boost.Python header.  Otherwise all
heck breaks loose.

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



From dgregor at cs.indiana.edu  Wed Nov 16 20:23:24 2005
From: dgregor at cs.indiana.edu (Doug Gregor)
Date: Wed, 16 Nov 2005 14:23:24 -0500
Subject: [C++-sig] Python help() not showing imported Boost.Python-wrapped
	classes
Message-ID: <c74eb713df2190504e2e3a6f9a492e4e@cs.indiana.edu>

Python's interactive help is not showing classes exported using 
Boost.Python. For instance, I have a module "_support" that uses 
Boost.Python like this:

BOOST_PYTHON_MODULE(_support)
{
   using boost::python::class_;
   using boost::python::enum_;
   using boost::python::init;

   class_<point2d>("Point2D");
   class_<point3d>("Point3D");
   enum_<default_color_type>("Color");
   def("distance", &point_2d_distance);
}

Now, that gets pulled into a package "bgl" that has this in its 
__init__.py:

	from _support import *

If I run help(bgl), something odd happens. Under the CLASSES heading, 
the "Color" enum gets documented but Point2D and Point3D do not. Under 
the FUNCTIONS heading, "distance" gets documented. \

If I run help(bgl._support), I get Point2D and Point3D documented (in 
the CLASSES section, of course) but neither "Color"  nor "distance".

Ideally, the _support module would be invisible to the user, and all of 
these classes and functions would show up when one uses help(bgl). Is 
this possible?

	Cheers,
	Doug



From dave at boost-consulting.com  Wed Nov 16 23:25:44 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 16 Nov 2005 17:25:44 -0500
Subject: [C++-sig] Python help() not showing imported
	Boost.Python-wrapped classes
References: <c74eb713df2190504e2e3a6f9a492e4e@cs.indiana.edu>
Message-ID: <u64qsfcwn.fsf@boost-consulting.com>

Doug Gregor <dgregor at cs.indiana.edu> writes:

> Python's interactive help is not showing classes exported using 
> Boost.Python. For instance, I have a module "_support" that uses 
> Boost.Python like this:
>
> BOOST_PYTHON_MODULE(_support)
> {
>    using boost::python::class_;
>    using boost::python::enum_;
>    using boost::python::init;
>
>    class_<point2d>("Point2D");
>    class_<point3d>("Point3D");
>    enum_<default_color_type>("Color");
>    def("distance", &point_2d_distance);
> }
>
> Now, that gets pulled into a package "bgl" that has this in its 
> __init__.py:
>
> 	from _support import *
>
> If I run help(bgl), something odd happens. Under the CLASSES heading, 
> the "Color" enum gets documented but Point2D and Point3D do not. Under 
> the FUNCTIONS heading, "distance" gets documented. \
>
> If I run help(bgl._support), I get Point2D and Point3D documented (in 
> the CLASSES section, of course) but neither "Color"  nor "distance".
>
> Ideally, the _support module would be invisible to the user, and all of 
> these classes and functions would show up when one uses help(bgl). Is 
> this possible?

>>> help(help)
Help on _Helper in module site object:

class _Helper(__builtin__.object)
 |  Define the built-in 'help'.
 |  This is a wrapper around pydoc.help (with a twist).
 
 <snip>

 http://www.python.org/doc/current/lib/module-pydoc.html

Getting pydoc to show the right things for Boost.Python has been an
ongoing struggle for some of us.  The cues pydoc takes from the
attributes in a module that it uses to decide what to document (and
how) are arcane and not well-understood.  I'd be happy if someone
wanted to take on the project of figuring out what was needed to make
everything come out right.

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



From amohr at pixar.com  Thu Nov 17 03:36:35 2005
From: amohr at pixar.com (Alex Mohr)
Date: Wed, 16 Nov 2005 18:36:35 -0800
Subject: [C++-sig] Something like boost::any ... custom conversions?
In-Reply-To: <u64r9l5rf.fsf@boost-consulting.com>
References: <43693C1F.9090101@pixar.com> <u64r9l5rf.fsf@boost-consulting.com>
Message-ID: <437BECB3.7010207@pixar.com>

So I'm back working on this now -- trying to register custom covnersions 
for my any-like object.

> You can do it for specific held types just by registering the
> appropriate from_python converters:

Cool -- that's all I want to do...

> I suggest you look at
> http://news.gmane.org/find-root.php?message_id=%3c1026918209.17167.2.camel%40hunsmac.mit.edu%3e
> and follow the links there.  It's probably about time to document this
> stuff for real and pull it out of the detail:: namespaces.

Thanks for the links.  I feel like I have a good general understanding 
of what needs doing.  The problem is I don't know how to tell whether 
the PyObject * that is passed to convertible() is pointing to (a wrapper 
of) my any-like object, and if it is, how to get it out.

I think this is what you were referring to when you said that the bigger 
issue is that I don't have the Python type object associated with my 
any-like class at compile time, correct?

If anybody has any hints for doing that, I'd greatly appreciate it.

Thanks,

Alex


From amohr at pixar.com  Thu Nov 17 19:22:20 2005
From: amohr at pixar.com (Alex Mohr)
Date: Thu, 17 Nov 2005 10:22:20 -0800
Subject: [C++-sig] Something like boost::any ... custom conversions?
In-Reply-To: <u64r9l5rf.fsf@boost-consulting.com>
References: <43693C1F.9090101@pixar.com> <u64r9l5rf.fsf@boost-consulting.com>
Message-ID: <437CCA5C.5040902@pixar.com>

> You can do it for specific held types just by registering the
> appropriate from_python converters:

Actually, I'm trying something like the code below now and it seems to 
be working so far.  I haven't tested extensively yet.  I'll have to 
register on the order of 100 converters like this and I'm wondering how 
much it might affect performance.

Is this a reasonable approach?  Is there a better way?  Is there some 
gap in my understanding?  Any comments are greatly appreciated.

Alex


// In this code, MyAny is a boost::any like object which has methods
// isHolding<T>() and get<T>()
//
using namespace boost::python;

template <typename T>
static void *convertible(PyObject *obj_ptr)
{
	extract<MyAny> getAny(obj_ptr);
	if (getAny.check()) {
		MyAny a = getAny();
		if (a.isHolding<T>())
			return obj_ptr;
	}
	return 0;
}

template <typename T>
static void construct(PyObject *obj_ptr,
	converter::rvalue_from_python_stage1_data *data)
{
	MyAny a = extract<MyAny>(obj_ptr);
	if (!a.isHolding<T>())
		// throw an exception: type error.
	void *storage =
		((converter::rvalue_from_python_storage<T>*)data)
		->storage.bytes;
	new (storage) T(a.get<T>());
	data->convertible = storage;
}

// Then I register converters like this
converter::registry::push_back(convertible<T>, construct<T>,
	boost::python::type_id<T>());


From meine at kogs1.informatik.uni-hamburg.de  Thu Nov 17 19:43:48 2005
From: meine at kogs1.informatik.uni-hamburg.de (Hans Meine)
Date: Thu, 17 Nov 2005 19:43:48 +0100
Subject: [C++-sig] Raw constructor (i.e. combination of make_constructor
	and raw_function)
In-Reply-To: <ud5o63iij.fsf@boost-consulting.com>
References: <200508220002.07595.hans_meine@gmx.net>
	<ud5o63iij.fsf@boost-consulting.com>
Message-ID: <200511171943.48729.meine@kogs.informatik.uni-hamburg.de>

Hi!

