[smart_holder] Change throw_if_uninitialized_or_disowned_holder() to also show C++ type name.
#4977
Conversation
|
@karagog could you please help reviewing this PR? |
| if (!holder().is_populated) { | ||
| pybind11_fail("Missing value for wrapped C++ type:" | ||
| " Python instance is uninitialized."); | ||
| throw value_error(missing_value_msg + clean_type_id(typeid_name) |
There was a problem hiding this comment.
I dont' suppose you can use f-strings?
There was a problem hiding this comment.
Interesting idea, but no. We don't have f-strings in C++ (I never thought about that before, not sure how feasible that is), and compared to relying on std::string operator+, the overhead of using the Python C API is not something I'd want to have in an error handling code path like here.
| @@ -9,23 +9,19 @@ | |||
|
|
|||
| @pytest.mark.xfail("env.PYPY", reason="gc after `del field` is apparently deferred") | |||
| @pytest.mark.parametrize("m_attr", ["m_valu_readonly", "m_valu_readwrite"]) | |||
| def test_valu_getter(msg, m_attr): | |||
| def test_valu_getter(m_attr): | |||
There was a problem hiding this comment.
nit: this function name is apparently misspelled
There was a problem hiding this comment.
It's intentional, see comment at top of this file, which points here:
pybind11/tests/test_class_sh_property.cpp
Lines 1 to 4 in feed8b1
(The original idea was actually to make this code in test_class_sh_basic.cpp more readable.)
|
Thanks for the review! |
rwgk
deleted the
sh_throw_if_uninitialized_or_disowned_holder_better_error
branch
Description
To aid debugging.
Suggested changelog entry: