stats.ApproximateUniqueCombineFn should handle numpy array
Open
bug
core
P3
python
- Dominant language
- Java
- Stars
- 8.7k
- Forks
- 4.7k
- Avg merge
- 1d 20h
- Merged PRs (30d)
- 196
Description
stats.ApproximateUniqueCombineFn uses DeterministicFastPrimitivesCoder to encode input data. DeterministicFastPrimitivesCoder only accepts python data types and iterables, so numpy data types throw an error.
Edit: for more detail, with numpy data, the coder registry assigns PickleCoder which returns DeterministicFastPrimitivesCoder when calling PickleCoder.as_deterministic_coder
Imported from Jira [BEAM-10900](https://issues.apache.org/jira/browse/BEAM-10900). Original Jira may contain additional context.
Reported by: monicadsong.
Contributor guide
Assessment
This issue has not been assessed yet.