Wenzel Jakob | 1e1f367 | 2016-04-07 10:31:01 +0200 | [diff] [blame] | 1 | Limitations |
| 2 | ########### |
| 3 | |
| 4 | pybind11 strives to be a general solution to binding generation, but it also has |
| 5 | certain limitations: |
| 6 | |
| 7 | - pybind11 casts away ``const``-ness in function arguments and return values. |
| 8 | This is in line with the Python language, which has no concept of ``const`` |
| 9 | values. This means that some additional care is needed to avoid bugs that |
| 10 | would be caught by the type checker in a traditional C++ program. |
| 11 | |
Wenzel Jakob | 4626b5a | 2016-06-30 14:45:57 +0200 | [diff] [blame] | 12 | - The NumPy interface ``pybind11::array`` greatly simplifies accessing |
| 13 | numerical data from C++ (and vice versa), but it's not a full-blown array |
| 14 | class like ``Eigen::Array`` or ``boost.multi_array``. |
| 15 | |
Wenzel Jakob | 8e5dceb | 2016-09-11 20:00:40 +0900 | [diff] [blame] | 16 | These features could be implemented but would lead to a significant increase in |
| 17 | complexity. I've decided to draw the line here to keep this project simple and |
| 18 | compact. Users who absolutely require these features are encouraged to fork |
| 19 | pybind11. |
Wenzel Jakob | 1e1f367 | 2016-04-07 10:31:01 +0200 | [diff] [blame] | 20 | |