Add CI liveness pipeline for SUQL stack - #48
Open
skyxiath wants to merge 2 commits into
Open
Conversation
skyxiath
force-pushed
the
add-testing-framework
branch
7 times, most recently
from
May 19, 2026 21:49
a6cceeb to
47a62ad
Compare
skyxiath
marked this pull request as ready for review
May 20, 2026 18:30
setup() takes `extras_require`, not `extra_requires`. The misspelled key was silently ignored by setuptools, so `pip install -e .[dev]` never installed the dev group. Also add an `embedding` extras group with `FlagEmbedding>=1.3` and `faiss-cpu>=1.7.4` — both required by faiss_embedding's EmbeddingStore but neither declared anywhere. Kept separate from `[dev]` so installs that only need the embedding server don't drag in spacy 3.6.0's pydantic<2 constraint (which conflicts with current litellm). FlagEmbedding pinned to >=1.3 because earlier 1.2.x patch versions shipped with BGE_M3/trainer.py referencing `Optional` without importing it from typing.
Brings up Postgres + SUQL servers from nothing on each push, ingests a private data sample fetched from a separate fixtures repo via a fine-grained PAT, runs one canned answer() query end-to-end. ~4 min warm cache, ~7 min cold. The point: catch plumbing breaks (server boot, SQL compilation, prompt loading, embedding/free-text servers, LLM proxy) at PR time rather than weeks later during research evals. Layout: .github/workflows/test.yml CI workflow tests/check_alive.py liveness probe (one query, one verdict) tests/ingest.sh applies schema + \COPY tests/start_embedding_server.py embedding-server bootstrap tests/fixtures/schema.sql example events table + roles tests/README.md, SETUP.md user docs Consuming fork must set three secrets (FIXTURES_TOKEN, OPENAI_API_KEY, OPENAI_API_BASE) + one variable (FIXTURES_REPO). Walkthrough in tests/SETUP.md.
skyxiath
force-pushed
the
add-testing-framework
branch
from
May 29, 2026 19:11
47a62ad to
1e9752f
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Brings up Postgres + the SUQL servers from nothing on each push, ingests
a private data sample, runs one canned
answer()query end-to-end,reports pass/fail. ~4 minutes warm cache, ~7 minutes cold.
Catches plumbing breaks (server boot, SQL compilation, prompt loading,
embedding server, free-text server, LLM proxy connectivity) at PR time
rather than later during research evals.
Two commits
setup.py: fix extras_require typo + add faiss-cpu to dev extrasStandalone fix — the misspelled
extra_requireswas silently ignoredby setuptools, so
pip install -e .[dev]never installed the devgroup. Adding
faiss-cpualongsideFlagEmbeddingsince theembedding server needs it.
Add CI liveness pipelineThe actual
tests/directory +.github/workflows/test.yml.Required GitHub Secrets on consuming fork
FIXTURES_TOKEN— fine-grained PAT withContents: Readon aprivate fixtures repo containing the sample CSV
OPENAI_API_KEY,OPENAI_API_BASE— LLM proxy credsRequired GitHub Variable
FIXTURES_REPO—owner/nameof the fixtures repoWalkthrough in
tests/SETUP.md.Scope
This is a smoke test — it catches plumbing failures, not result quality.
Quality belongs in separate research evals.
Note
faiss_embedding.py'suse_fp16should also be gated ontorch.cuda.is_available()to make the embedding server boot onCPU-only runners. That's posted separately as a one-line PR rather than
bundled here.