Is the example of Object ownership wrong?

[LiuYinCarl]

I am testing the example of section Object ownership A problematic example

#include <nanobind/nanobind.h>

namespace nb = nanobind;

struct Data{};
Data data;
Data *get_data() { return &data; }

NB_MODULE(my_ext, m) {
  nb::class_<Data>(m, "Data");
  m.def("get_data", &get_data());
}

but compile failed.

cmake -S . -B build && cmake --build build/
-- Configuring done
-- Generating done
-- Build files have been written to: /home/lzh/test/nanobind/build
Consolidate compiler generated dependencies of target nanobind-static
[ 83%] Built target nanobind-static
Consolidate compiler generated dependencies of target my_ext
[ 91%] Building CXX object CMakeFiles/my_ext.dir/my_ext.cpp.o
/home/lzh/test/nanobind/my_ext.cpp: In function ‘void nanobind_init_my_ext(nanobind::module_&)’:
/home/lzh/test/nanobind/my_ext.cpp:12:30: error: lvalue required as unary ‘&’ operand
   12 |   m.def("get_data", &get_data());
      |                      ~~~~~~~~^~
gmake[2]: *** [CMakeFiles/my_ext.dir/build.make:76: CMakeFiles/my_ext.dir/my_ext.cpp.o] Error 1
gmake[1]: *** [CMakeFiles/Makefile2:84: CMakeFiles/my_ext.dir/all] Error 2
gmake: *** [Makefile:91: all] Error 2

So I think the example is error, then I edit the example then compile success.

#include <nanobind/nanobind.h>

namespace nb = nanobind;

struct Data{};
Data data;
Data *get_data() { return &data; }


NB_MODULE(my_ext, m) {
  nb::class_<Data>(m, "Data");
  m.def("get_data", &get_data); // remove ()
}

then test

lzh@liuzuhua:~/test/nanobind/build$ python3
Python 3.10.6 (main, Mar 10 2023, 10:55:28) [GCC 11.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import my_ext
>>> import gc
>>> a = my_ext.get_data()
>>> a
<my_ext.Data object at 0x7fe8bbddd330>
>>> del a
>>> gc.collect()
free(): invalid pointer
Aborted

[wjakob]

Good catch, thanks. It's fixed in f197571.