Andreas Klöckner
Andreas Klöckner
> I would love to get rid of hacks such as: Whoa. I had missed that during review. That is pretty gross.
Add a test like https://github.com/kaushikcfd/pytato/blob/1d315637ee237631b30a24228375c33257ea7dbe/test/test_pytato.py#L307-L330
I need to do some measuring.
I think the most reasonable thing to do would be to implement this via pybind, cf. https://github.com/pybind/pybind11/issues/945. Pybind has all the information it needs, it just needs to make it...
> I'm a bit stuck on how `islpy` is merging the `.inc` files into the final wrapper? They just get included, e.g. here: https://github.com/inducer/islpy/blob/33cdc99021b1eed9be0174911202601d84ebff82/src/wrapper/wrap_isl_part3.cpp#L5 > I did a quick implementation...
Alternatively, is it possible to set `.__annotations__` on the methods from the wrapper? (I haven't tried, but `cls.attr("methodname").attr("__annotations__") = ...` might work?)
pybind has `py::object` that wraps and makes C++ handling of `PyObject`s easier, e.g. `py::make_tuple("a", "b", "c")`.
This seems to fail most of the CI... request review when you think it's ready.
Hmm. So my whole reason for starting pymbolic (and not using sympy) was because sympy made 'too many assumptions' and 'did too much stuff' to my expressions behind my back....
Could you use `flattened_sum` and `flattened_product` instead of the constructor? Those *do* simplify, and I'd prefer to keep the constructors non-simplifying.