I want to pick up the discussion again.  Today, I discussed the problem 
(defining an __init__ which takes an arbitrary number of arguments) again 
with Ullrich K?the, and thanks to his stamina when digging through the depths 
of boost::python, we now have a working solution.

On Monday 22 August 2005 02:47, David Abrahams wrote:
> > I am trying to do something like
> >
> >     python::class_<MyVector>("PythonVector", python::no_init)
> >         .def("__init__", python::raw_function(&createVector, 1));
> >
> > But this gives the vector as first element of the args tuple to
> > createVector().
>
> What's wrong with that?

We could not find out what to do with args[0] or what it actually contains.  
It's an object which is said to be of type "PythonVector" in error messages, 
but one cannot extract<> anything from it, and we could not find any other 
use for it.  (See also my messages from the 22th of August)

After unsuccessfully trying to work out how the callable objects created by 
make_constructor initialize this "self" argument, Ullrich's final idea was to 
just pass this argument from our raw constructor function to another one 
created by make_constructor.  The rest of the arguments are passed as 
(sliced-off) tuple to a factory function (similar to what Ralf proposed back 
then).

Attached, you find the our new make_constructor implementation in the .hpp 
file, and a small testcase in the .cpp.  (Compile into rawconstructor.so)

Use like this from Python:

import rawconstructor


-- 
Ciao, /  /
     /--/
    /  / ANS
-------------- next part --------------
A non-text attachment was scrubbed...
Name: raw_constructor.hpp
Type: text/x-c++hdr
Size: 1243 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051117/10d7bae2/attachment.hpp>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: raw_constructor.cpp
Type: text/x-c++src
Size: 1124 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051117/10d7bae2/attachment.cpp>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: raw_constructor.py
Type: application/x-python
Size: 130 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051117/10d7bae2/attachment.bin>

From luvangel1002 at gmail.com  Fri Nov 18 07:06:48 2005
From: luvangel1002 at gmail.com (Petya Vasileva)
Date: Fri, 18 Nov 2005 00:06:48 -0600
Subject: [C++-sig] (no subject)
Message-ID: <39d50a260511172206y114f2298id9dfaf1a3d17ff0c@mail.gmail.com>

hey, i am trying to run this program, but it's not letting me. Would you
help me, please!
Thank you!!!

//This program reads in a customer's name and time *

//they were in a parking garage. It then calculates*

//and prints out the average payment of all the *

//people, the payment each person owes, and the *

//change each person gets back. It also calculates*

//and prints out anyone that is under the average *

//payment. *

//**************************************************

# include <iostream>

# include <fstream>

# include <iomanip>

using namespace std;

//****************************************************

// Stubs or prototypes of functions

double AveragePayment();

void Initialize(int & iNumCustomers, int & iNumUnderAverage);

void Titles(ofstream & OutputFile);

void ReadName(ifstream& InputFile, char strNm[]);

int ReadMinutes(ifstream & InputFile);

int Payment(int iMinutes);

void ChangeDue (int iPayment, int & iNumQuarters, int & iNumDimes, int &
iNumNickels, int & iNumPennies);

void PrintIt (ofstream & OutputFile, char strNm[], int iNumMin, intiPayment,
int iNumQuarters, int iNumDimes, int iNumNickels, int iNumPennies);

void UnderAverage(int iPayment, double dAveragePayment, int &iNumCustomers,
int & iNumUnderAverage, ofstream &ofPaymentFile);

void PrintAverage(double dAveragePayment, int iNumCustomers,
intiNumUnderAverage, ofstream &OutputFile);

void FreezeScreen();

//****************************************************

void main()

{

char strNm[20];//Customer name

int iNumMin; //Number of minutes the customer was in the garage

int iNumQuarters;

int iNumDimes;

int iNumNickels;

int iNumPennies;

int iNumCustomers;

int iNumUnderAverage;

//int iNumUnderAverage;

double dAveragePayment; //Average amount paid for parking

int iPayment; //Payment due

double iChangeDue; //Change due to the customer

double dPercentCust; //Percent of customers under the average

ifstream InputFile;

//Input file name

ofstream OutputFile;

//Output file name

AveragePayment();

InputFile.open("ASS9.txt");

Initialize (iNumCustomers, iNumUnderAverage);

 if (! InputFile)

{

cout << "Error in opening file" << endl;

cout << endl;

FreezeScreen();

return;

}

OutputFile.open("Output.txt");

Titles(OutputFile);

while(! InputFile.eof())

{

ReadName(InputFile, strNm);

ReadMinutes(InputFile);

Payment(iNumMin);

ChangeDue (iPayment, iNumQuarters, iNumDimes, iNumNickels, iNumPennies);

PrintIt (OutputFile, strNm, iNumMin, iPayment, iNumQuarters, iNumDimes,
iNumNickels, iNumPennies);

UnderAverage(iPayment, dAveragePayment, iNumCustomers, iNumUnderAverage,
OutputFile);

PrintAverage(dAveragePayment, iNumCustomers, iNumUnderAverage, OutputFile);

}

InputFile.close();

OutputFile.close();

FreezeScreen();

}

//****************************************************

//This function freezes the screen and stops the program.

void FreezeScreen()

{

fflush(stdin);

cout << "Hit enter to continue";

getchar();

}

//****************************************************

//This function returns the average amount paid by customers

//to park in the garage.

double AveragePayment()

{

ifstream InputFile;

char strNm[20];

int iNumCustomers;

int iNumMin;

int iPayment;

double dAveragePayment;

InputFile.open("ASS9.txt");

if (InputFile.fail())

{

cout << "Error Opening Input File" << endl;

FreezeScreen();

return -1;

}

else

{

while (! InputFile.eof())

{

InputFile.getline(strNm, 20);

Payment(iNumMin);

iNumCustomers++;

}

}

InputFile.close();

dAveragePayment = iPayment / iNumCustomers;

return dAveragePayment;

}

//****************************************************

//This function initializes all the variables

void Initialize(int & iNumCustomers, int & iNumUnderAverage)

{

int iCntr = 0;

int iNumMin = 0; //Number of minutes the customer was in the garage

int iNumQuarters = 0;

int iNumDimes = 0;

int iNumNickels = 0;

int iNumPennies = 0;

double dAveragePayment = 0; //Average amount paid for parking

double dPayment = 0; //Payment due

int iChangeDue = 0; //Change due to the customer

double dPercentCust = 0;

//int iNumCustomers = 0;

//int iNumUnderAverage = 0;

//return iNumCustomers, iNumUnderAverage;

}

//****************************************************

//This function prints a title for the table in the output file.

void Titles(ofstream & OutputFile)

{

OutputFile << setw(30) << "Parking Garage Payment" << endl;

}

//****************************************************

//This function reads in the name of the customer from the input file

void ReadName(ifstream & InputFile, char strNm[20])

{

InputFile.getline(strNm, 20);

}

//****************************************************

//This function reads in the number of minutes the customer was in the

//garage.

int ReadMinutes(ifstream & InputFile)

{

int iNumMin;

InputFile >> iNumMin;

return iNumMin;

}

//****************************************************

int Payment(int iNumMin)

{

int iPayment;

if(iNumMin <= 60)

{

iPayment = 50;

}

else

{

iPayment = (iNumMin - 60)* 1;

}

if(iPayment >= 1000)

{

iPayment = 1000;

}

return iPayment;

}

//****************************************************

void ChangeDue (int iPayment, int & iNumQuarters, int & iNumDimes, int &
iNumNickels, int & iNumPennies)

