Add const T to docstring generation. (#3020)

* Add const T to docstring generation.

* Change order.

* See if existing test triggers for a const type.

* Add tests.

* Fix test.

* Remove experiment.

* Reformat.

* More tests, checks run.

* Adding `test_fmt_desc_` prefix to new test functions.

* Using pytest.mark.parametrize to 1. condense test; 2. exercise all functions even if one fails; 3. be less platform-specific (e.g. C++ float is not necessarily float32).

Co-authored-by: Ralf W. Grosse-Kunstleve <rwgk@google.com>
This commit is contained in:
Jack S. Hale
2021-06-08 20:56:45 +02:00
committed by GitHub
parent e25b1505db
commit 4c7697dbe9
3 changed files with 26 additions and 2 deletions

View File

@@ -437,4 +437,10 @@ TEST_SUBMODULE(numpy_array, sm) {
sm.def("accept_double_f_style_forcecast_noconvert",
[](py::array_t<double, py::array::forcecast | py::array::f_style>) {},
"a"_a.noconvert());
// Check that types returns correct npy format descriptor
sm.def("test_fmt_desc_float", [](py::array_t<float>) {});
sm.def("test_fmt_desc_double", [](py::array_t<double>) {});
sm.def("test_fmt_desc_const_float", [](py::array_t<const float>) {});
sm.def("test_fmt_desc_const_double", [](py::array_t<const double>) {});
}