Release develop → master: ingestor v0.8.0 (RFC-0003 I2 per-ingestion write path)

[LukasWodka]

Promotes develop → master so the ingestor image gets built. The image only exists as a byproduct of a master release: release-image.yml is gated on the PyPI publish, which happens on master (publish-master.yml). Divya's staging test surfaced that the deployed :0.7 ingestor predates #408, and my direct v0.8.0 tag couldn't build because 0.8.0 isn't on PyPI.

What merging this does (automatically)

1. publish-master → tracebloc-ingestor==0.8.0 to PyPI.
2. auto-release-on-master sees the v0.8.0 tag already exists (pushed manually) with no Release → re-dispatches release-image.yml, which now passes the PyPI gate and publishes ghcr.io/tracebloc/ingestor:0.8.0 (+ :0.8, :0), signed.

Why this is SAFE / inert — it changes what zero edges run

• Non-prod edges pin tag: "0.7"; prod pins the 0.7 prodDigest. The moving :0.8/:0 floats don't touch :0.7.
• The chart pin bump to 0.8 (chore(chart): point spawned ingestor at the 0.8 line (D16 write path) — HOLD until v0.8.0 image client#490) is held as draft.
• So the 0.8.0 image sits adopted-by-nobody until we bump the pin after Divya's staging validation. She overrides to :0.8.0 deliberately to test — that's the only thing that runs it. Test-first is preserved (artifact exists, deployment gate stays closed).

Carries (8 commits)

• a831d78 chore(release): bump ingestor to 0.8.0 — release the per-ingestion write path (feat(ingestor): per-ingestion immutable tables ds_<uuid4().hex> behind PER_INGESTION_TABLES (backend#1205) #408) (chore(release): bump ingestor to 0.8.0 — release the #408 per-ingestion write path #419)
• 04026fc ci: arm the code-quality gate — soft-fail off, findings block (backend#1303) (ci: arm the code-quality gate (soft-fail off) + all-files dispatch #417)
• 3903fa4 ci(release): let the release train own the version tag (ci(release): let the release train own the version tag #416)
• 4197868 chore(release): bump version to 0.7.9 (v0.7.8 already tagged — Bugbot on release-train: staging -> master #413) (chore(release): bump version to 0.7.9 #414)
• 07ff47f chore: clear the ruff backlog (E4,E7,E9,F) (chore: clear the ruff backlog (E4,E7,E9,F) #412)
• 44ebcb4 Add pre-commit hooks (Layer 0 local lint guard) (Add pre-commit hooks (Layer 0 local lint guard) #411)
• ecf8aef docs: add .cursor/BUGBOT.md — per-repo Bugbot review context (docs: add .cursor/BUGBOT.md — per-repo Bugbot review context #410)
• 3f84eda ci: add code-quality caller workflow (advisory) (ci: code-quality caller (advisory) #409)

Notes

• fr-gate will block until items are Ready-for-prod — conscious skip-fr-gate is defensible here (the release is inert).
• The stale Sync develop → master for v0.7.8 release #400 (sync for v0.7.8) can be closed — v0.7.8 already released.

Epic: tracebloc/backend#1151 · I2: #408 · image: #419

🤖 Generated with Claude Code

---

Note

Medium Risk
Release pipeline behavior changes (who tags, when images build) are operationally critical but explicitly gated on PyPI; ingest/runtime logic changes in this diff are limited to lint/exception-handling hygiene.

Overview
Release & CI: __version__ moves to 0.8.0. auto-release-on-master.yml is removed — v* tags come from the org release train, not github-actions[bot]. release-image.yml adds a verify-published job (poll PyPI, strict vX.Y.Z ref validation, minimal permissions) before build/sign, and concurrency is keyed on the tag so manual re-drives don’t race tag-triggered runs. RELEASING.md and .cursor/BUGBOT.md document the new flow and review rules.

Quality gates: New code-quality-caller.yml (org workflow, soft-fail: false), .pre-commit-config.yaml, ruff.toml, Black 26.x in dev deps, plus widespread unused-import / lint fixes in tests and small edits in data_validator (except Exception), json_ingestor import order, and validators/base typing.

^{Reviewed by Cursor Bugbot for commit a831d78. Bugbot is set up for automated code reviews on this repo. Configure here.}

[LukasWodka]

👋 Heads-up — Code review queue is at 40 / 30

Above the WIP limit. The team convention is to review existing PRs before opening new work.

Open PRs currently in Code review (oldest first):

• averaging-service#222 — fix(sklearn): unwrap Pipeline to terminal estimator before merge dispatch (docs(yaml): time_series_forecasting timestamp must be TIMESTAMP not VARCHAR (#200) #209) · author: @divyasinghds · no reviewer assigned
• averaging-service#225 — fix(averaging): fail fast + dead-letter on permission-denied weight write (fix(base): concurrent-ingest table-name lock (backend #772 P2) #221) · author: @divyasinghds · no reviewer assigned
• averaging-service#245 — chore(ci): record gitleaks history baseline (triaged false positives) · author: @LukasWodka · no reviewer assigned
• averaging-service#246 — feat(chart): ServiceAccount + IRSA + blobStore env for the S3 backend (#1162) · author: @saadqbal · reviewer: @aptracebloc
• averaging-service#247 — fix(averaging): classify S3 access-denied weight writes as permanent (Non-deterministic data_id causes row duplication on K8s Job retry (from backend#772 P0.3) #225 follow-up) · author: @divyasinghds · reviewer: @saadqbal
• backend#1201 — docs(rfc): Data spaces & federated dataset composition (draft) · author: @LukasWodka · reviewer: @saadqbal
• backend#1217 — docs(rfc): Dataset integrity & score reproducibility (draft) · author: @LukasWodka · reviewer: @saadqbal
• backend#1261 — RFC-BACKEND-0008: board & process reset (decision RFC — for markup) · author: @LukasWodka · reviewer: @saadqbal
• backend#1284 — docs(deploy): backend deploy runbook skeleton (#1263) · author: @LukasWodka · no reviewer assigned
• backend#1293 — fix(heartbeat): tolerate unknown experiment_key in training_info (#1287) · author: @divyasinghds · no reviewer assigned

Pull from review before opening new work. (This is a nudge from the kanban WIP check, not a block.)