{

int iChangeDue;

iChangeDue = 1000 - iPayment;

iNumQuarters = iChangeDue % 25;

iNumDimes = iNumQuarters % 10;

iNumNickels = iNumDimes % 5;

iNumPennies = iNumNickels % 1;

//return iChangeDue, iNumQuarters, iNumDimes, iNumNickels, iNumPennies;

}

//****************************************************

void PrintIt (ofstream & OutputFile, char strNm[], int iNumMin, intiPayment,
int iNumQuarters, int iNumDimes, int iNumNickels, int iNumPennies)

{

OutputFile << setw(7) << strNm

<< setw(15)<< iNumMin

<< setw(21)<< iPayment

<< setw(20)<< iNumQuarters

<< setw(20)<< iNumDimes

<< setw(21)<< iNumNickels

<< setw(20)<< iNumPennies

<< endl;

}

//****************************************************

void UnderAverage(int iPayment, double dAveragePayment, int &iNumCustomers,
int & iNumUnderAverage, ofstream &OutputFile)

{

double dPercentCust;

while (iPayment < dAveragePayment)

{

OutputFile << "Under Average";

iNumUnderAverage ++;

}

dPercentCust = iNumUnderAverage/iNumCustomers;

//return iNumUnderAverage, iNumCustomers, dPercentCust;

}

//***************************************************

void PrintAverage(double dAveragePayment, int iNumCustomers,
intiNumUnderAverage, ofstream &OutputFile)

{

double dAveAmtPaid;

double dPercentCust;

OutputFile << "The average payment was: "

<< dAveAmtPaid

<< "."

<< endl;

OutputFile << "The percent of customers that are under the average is: "

<< dPercentCust

<< "."

<< endl;

}

//***************************************************
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051118/47767863/attachment.htm>

From hans_meine at gmx.net  Fri Nov 18 09:02:51 2005
From: hans_meine at gmx.net (Hans Meine)
Date: Fri, 18 Nov 2005 09:02:51 +0100
Subject: [C++-sig] Limits of implicitly_convertible?
Message-ID: <200511180903.02598.hans_meine@gmx.net>

Hi!

I could not get implicitly_convertible<list, MyVector>() to work *at all*, so 
I tried the example from the documentation.  (BTW: My compiler told me that 
operator int() had to be const, that should probably be fixed in the docs.)  
That simple example worked, but when I changed the constructor to take a list 
instead of an int, it ceased working.  Attached you'll find a minimal 
testcase.

-- 
Ciao, /  /                                                    .o.
     /--/                                                     ..o
    /  / ANS                                                  ooo
-------------- next part --------------
A non-text attachment was scrubbed...
Name: implicit.cpp
Type: text/x-c++src
Size: 412 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051118/74facba2/attachment.cpp>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: implicit.py
Type: application/x-python
Size: 86 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051118/74facba2/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051118/74facba2/attachment.pgp>

From noam.seker at intel.com  Sat Nov 19 00:19:22 2005
From: noam.seker at intel.com (Seker, Noam)
Date: Sat, 19 Nov 2005 01:19:22 +0200
Subject: [C++-sig] porting Qt 3.3.x to python
Message-ID: <6230DAB319ED764C98371CCE53E263CE081D3302@hasmsx403.ger.corp.intel.com>

Hi,
Is there an available boost/python package for Qt 3.3.x?
thanks

Noam Seker-Gafni
phone: +972 - 4 - 865 5260
inet:     8 - 465 5260
email:   noam.seker at intel.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051119/8a8029fa/attachment.htm>

From jan at langernetz.de  Sun Nov 20 13:19:50 2005
From: jan at langernetz.de (Jan Langer)
Date: Sun, 20 Nov 2005 13:19:50 +0100
Subject: [C++-sig] [Boost.Python] map_indexing_suite problem
Message-ID: <dlppl6$2sa$1@sea.gmane.org>

Hello,
I ran into a problem with Boost.Python exporting a std::map. I used the
following code below. I use Boost 1.33.0 and everything compiles fine.

------------
#include <boost/python.hpp>
#include <boost/python/suite/indexing/map_indexing_suite.hpp>

using namespace boost::python;

typedef std::map <int, std::string> map_t;

struct A
{
   map_t data;
   std::string str;
};

BOOST_PYTHON_MODULE(test)
{
   class_ <map_t> ("test_map")
     .def (map_indexing_suite <map_t> ());

   class_ <A> ("A")
     .def_readonly ("data", &A::data)
     .def_readwrite ("str", &A::str);
}
-------------

Then I run Python and try to use the module:

------------
laja at bridge:python> python
Python 2.4.1 (#1, Sep  2 2005, 09:48:20)
[GCC 2.95.2 19991024 (release)] on sunos5
Type "help", "copyright", "credits" or "license" for more information.
\>>> import test
\>>> a = test.A()
\>>> a.str="text"
\>>> a.str
'text'
\>>> a.data[0]="text2"
\>>> a.data[0]
Traceback (most recent call last):
   File "<stdin>", line 1, in ?
TypeError: No Python class registered for C++ class
t12basic_string3ZcZt18string_char_traits1ZcZt24__default_alloc_template2b0i0
-----------

I played around for a while but I find no solution.
Regards,
Jan



From jan at langernetz.de  Fri Nov 18 18:42:18 2005
From: jan at langernetz.de (Jan Langer)
Date: Fri, 18 Nov 2005 18:42:18 +0100
Subject: [C++-sig] [Boost.Python] map_indexing_suite problem
Message-ID: <dll3pr$a1v$1@sea.gmane.org>

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

Hello,
I ran into a problem with Boost.Python exporting a std::map. I used the
following code below. I use Boost 1.33.0 and everything compiles fine.

>>>>>>>>>>>
#include <boost/python.hpp>
#include <boost/python/suite/indexing/map_indexing_suite.hpp>

using namespace boost::python;

typedef std::map <int, std::string> map_t;

struct A
{
  map_t data;
  std::string str;
};

BOOST_PYTHON_MODULE(test)
{
  class_ <map_t> ("test_map")
    .def (map_indexing_suite <map_t> ());

  class_ <A> ("A")
    .def_readonly ("data", &A::data)
    .def_readwrite ("str", &A::str);
}
<<<<<<<<<<<<

Then I run Python and try to use the module:

>>>>>>>>>>>>
laja at bridge:python> python
Python 2.4.1 (#1, Sep  2 2005, 09:48:20)
[GCC 2.95.2 19991024 (release)] on sunos5
Type "help", "copyright", "credits" or "license" for more information.
>>> import test
>>> a = test.A()
>>> a.str="text"
>>> a.str
'text'
>>> a.data[0]="text2"
>>> a.data[0]
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
TypeError: No Python class registered for C++ class
t12basic_string3ZcZt18string_char_traits1ZcZt24__default_alloc_template2b0i0
<<<<<<<<<<<<

I played around for a while but I find no solution.
Regards,
Jan
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (SunOS)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFDfhJ5yJOS3PG4wcwRAihcAJsFhjuOtOykJlyC9JKsg1yobwOsKgCgxQa5
l6GQlwbGXfjizZ3BVVo81Io=
=Jebw
-----END PGP SIGNATURE-----



From djowel at gmail.com  Mon Nov 21 04:54:29 2005
From: djowel at gmail.com (Joel de Guzman)
Date: Mon, 21 Nov 2005 11:54:29 +0800
Subject: [C++-sig] [Boost.Python] map_indexing_suite problem
In-Reply-To: <dlppl6$2sa$1@sea.gmane.org>
References: <dlppl6$2sa$1@sea.gmane.org>
Message-ID: <438144F5.2020805@boost-consulting.com>

