NVIDIA / NVIDIA/cudf

[BUG] Without providing metadata cannot round-trip struct columns via interop/from/to_arrow

Open
#17,061 1 comment 0 reactions 1 assignee Claimed by @vyasr View on GitHub
bug pylibcudf Python
Dominant language
C++
Stars
9.8k
Forks
1.1k
Avg merge
3d 6m
Merged PRs (30d)
278

Description

**Describe the bug**

```python
import pyarrow as pa
import pylibcudf as plc

host_c = pa.array([{"a": [1, 2], "b": [3, 4]}])
device_c = plc.interop.from_arrow(host_c)
on_host = plc.interop.to_arrow(device_c)

# > RuntimeError: CUDF failure at: /home/coder/cudf/cpp/src/interop/to_arrow_schema.cpp:146: Number of field names and number of children doesn't match
```

We either need to construct appropriate (empty) metadata for child columns, or handle it in to_arrow_schema on the C++ side.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.