On Friday, July 29, 2011 05:48:32 AM Johan Råde wrote:

When I run the following code

C++:

#include #include

using boost::noncopyable; using namespace boost::python;

class X : noncopyable {};

PyObject* f() { return api::object(new X).ptr(); }

BOOST_PYTHON_MODULE(Foo) { class_("X"); def("f", &f); }

Python:

import Foo Foo.f()

then an exception, with the following error string, is thrown from the api::object constructor:

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

Am I using api::object incorrectly, or should I not use api::object at all but do something completely different?

As far as I can tell, the above should work. Could you please file a ticket on trac? Ravi

On 12/3/2011 1:13 PM, Johan Råde wrote:

On 8/8/2011 6:21 AM, Ravi wrote:

...

On Friday, July 29, 2011 05:48:32 AM Johan Råde wrote: ...

...

Am I using api::object incorrectly, or should I not use api::object at all but do something completely different?

As far as I can tell, the above should work. Could you please file a ticket on trac?

Ravi

Now I have finally done that: https://svn.boost.org/trac/boost/ticket/6203

--Johan

It turned out not to be a bug. The python::object constructor does not take a raw pointer, you must give it a shared pointer. See https://svn.boost.org/trac/boost/ticket/6203

I noticed the below crashes[run time error because of space in string] Is there a possibility to handle this situation? [without explicitly calling split]

  string s = "   12252001";   int offsets[] = {5,2,4};   offset_separator f(offsets, offsets+3);   tokenizer tok(s,f);   for(tokenizer::iterator beg=tok.begin(); beg!=tok.end();++beg)           int val = boost::lexical_cast<int>( *beg);

You should explicitly trim every token (use boost/algorithm/string/trim.hpp). Alternatively, you could use more flexible facility, like Spirit.