Jan Langer wrote:
> Hello,
> I ran into a problem with Boost.Python exporting a std::map. I used the
> following code below. I use Boost 1.33.0 and everything compiles fine.
> 

Have you tried setting NoProxy template parameter to true?

    class_ <map_t, true> ("test_map")

Regards,
-- 
Joel de Guzman
http://www.boost-consulting.com
http://spirit.sf.net


From djowel at gmail.com  Mon Nov 21 05:54:22 2005
From: djowel at gmail.com (Joel de Guzman)
Date: Mon, 21 Nov 2005 12:54:22 +0800
Subject: [C++-sig] [Boost.Python] map_indexing_suite problem
In-Reply-To: <438144F5.2020805@boost-consulting.com>
References: <dlppl6$2sa$1@sea.gmane.org>
	<438144F5.2020805@boost-consulting.com>
Message-ID: <438152FE.9090704@boost-consulting.com>

Joel de Guzman wrote:
> Jan Langer wrote:
> 
>>Hello,
>>I ran into a problem with Boost.Python exporting a std::map. I used the
>>following code below. I use Boost 1.33.0 and everything compiles fine.
>>
> 
> 
> Have you tried setting NoProxy template parameter to true?
> 
>     class_ <map_t, true> ("test_map")

This is the same issue as discussed in:
http://mail.python.org/pipermail/c++-sig/2004-March/007029.html

I've just added a fix for this that regards std::string and
std::complex as no-proxy types.

Regards,
-- 
Joel de Guzman
http://www.boost-consulting.com
http://spirit.sf.net


From duranlef at iro.umontreal.ca  Mon Nov 21 18:04:50 2005
From: duranlef at iro.umontreal.ca (=?ISO-8859-1?Q?Fran=E7ois_Duranleau?=)
Date: Mon, 21 Nov 2005 12:04:50 -0500 (EST)
Subject: [C++-sig] porting Qt 3.3.x to python
In-Reply-To: <6230DAB319ED764C98371CCE53E263CE081D3302@hasmsx403.ger.corp.intel.com>
References: <6230DAB319ED764C98371CCE53E263CE081D3302@hasmsx403.ger.corp.intel.com>
Message-ID: <Pine.LNX.4.63.0511211204070.4927@brocoli.iro.umontreal.ca>

On Sat, 19 Nov 2005, Seker, Noam wrote:

> Hi,
> Is there an available boost/python package for Qt 3.3.x?
> thanks

Here you go:  http://www.riverbankcomputing.co.uk/pyqt/

-- 
Fran?ois Duranleau
LIGUM, Universit? de Montr?al

"[...] consider McDonalds -- it is fast, cheap, and used by millions.
  However, it also contributes to obesity, heart disease, and (arguably)
  deforestation."
                  - Prof. Gene Spafford, in email about Windows supplanting
                    other operating systems in Universities

From n_habili at hotmail.com  Tue Nov 22 04:20:52 2005
From: n_habili at hotmail.com (Nariman Habili)
Date: Tue, 22 Nov 2005 03:20:52 +0000
Subject: [C++-sig] SciPy Core
Message-ID: <BAY102-F1625CDF09362BB0DE5E715F9520@phx.gbl>

I'm wondering whether the boost python numeric wrappers will be updated to 
scipy core?
Thanks.




From noam.seker at intel.com  Tue Nov 22 08:23:45 2005
From: noam.seker at intel.com (Seker, Noam)
Date: Tue, 22 Nov 2005 09:23:45 +0200
Subject: [C++-sig] porting Qt 3.3.x to python
Message-ID: <6230DAB319ED764C98371CCE53E263CE082284BB@hasmsx403.ger.corp.intel.com>

Thanks.
PyQt package is built with SIP. I tried to use PyQt objects from a code wrapped with boost/python but, at least in my na?ve example, same object (name) in boost and in PyQt have different 'signature'. 
Is there a way to bridge between the 2 python wrappers?
Can I tell boost to use qt objects from PyQt instead of generating it's own version of them?

-----Original Message-----
From: c++-sig-bounces at python.org [mailto:c++-sig-bounces at python.org] On Behalf Of Fran?ois Duranleau
Sent: Monday, November 21, 2005 7:05 PM
To: Development of Python/C++ integration
Subject: Re: [C++-sig] porting Qt 3.3.x to python

On Sat, 19 Nov 2005, Seker, Noam wrote:

> Hi,
> Is there an available boost/python package for Qt 3.3.x?
> thanks

Here you go:  http://www.riverbankcomputing.co.uk/pyqt/

-- 
Fran?ois Duranleau
LIGUM, Universit? de Montr?al

"[...] consider McDonalds -- it is fast, cheap, and used by millions.
  However, it also contributes to obesity, heart disease, and (arguably)
  deforestation."
                  - Prof. Gene Spafford, in email about Windows supplanting
                    other operating systems in Universities


From etaurel at cells.es  Tue Nov 22 12:31:37 2005
From: etaurel at cells.es (Emmanuel Taurel)
Date: Tue, 22 Nov 2005 12:31:37 +0100
Subject: [C++-sig] returning reference to enum
Message-ID: <0IQC008YYTDH63@cells-s02.cells.es>

Hello everybody,

 

I have a very simple example where I try to wrap a C++ class. In this class
there is a method which returns a reference to an internal variable which is
an enumeration. I try to wrap this method with the return_internal_reference
Call Policy but I have a compiler error.

 

First here is my little C++ file

 

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

 

#include <boost/python.hpp>

using namespace boost::python;

 

enum DispLevel

{

            OPERATOR = 0,

            EXPERT

};

 

class MyClass

{

public:

            MyClass() {}

            ~MyClass() {}

 

            void set_enum(DispLevel dl) {disp=dl;}

            DispLevel &get_enum() {return disp;}

 

protected:

            DispLevel disp;

};

 

BOOST_PYTHON_MODULE(PyTangoDs)

{

            enum_<DispLevel>("DispLevel")

            .value("OPERATOR",OPERATOR)

            .value("EXPERT",EXPERT)

            ;

                        

            class_<MyClass>("MyClass")

            .def("set_enum",&MyClass::set_enum)

 
.def("get_enum",&MyClass::get_enum,return_internal_reference<>())

            ;

}

 

The error I get from the compiler is : 

 

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

 

gcc-C++-action bin/enum/PyTangoDs.so/gcc/debug/shared-linkable-true/py_ds.o

/home/etaurel/src/boost_1_33_0/boost/python/object/make_instance.hpp: In
static

   member function `static PyObject*

   boost::python::objects::make_instance_impl<T, Holder,

   Derived>::execute(Arg&) [with Arg = DispLevel*, T = DispLevel, Holder =

   boost::python::objects::pointer_holder<DispLevel*, DispLevel>, Derived =

   boost::python::objects::make_ptr_instance<DispLevel,

   boost::python::objects::pointer_holder<DispLevel*, DispLevel> >]':

/home/etaurel/src/boost_1_33_0/boost/python/to_python_indirect.hpp:96:
instantiated from `static PyObject*
boost::python::detail::make_reference_holder::execute(T*) [with T =
DispLevel]'

/home/etaurel/src/boost_1_33_0/boost/python/to_python_indirect.hpp:60:
instantiated from `PyObject* boost::python::to_python_indirect<T,
MakeHolder>::execute(const U&, mpl_::bool_<false>) const [with U =
DispLevel, T = DispLevel&, MakeHolder =
boost::python::detail::make_reference_holder]'

