David Hewitt
David Hewitt
We took the hard-coded value to match the original `serde` behaviour, however there's no reason we couldn't make this configurable. A PR in `jiter` would be welcome; I suppose there...
Note to self: this might need to be opt-out (for a few versions) and be mentioned in the migration guide.
I think it's sorta fine to leave the type objects in statics for now given we don't support subinterpreters, but yes: - we should seek to use module state if...
I agree, I think there should be ways to take inspiration from how `pybind11` makes that possible, with the caveat that Rust's lack of stable ABI makes things more tricky....
Yep `PyModule_ExecDef` our definition is wrong, but we also need `PyModule_FromDefAndSpec2` which I couldn't see in the pypy exports.
Thanks, that's perfect, I will seek to retest with pypy nightly. > I will point out that cython does not use these low-level interfaces, rather it uses `PyModuleDef_Init`. Indeed, we...
Superseded by #5525
Superseded by #5592
We could probably avoid copying from Python to Rust by using [`Bytes::from_owner`](https://docs.rs/bytes/latest/bytes/struct.Bytes.html#method.from_owner) (maybe with `PyBackedBytes` as the owner) but going from Rust to Python would almost certainly require a copy....
... so while I think this is probably a useful thing to have, it might be that it's a future useful thing to have?