`NamedRowTuple` does not support pickling

[ragnard]

Expected behavior

The custom type NamedRowTuple does not implement the methods necessary for a successful pickle roundtrip. When unpickling (pickle.load[s]) it raises a TypeError.

It should work with pickle.

Some frameworks, for example https://github.com/streamlit/streamlit rely on pickling results/rows for caching which currently fails when using this library.

Actual behavior

Exception when calling pickle.load[s]:

 TypeError: NamedRowTuple.__new__() missing 2 required positional arguments: 'names' and 'types'

Steps To Reproduce

import pickle
from trino import types

pickle.loads(pickle.dumps(types.NamedRowTuple(["Alice", 38], ["name", "age"], ["varchar", "integer"])))

Log output

No response

Operating System

Linux

Trino Python client version

caff0e8

Trino Server version

461

Python version

Python 3.12.5

Are you willing to submit PR?

• Yes I am willing to submit a PR!