/home/etaurel/src/boost_1_33_0/boost/python/to_python_indirect.hpp:37:
instantiated from `PyObject* boost::python::to_python_indirect<T,
MakeHolder>::operator()(const U&) const [with U = DispLevel, T = DispLevel&,
MakeHolder = boost::python::detail::make_reference_holder]'

/home/etaurel/src/boost_1_33_0/boost/python/detail/invoke.hpp:88:
instantiated from `PyObject*
boost::python::detail::invoke(boost::python::detail::invoke_tag_<false,
true>, const RC&, F&, TC&) [with RC =
boost::python::to_python_indirect<DispLevel&,
boost::python::detail::make_reference_holder>, F = DispLevel&(MyClass::*)(),
TC = boost::python::arg_from_python<MyClass&>]'

/home/etaurel/src/boost_1_33_0/boost/python/detail/caller.hpp:199:
instantiated from `PyObject* boost::python::detail::caller_arity<1>::impl<F,
Policies, Sig>::operator()(PyObject*, PyObject*) [with F =
DispLevel&(MyClass::*)(), Policies =
boost::python::return_internal_reference<1,
boost::python::default_call_policies>, Sig = boost::mpl::vector2<DispLevel&,
MyClass&>]'

/home/etaurel/src/boost_1_33_0/boost/python/object/py_function.hpp:38:
instantiated from `PyObject*
boost::python::objects::caller_py_function_impl<Caller>::operator()(PyObject
*, PyObject*) [with Caller =
boost::python::detail::caller<DispLevel&(MyClass::*)(),
boost::python::return_internal_reference<1,
boost::python::default_call_policies>, boost::mpl::vector2<DispLevel&,
MyClass&> >]'

/home/etaurel/src/boost_1_33_0/boost/detail/call_traits.hpp:83:
instantiated from here

/home/etaurel/src/boost_1_33_0/boost/python/object/make_instance.hpp:24:
error: incomplete

   type `boost::STATIC_ASSERTION_FAILURE<false>' does not have member
`value'

 

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

 

Any idea welcome. Thank's very much in advance

 

Emmanuel Taurel (etaurel at cells.es)

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051122/2e383d6e/attachment.htm>

From roman.yakovenko at gmail.com  Tue Nov 22 12:44:13 2005
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 22 Nov 2005 13:44:13 +0200
Subject: [C++-sig] returning reference to enum
In-Reply-To: <0IQC008YYTDH63@cells-s02.cells.es>
References: <0IQC008YYTDH63@cells-s02.cells.es>
Message-ID: <7465b6170511220344v6260f4eet9f3d9284e15ca80@mail.gmail.com>

On 11/22/05, Emmanuel Taurel <etaurel at cells.es> wrote:
> Hello everybody,
>
> I have a very simple example where I try to wrap a C++ class. In this class
> there is a method which returns a reference to an internal variable which is
> an enumeration. I try to wrap this method with the return_internal_reference
> Call Policy but I have a compiler error.
>
> Any idea welcome. Thank's very much in advance

If you can change the class, then change it to return that enum value by value.
In this case you don't need to specify call policies.

If you can not to change it, then return_value_policy<
copy_non_const_reference > will do the job.

>
>
> Emmanuel Taurel (etaurel at cells.es)
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>
>
>


From jan at langernetz.de  Tue Nov 22 14:07:06 2005
From: jan at langernetz.de (Jan Langer)
Date: Tue, 22 Nov 2005 14:07:06 +0100
Subject: [C++-sig] [Boost.Python] map_indexing_suite problem
In-Reply-To: <438152FE.9090704@boost-consulting.com>
References: <dlppl6$2sa$1@sea.gmane.org>	<438144F5.2020805@boost-consulting.com>
	<438152FE.9090704@boost-consulting.com>
Message-ID: <dlv55r$oac$1@sea.gmane.org>

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

Joel de Guzman wrote:
>>>I ran into a problem with Boost.Python exporting a std::map. I used the
>>>following code below. I use Boost 1.33.0 and everything compiles fine.
>>
>>Have you tried setting NoProxy template parameter to true?
>>
>>    class_ <map_t, true> ("test_map")
> 
> This is the same issue as discussed in:
> http://mail.python.org/pipermail/c++-sig/2004-March/007029.html
> 
> I've just added a fix for this that regards std::string and
> std::complex as no-proxy types.

Hello,
I tried setting NoProxy to true and it worked in the example I posted.
However my real example is a little more complicated. I copied the
relevant part below:

  struct domain
  {};

  struct graph
  {
    typedef std::map <int, boost::shared_ptr <domain> > domains_t;
    domains_t domains_;
  };

  class_<domain, boost::shared_ptr <domain> > ("domain");

//  objects::class_value_wrapper <boost::shared_ptr<domain>,
//     objects::make_ptr_instance <domain,
//     objects::pointer_holder <boost::shared_ptr<domain>, domain> > > ();

  class_<graph::domains_t> ("domains_map")
    .def (map_indexing_suite<graph::domains_t, true> ());

  class_<graph> ("graph")
    .def_readonly ("domains", &graph::domains_);

This code (used in a module called pdg) can be compiled and imported in
Python:

import pdg
g = pdg.graph ()
for el in g.domains:
    print el.key(),
    dom = el.data()

But the last line gives an error:

TypeError: No Python class registered for C++ class
Q25boostt10shared_ptr1ZQ23pdg6domain

If I comment out the three lines with class_value_wrapper (I saw
something like that in the examples) and remove shared_ptr from the
class_ definition of domain, I get the same error.

I hope that makes my problem clearer.
Regards,
Jan Langer
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (SunOS)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFDgxf5yJOS3PG4wcwRAjrgAJ9obR9b3g/YA21d4qr8yLH3pMx2SwCgpCQY
eXHTX+NCzzMWSpRdHdgLoIk=
=vFO5
-----END PGP SIGNATURE-----



From etaurel at cells.es  Tue Nov 22 14:18:00 2005
From: etaurel at cells.es (Emmanuel Taurel)
Date: Tue, 22 Nov 2005 14:18:00 +0100
Subject: [C++-sig] returning reference to enum
In-Reply-To: <7465b6170511220344v6260f4eet9f3d9284e15ca80@mail.gmail.com>
Message-ID: <0IQC008BUYAS63@cells-s02.cells.es>

Hello,

I can't change the class therefore, as suggested I have used the
copy_non_const_reference ResultConverterGenerator.
It works fine. Thank's for the tip but in this case, my enum is copied in a
new Python object. Is this correct ?

A last question for my understanding. Why using return_internal_reference
works fine when returning a reference to an already wrapped class (like in
the return_internal_reference documentation example) and it does not when
returning an enumeration ?

Thank's for your answer

Emmanuel Taurel (etaurel at cells.es)

-----Original Message-----
From: c++-sig-bounces at python.org [mailto:c++-sig-bounces at python.org] On
Behalf Of Roman Yakovenko
Sent: Tuesday, November 22, 2005 12:44 PM
To: Development of Python/C++ integration
Subject: Re: [C++-sig] returning reference to enum

On 11/22/05, Emmanuel Taurel <etaurel at cells.es> wrote:
> Hello everybody,
>
> I have a very simple example where I try to wrap a C++ class. In this
class
> there is a method which returns a reference to an internal variable which
is
> an enumeration. I try to wrap this method with the
return_internal_reference
> Call Policy but I have a compiler error.
>
> Any idea welcome. Thank's very much in advance

If you can change the class, then change it to return that enum value by
value.
In this case you don't need to specify call policies.

