[cctbxbb] boost python exports

[David Waterman]

Ah yes, I have to remember to actually import flex in python for the
get_data() accessor to work.

foo.get_indices() is a different issue. I should be able to copy-and-paste
whatever I need from somewhere else a scitbx::af::shared< std::size_t > is
returned.

Cheers
-- David


On Wed, 2 Mar 2022 at 14:27, David Waterman <dgwaterman at gmail.com> wrote:

> Hi folks,
>
> I'm trying to create a boost python class that works with either vec3 or
> mat3 arrays (and I only need those two types). I made a templated class and
> then exported it twice with different names, for the vec3 and mat3
> versions. For example, for mat3:
>
>     class_<ReconstituteDerivatives< mat3<double> >
> >("ReconstituteDerivativesMat3", no_init)
>       .def("get_data", &ReconstituteDerivatives< mat3<double> >::get_data)
>       .def("get_indices", &ReconstituteDerivatives< std::size_t
> >::get_indices);
>
> I can create one of these objects from Python
>
> >>> foo
> <dials_refinement_helpers_ext.ReconstituteDerivativesMat3 object at
> 0x7f3ffa421ac0>
>
> but the accessors are not exported properly:
>
> >>> foo.get_data()
> Traceback (most recent call last):
>   File "<stdin>", line 1, in <module>
> TypeError: No to_python (by-value) converter found for C++ type:
> scitbx::af::shared<scitbx::mat3<double> >
>
> I expected a scitbx_array_family_flex_ext.mat3_double here
>
> >>> foo.get_indices()
> Traceback (most recent call last):
>   File "<stdin>", line 1, in <module>
> Boost.Python.ArgumentError: Python argument types in
>     ReconstituteDerivativesMat3.get_indices(ReconstituteDerivativesMat3)
> did not match C++ signature:
>     get_indices(dials::refinement::ReconstituteDerivatives<unsigned long>
> {lvalue})
>
> and I expected a scitbx_array_family_flex_ext.size_t here.
>
> Is it likely there is something simple I am missing, like a #include? I do
> include flex_wrapper.h. Or perhaps I am just doing something wrong with
> templated classes and boost python. Looking at other code isn't helping to
> reveal the mysteries to me yet.
>
> Cheers
> -- David
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/cctbxbb/attachments/20220302/5ecae03f/attachment.htm>