If you can not to change it, then return_value_policy<
copy_non_const_reference > will do the job.

>
>
> Emmanuel Taurel (etaurel at cells.es)
> _______________________________________________
> C++-sig mailing list
> C++-sig at python.org
> http://mail.python.org/mailman/listinfo/c++-sig
>
>
>
_______________________________________________
C++-sig mailing list
C++-sig at python.org
http://mail.python.org/mailman/listinfo/c++-sig



From rwgk at yahoo.com  Tue Nov 22 18:03:34 2005
From: rwgk at yahoo.com (Ralf W. Grosse-Kunstleve)
Date: Tue, 22 Nov 2005 09:03:34 -0800 (PST)
Subject: [C++-sig] porting Qt 3.3.x to python
In-Reply-To: <6230DAB319ED764C98371CCE53E263CE082284BB@hasmsx403.ger.corp.intel.com>
Message-ID: <20051122170334.15526.qmail@web31504.mail.mud.yahoo.com>

--- "Seker, Noam" <noam.seker at intel.com> wrote:

> Thanks.
> PyQt package is built with SIP. I tried to use PyQt objects from a code
> wrapped with boost/python but, at least in my na?ve example, same object
> (name) in boost and in PyQt have different 'signature'. 
> Is there a way to bridge between the 2 python wrappers?

I am not aware of interfaces between SIP-generated wrappers and Boost.Python.

> Can I tell boost to use qt objects from PyQt instead of generating it's own
> version of them?

That's probably something you'd have to figure out yourself. A few months ago
someone asked a similar question regarding SWIG. It was surprisingly easy to
support passing SWIG-wrapped objects to Boost.Python wrapped functions:

http://cvs.sourceforge.net/viewcvs.py/cctbx/boost_adaptbx/swig_arg.hpp?view=markup

http://cvs.sourceforge.net/viewcvs.py/cctbx/boost_adaptbx/swig_args_ext.cpp?view=markup

If you are lucky, maybe something similar works for SIP. You'd have to find out
how the SIP wrapping mechanism works.

HTH,
        Ralf



	
		
__________________________________ 
Yahoo! Mail - PC Magazine Editors' Choice 2005 
http://mail.yahoo.com


From mjkeyes at sbcglobal.net  Wed Nov 23 05:39:25 2005
From: mjkeyes at sbcglobal.net (Matt)
Date: Tue, 22 Nov 2005 22:39:25 -0600
Subject: [C++-sig] Any word on when Py_Finalize will be implemented in
	boost::python yet?
Message-ID: <dm0rln$ccs$1@sea.gmane.org>

Just curious. 





From pth at suse.de  Fri Nov 25 18:16:15 2005
From: pth at suse.de (Philipp Thomas)
Date: Fri, 25 Nov 2005 18:16:15 +0100
Subject: [C++-sig] type-punning in boost python code.
Message-ID: <20051125171615.GE7473@paradies.suse.de>

our gcc 4.1 also analyses C++ code for violation of the C/C++ aliasing
rules (mainline doesn't do so yet). For the code in boost 1.33.1 I get 

warning: dereferencing type-punned pointer will break strict-aliasing rules

for the following files:

 /usr/src/packages/BUILD/boost_1_33_1/libs/python/build/../src/long.cpp: 12,19, 26
 /usr/src/packages/BUILD/boost_1_33_1/libs/python/build/../src/dict.cpp: 32
 /usr/src/packages/BUILD/boost_1_33_1/libs/python/build/../src/tuple.cpp:12
 /usr/src/packages/BUILD/boost_1_33_1/libs/python/build/../src/str.cpp: 12
 /usr/src/packages/BUILD/boost_1_33_1/libs/python/build/../src/object/class.cpp:543, 553
 /usr/src/packages/BUILD/boost_1_33_1/libs/python/build/../src/list.cpp: 15

as our package build system treats this warning as an error, I will use -fno-strict-aliasing
for gcc as I'm not familiar with that code and I want to get the package built.

For boost you should also consider either using said gcc switch or fix the
code to not break strict-aliasing rules. I'd recommend doing the latter as
switching off aliasing analysis will inhibit some optimisations.

Philipp


From djowel at gmail.com  Sat Nov 26 17:26:34 2005
From: djowel at gmail.com (Joel de Guzman)
Date: Sun, 27 Nov 2005 00:26:34 +0800
Subject: [C++-sig] [Boost.Python] map_indexing_suite problem
In-Reply-To: <dlv55r$oac$1@sea.gmane.org>
References: <dlppl6$2sa$1@sea.gmane.org>	<438144F5.2020805@boost-consulting.com>	<438152FE.9090704@boost-consulting.com>
	<dlv55r$oac$1@sea.gmane.org>
Message-ID: <43888CBA.9070907@boost-consulting.com>

Jan Langer wrote:

> Hello,
> I tried setting NoProxy to true and it worked in the example I posted.
> However my real example is a little more complicated. I copied the
> relevant part below:

[snips]
> 
> TypeError: No Python class registered for C++ class
> Q25boostt10shared_ptr1ZQ23pdg6domain

Ok it is a bug. It's fixed now in the CVS. I updated 
map_indexing_suite.cpp/py to prove the fix. All tests
are passing.

Hope that helps. Thanks for the report.

Regards,
-- 
Joel de Guzman
http://www.boost-consulting.com
http://spirit.sf.net


From dave at boost-consulting.com  Sat Nov 26 22:59:02 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Sat, 26 Nov 2005 16:59:02 -0500
Subject: [C++-sig] Any word on when Py_Finalize will be implemented in
	boost::python yet?
References: <dm0rln$ccs$1@sea.gmane.org>
Message-ID: <uveyfnk9l.fsf@boost-consulting.com>

"Matt" <mjkeyes at sbcglobal.net> writes:

> Just curious. 

No word.  It could be done by an ambitious volunteer (or hired
professional) in a few days, though.

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



From s_sourceforge at nedprod.com  Sun Nov 27 11:08:45 2005
From: s_sourceforge at nedprod.com (Niall Douglas)
Date: Sun, 27 Nov 2005 10:08:45 -0000
Subject: [C++-sig] Patch implementing void * support
In-Reply-To: <436E4390.14657.548C6F9@localhost>
Message-ID: <438985AD.32440.B7578@s_sourceforge.nedprod.com>

Attached is a slightly fixed version of the patch. When is this patch 
going to be applied?

On 6 Nov 2005 at 17:55, Niall Douglas wrote:

> The enclosed patch adds void * support to Boost.Python by making void 
> an opaque type. Code that uses void * "just works" with this patch 
> with no further effort required.
> 
> Also enclosed is a testcase and the docs have been modified.
> 
> Ran full testsuite on msvc7.1 and msvc6 - everything works. I don't 
> have GCC to hand ATM, but apart from maybe a missing typename 
> specifier I can't see anything which would fault.

I understand from Roman that I have too many typenames in there, but 
other than this trivial fix the patch works perfectly on GCC.

Cheers,
Niall



-------------- next part --------------
The following section of this message contains a file attachment
prepared for transmission using the Internet MIME message format.
If you are using Pegasus Mail, or any other MIME-compliant system,
you should be able to save it or view it from within your mailer.
If you cannot, please ask your system administrator for assistance.

   ---- File information -----------
     File:  voidptrpatch2.diff
     Date:  27 Nov 2005, 10:05
     Size:  4781 bytes.
     Type:  Unknown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: voidptrpatch2.diff
Type: application/octet-stream
Size: 4780 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051127/2cc0d7b3/attachment.obj>
-------------- next part --------------
The following section of this message contains a file attachment
prepared for transmission using the Internet MIME message format.
If you are using Pegasus Mail, or any other MIME-compliant system,
you should be able to save it or view it from within your mailer.
If you cannot, please ask your system administrator for assistance.

   ---- File information -----------
     File:  voidptr.cpp
     Date:  6 Nov 2005, 17:38
     Size:  892 bytes.
     Type:  Program-source
-------------- next part --------------
A non-text attachment was scrubbed...
Name: voidptr.cpp
Type: application/octet-stream
Size: 891 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051127/2cc0d7b3/attachment-0001.obj>
-------------- next part --------------
The following section of this message contains a file attachment
prepared for transmission using the Internet MIME message format.
If you are using Pegasus Mail, or any other MIME-compliant system,
you should be able to save it or view it from within your mailer.
If you cannot, please ask your system administrator for assistance.

   ---- File information -----------
     File:  voidptr.py
     Date:  6 Nov 2005, 17:24
     Size:  1114 bytes.
     Type:  Unknown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: voidptr.py
Type: application/octet-stream
Size: 1114 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051127/2cc0d7b3/attachment-0002.obj>

From roman.yakovenko at gmail.com  Sun Nov 27 11:34:54 2005
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Sun, 27 Nov 2005 12:34:54 +0200
Subject: [C++-sig] Patch implementing void * support
In-Reply-To: <438985AD.32440.B7578@s_sourceforge.nedprod.com>
References: <436E4390.14657.548C6F9@localhost>
	<438985AD.32440.B7578@s_sourceforge.nedprod.com>
Message-ID: <7465b6170511270234o748fcc97lc644075b95d613b9@mail.gmail.com>

On 11/27/05, Niall Douglas <s_sourceforge at nedprod.com> wrote:
> Attached is a slightly fixed version of the patch. When is this patch
> going to be applied?
>
> I understand from Roman that I have too many typenames in there, but
> other than this trivial fix the patch works perfectly on GCC.

Small mistake: you don't have to declare void as an opaque type for
automatic void * usage

You don't need those lines ( at least on GCC 3.3 ):
+
+// Declare void as an opaque type for automatic void * usage
+namespace boost { namespace python {
+    template<>
+    inline type_info type_id<void>(BOOST_PYTHON_EXPLICIT_TT_DEF(void))
+    {
+        return type_info (typeid (void *));
+    }
+    template<>
+    inline type_info type_id<const volatile void>(
+        BOOST_PYTHON_EXPLICIT_TT_DEF(const volatile void))
+    {
+        return type_info (typeid (void *));
+    }
+}}


> Cheers,
> Niall


Roman Yakovenko.


From Yves_Secretan at inrs-ete.uquebec.ca  Mon Nov 28 17:46:58 2005
From: Yves_Secretan at inrs-ete.uquebec.ca (Yves Secretan)
Date: Mon, 28 Nov 2005 11:46:58 -0500
Subject: [C++-sig] iterator over list and sequences
Message-ID: <IOENLEGGNAMHEFJODOADIEMHEDAA.Yves_Secretan@inrs-ete.uquebec.ca>

Hello all,

In search for a simple way to iterate in C++ over a boost::python::list, I
only found a discussion about how nice it would be to have one (
http://mail.python.org/pipermail/c++-sig/2003-March/003528.html). So I wrote
a set of C++ iterators and ended up with a syntax a little more complex than
what was originally proposed by David Abrahams, but maybe I missed
something.

typedef value_iterator<int> iter_type;
for (iter_type i = begin<iter_type>(lst);
   i != end<iter_type>();
   ++i)
{
   whatever(*i);
}

The tests provided have been run with gcc 3.3 mingw, visual c++ 7.0 and
Intel 7.1 on WindowsXP (Boost 1.33.0, Python 2.4).

Is there any interest?

---------------------------------------------------------------
Yves Secretan, Professeur
INRS-ETE
490, rue de la Couronne
Qu?bec, Qu?bec
G1K 9A9 CANADA

tel: (418) 654 38 48   fax: (418) 654 26 00
e-mail: Yves_Secretan at ete.inrs.ca
---------------------------------------------------------------

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051128/029655fe/attachment.htm>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: sequence_iterator.txt
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051128/029655fe/attachment.txt>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test_sequence_iterator.cpp
Type: application/octet-stream
Size: 4187 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051128/029655fe/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: sequence_iterator.hpp
Type: application/octet-stream
Size: 7325 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051128/029655fe/attachment-0001.obj>

From tippetts at lanl.gov  Tue Nov 29 02:15:47 2005
From: tippetts at lanl.gov (Trevor Tippetts)
Date: Mon, 28 Nov 2005 18:15:47 -0700
Subject: [C++-sig] boost.python and std::valarray
Message-ID: <1133226947.4218.8.camel@localhost>

I'm relatively new (and very impressed) user of boost.python. I do have
a question, however. When I try to access members typed in c++ as
std::valarray<> in python, I get errors like this:

No Python class registered for C++ class std::valarray<int>

After searching through list archives and online docs and wikis, I have
come to the conclusion that pyste and boost.python do not have a
built-in wrapper for std::valarray<> and std::vector<> types. Is that a
correct assumption?

I found some examples people have given for wrapping std::vector<>, but
not for std::valarray<>. Does anyone know of a good example
implementation to which the rest of us could refer?

Thanks,
Trevor
-- 
After due consideration, this message was judged by Trevor Tippetts not
to contain technical or programmatic information and therefore does not
require classification review.
--
Trevor Tippetts
Los Alamos National Laboratory -- ESA-WR
T080
Los Alamos NM  87545

505-667-9009
<tippetts at lanl.gov>



From eric at boost-consulting.com  Tue Nov 29 05:48:36 2005
From: eric at boost-consulting.com (Eric Niebler)
Date: Mon, 28 Nov 2005 20:48:36 -0800
Subject: [C++-sig] iterator over list and sequences
In-Reply-To: <IOENLEGGNAMHEFJODOADIEMHEDAA.Yves_Secretan@inrs-ete.uquebec.ca>
References: <IOENLEGGNAMHEFJODOADIEMHEDAA.Yves_Secretan@inrs-ete.uquebec.ca>
Message-ID: <438BDDA4.3020507@boost-consulting.com>

Yves Secretan wrote:
> 
> In search for a simple way to iterate in C++ over a boost::python::list, 
> I only found a discussion about how nice it would be to have one 
> (http://mail.python.org/pipermail/c++-sig/2003-March/003528.html). So I 
> wrote a set of C++ iterators and ended up with a syntax a little more 
> complex than what was originally proposed by David Abrahams, but maybe I 
> missed something.
> 


We've already got one. :-) I wrote a STL iterator for python sequences. 
It is an input iterator over any iterable python object, not just lists. 
It has not yet been part of an official Boost release, but you can find 
it in Boost main CVS at boost/python/stl_iterator.hpp. It'll be part of 
the next major release.

Hope that helps,

-- 
Eric Niebler
Boost Consulting
www.boost-consulting.com


From roman.yakovenko at gmail.com  Tue Nov 29 06:27:03 2005
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 29 Nov 2005 07:27:03 +0200
Subject: [C++-sig] boost.python and std::valarray
In-Reply-To: <1133226947.4218.8.camel@localhost>
References: <1133226947.4218.8.camel@localhost>
Message-ID: <7465b6170511282127k3e18c53flca5bf1064364b79e@mail.gmail.com>

On 11/29/05, Trevor Tippetts <tippetts at lanl.gov> wrote:
> I'm relatively new (and very impressed) user of boost.python. I do have
> a question, however. When I try to access members typed in c++ as
> std::valarray<> in python, I get errors like this:
>
> No Python class registered for C++ class std::valarray<int>
>
> After searching through list archives and online docs and wikis, I have
> come to the conclusion that pyste and boost.python do not have a
> built-in wrapper for std::valarray<> and std::vector<> types. Is that a
> correct assumption?

Partial, boost.python have this:

http://boost.org/libs/python/doc/v2/indexing.html

Also you can write ( and contribute ) template class similar to
indexing_suite::Container
that will expose std::valarray to python.

An other valid approach, is to copy std::valarray to
boost::python::list and return it.

See http://boost.org/libs/python/pyste/doc/wrappers.html

> I found some examples people have given for wrapping std::vector<>, but
> not for std::valarray<>. Does anyone know of a good example
> implementation to which the rest of us could refer?
>
> Thanks,
> Trevor


From roman.yakovenko at gmail.com  Tue Nov 29 09:47:26 2005
From: roman.yakovenko at gmail.com (Roman Yakovenko)
Date: Tue, 29 Nov 2005 10:47:26 +0200
Subject: [C++-sig] [Announce] pyplusplus 0.6.0
Message-ID: <7465b6170511290047u65ff91b4sc704950062fc3f88@mail.gmail.com>

Hi. I would like announce of new version of pyplusplus (0.5).

What is pyplusplus?

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

You can find pyplusplus here: http://www.language-binding.net/
You can download pyplusplus from
http://sourceforge.net/project/showfiles.php?group_id=118209.

Changes:

    1. Code repository has been introduced. This repository contains classes
       and functions that will help users to export different C++ classes and
       declarations. Right now this repository contains two classes:

       array_1_t

      const_array_1_t

       Those classes helps to export static, single dimension arrays.
As you can
       guess static, single dimension arrays are now supported.

    2. Code generation has been improved.

    3. Code generation speed has been improved.

    4. If you have Niall Douglas void* patch, then you can enjoy from
       automatically set call policies.

    5. Bit fields can be accessed ( get\\set ) from Python.

    6. Creating custom code creater example has been added.

    7. To Pyste comparison has been wrote.

    8. Using this version it is possible to export most of TnFOX Python
       bindings.

Roman Yakovenko


From jan at langernetz.de  Tue Nov 29 12:20:05 2005
From: jan at langernetz.de (Jan Langer)
Date: Tue, 29 Nov 2005 12:20:05 +0100
Subject: [C++-sig] [Boost.Python] exporting GMP numbers to Python
Message-ID: <dmhdha$lv2$1@sea.gmane.org>

Hello,
I have a c++ library I want to use as a python module. This works quite
nice now (thanks to Joel's bugfix). However, some functions of the
library return arbitrary precision numbers (GMP) and I need to access
them in Python.

I found a working GMP Python wrapper in the GMPY module, but I have
simply no idea how use it for my purpose.
Maybe someone has a link or something similar for me.
Jan
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 252 bytes
Desc: OpenPGP digital signature
URL: <http://mail.python.org/pipermail/cplusplus-sig/attachments/20051129/b0ac4f2a/attachment.pgp>

From meine at kogs1.informatik.uni-hamburg.de  Tue Nov 29 15:30:20 2005
From: meine at kogs1.informatik.uni-hamburg.de (Hans Meine)
Date: Tue, 29 Nov 2005 15:30:20 +0100
Subject: [C++-sig] [Boost.Python] exporting GMP numbers to Python
In-Reply-To: <dmhdha$lv2$1@sea.gmane.org>
References: <dmhdha$lv2$1@sea.gmane.org>
Message-ID: <200511291530.21987.meine@kogs.informatik.uni-hamburg.de>

On Tuesday 29 November 2005 12:20, Jan Langer wrote:
> I found a working GMP Python wrapper in the GMPY module, but I have
> simply no idea how use it for my purpose.

Short answer:
You can either export GMP numbers yourself with Boost::Python, or try to find 
out how to/from-python conversion is done with the wrapper you found.  In the 
latter case, export boost::python wrappers returning/accepting 
boost::python::object's and do the conversion manually.

-- 
Ciao, /  /
     /--/
    /  / ANS


From meine at kogs1.informatik.uni-hamburg.de  Tue Nov 29 15:35:36 2005
From: meine at kogs1.informatik.uni-hamburg.de (Hans Meine)
Date: Tue, 29 Nov 2005 15:35:36 +0100
Subject: [C++-sig] porting Qt 3.3.x to python
In-Reply-To: <20051122170334.15526.qmail@web31504.mail.mud.yahoo.com>
References: <20051122170334.15526.qmail@web31504.mail.mud.yahoo.com>
Message-ID: <200511291535.37200.meine@kogs.informatik.uni-hamburg.de>

On Tuesday 22 November 2005 18:03, Ralf W. Grosse-Kunstleve wrote:
> > Can I tell boost to use qt objects from PyQt instead of generating it's
> > own version of them?
>
> That's probably something you'd have to figure out yourself. A few months
> ago someone asked a similar question regarding SWIG. It was surprisingly
> easy to support passing SWIG-wrapped objects to Boost.Python wrapped
> functions:
>
> http://cvs.sourceforge.net/viewcvs.py/cctbx/boost_adaptbx/swig_arg.hpp?view
>=markup
>
> http://cvs.sourceforge.net/viewcvs.py/cctbx/boost_adaptbx/swig_args_ext.cpp
>?view=markup
>
> If you are lucky, maybe something similar works for SIP. You'd have to find
> out how the SIP wrapping mechanism works.

Yes, it is possible; for example I did it with code fragments like these (this 
is also a hint for Jan Langer how such a thing could look like):

#include <sipvigraqtDeclvigraqt.h> // my SIP module was called vigraqt

PyObject *getViewer()
{
    return sipMapCppToSelfSubClass(
        somehowGetViewer(), sipClass_QImageViewer);
}

BOOST_PYTHON_MODULE_INIT(sipbridge)
{
    def("viewer", &getViewer);
}

-- 
Ciao, /  /
     /--/
    /  / ANS


From dave at boost-consulting.com  Wed Nov 30 13:18:20 2005
From: dave at boost-consulting.com (David Abrahams)
Date: Wed, 30 Nov 2005 07:18:20 -0500
Subject: [C++-sig] [Boost.Python] exporting GMP numbers to Python
References: <dmhdha$lv2$1@sea.gmane.org>
	<200511291530.21987.meine@kogs.informatik.uni-hamburg.de>
Message-ID: <uy83672ib.fsf@boost-consulting.com>

Hans Meine <meine at kogs1.informatik.uni-hamburg.de> writes:

> On Tuesday 29 November 2005 12:20, Jan Langer wrote:
>> I found a working GMP Python wrapper in the GMPY module, but I have
>> simply no idea how use it for my purpose.
>
> Short answer:
> You can either export GMP numbers yourself with Boost::Python, or try to find 
> out how to/from-python conversion is done with the wrapper you found.  In the 
> latter case, export boost::python wrappers returning/accepting 
> boost::python::object's and do the conversion manually.

It's also possible to register custom from_/to_python converters that
will automatically convert the GMP numbers using the wrapper from the
other extension.

http://mail.python.org/pipermail/c++-sig/2002-May/001211.html
http://www.boost.org/libs/python/doc/v2/to_python_converter.html

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