feat(m1): wire run events into commands (Task 2/2) by JesuFemi-O · Pull Request #137 · Database-Tycoon/tycoon-cli

JesuFemi-O · 2026-07-09T17:01:48Z

What this PR does

This is the second and final PR for M1. It wires the event infrastructure from Task 1/2 (PR #136) into the four live command paths.

Closes #95, #97, #101, #103.

Issue breakdown

#95 — Emit RunStarted / RunCompleted / RunFailed from `run_source()`

runner.py now emits three lifecycle events around every dlt dispatch path:

RunStarted before dispatch (best-effort; never blocks the run)
RunCompleted after a successful load with load_id, duration_seconds, and rows_loaded extracted from load_info.load_packages
RunFailed in the outer except block so any pipeline exception is captured

Two module-level helpers added:

_emit_event_safe(metadata_db, event) — opens DuckDBFileBackend, appends the event, swallows all exceptions
_build_run_completed(name, load_info, elapsed) — extracts user-table row counts from dlt load_packages (filters _dlt_* internal tables)

The metadata DB path is resolved once via config.root inside a try/except, so it degrades silently in test environments where config.root is not available.

#97 — Emit DbtRunCompleted from `_capture_dbt_and_refresh_safe()`

After each dbt run / build / test, transform.py reads <dbt_project_dir>/target/run_results.json and appends a DbtRunCompleted event with:

command, target, models_run, models_passed, models_errored, duration_seconds

Wrapped in its own try/except so a missing results file never fails the build.

#101 — Replace raw-DB queries in `status.py` with events backend

import duckdb removed from the top of the file. It now appears only inside _query_run_counts() and _query_layer_last_build(), which still read from the old observability schema (needed for backward compat with the Runs column).
_sources_from_backend(metadata_db) replaces _query_last_sync() and _query_row_counts(). It opens DuckDBFileBackend read-only, queries event_type = 'run_completed', and returns {source_id: {last_sync, rows}}. The Sources panel now reads freshness and row counts from the events table.

_render_sources_panel() signature changes: raw_db: Path → metadata_db: Path.

#103 — Rewrite `history.py` using HistoryRepository; add `core/history.py`

src/tycoon/core/history.py (new file):

RunSummary — frozen dataclass: run_id, source_id, runtime_id, status, started_at, duration_seconds, rows_total, command
RunDetail — frozen dataclass: summary, rows_by_table, tables_created, error
HistoryRepository — list_runs(limit) returns sorted list[RunSummary]; get_run(prefix) returns RunDetail | None (None on zero or ambiguous matches)

src/tycoon/commands/history.py — full rewrite:

import duckdb gone from the top level
All raw SQL data-access functions removed: _load_dlt_rows, _load_dbt_rows, _resolve_id, _fmt_bytes, _show_dlt_run, _show_dbt_run
_render_history_table(runs: list[RunSummary]) — dbt rows use command (short_id) as Ref; dlt rows use source_id/short_id
_list_history() — uses HistoryRepository; --layer validates manifest and filters to dbt runs (per-model filtering deferred: DbtRunCompleted does not carry per-node detail yet)
_show_run() — uses HistoryRepository.get_run(); shows rows-by-table for RunCompleted, error for RunFailed

--source filter semantic change: events store source_id = config_key (e.g. "pokeapi", not "raw_pokeapi"), so filtering is direct with no schema translation.

--layer M1 limitation: validates the layer and requires a compiled manifest, but shows all dbt runs rather than filtering to only those that touched the given layer. Per-node detail will be added to DbtRunCompleted in a later milestone.

Tests

tests/test_history.py migrated from _seed_metadata() (old schema) to _seed_events() (writes via DuckDBFileBackend). _seed_metadata() kept for TestStatusRunsColumn, which validates the _query_run_counts() → old dlt_runs path that is intentionally preserved.

Updated/replaced tests:

test_show_dlt_surfaces_trace_details_when_present → test_show_dlt_surfaces_duration (events carry duration_seconds; trace bytes not in M1)
test_show_dbt_surfaces_schema_changes_when_present → test_show_dbt_invocation_exits_zero
test_layer_filter_restricts_to_invocations_touching_layer — updated to reflect M1 layer-filter behavior
test_source_resolves_config_name_to_schema — assertion updated from "raw_pokeapi" to "pokeapi"

82 tests pass across test_events.py, test_metadata_contract.py, test_history.py, and test_cli_surface.py. One pre-existing unrelated failure (test_nao_uninitialised_warn_points_at_register_llm) excluded.

Stacking note

This PR targets feat/m1-metadata-backend-protocol. Once PR #136 merges to main, this PR will be retargeted to main.

^{Need help on this PR? Tag /codesmith with what you need. Autofix is disabled.}

Completes M1 Task 2/2. Four issues shipped in one commit: #95 — runner.py emits RunStarted before dispatch, RunCompleted after success, RunFailed on any exception. Helpers: _emit_event_safe (wraps DuckDBFileBackend, silently no-ops) and _build_run_completed (extracts rows from dlt load_info.load_packages). #97 — transform.py reads target/run_results.json after each dbt run/build/test and appends a DbtRunCompleted event. Wrapped in its own try/except so a missing results file never fails the build. #101 — status.py: import duckdb removed from top level; moved inside _query_run_counts() and _query_layer_last_build(). New _sources_from_backend(metadata_db) reads RunCompleted events to populate Last Sync and row counts in the Sources panel. _render_sources_panel() signature changes raw_db → metadata_db. #103 — history.py rewritten to use HistoryRepository (new core/history.py). No raw SQL, no top-level import duckdb. List view renders list[RunSummary]; show view renders RunDetail. _resolve_source_id replaces _resolve_source_schema — source_id in events IS the config key, no schema translation needed. --layer validates the manifest and filters to dbt runs; per-model filtering deferred to a later milestone. core/history.py: new module — RunSummary, RunDetail, HistoryRepository. list_runs() sorts all events newest-first; get_run() resolves by prefix across RunCompleted/RunFailed/DbtRunCompleted. tests/test_history.py: all history tests migrated to _seed_events() via DuckDBFileBackend. _seed_metadata() kept for TestStatusRunsColumn which still reads dlt_runs via _query_run_counts(). Tests that tested old trace/schema-change detail replaced with M1-appropriate equivalents. 32 new tests + 22 updated tests. All pass (1 pre-existing unrelated failure in test_nao_uninitialised_warn_points_at_register_llm excluded).

gemini-code-assist

Code Review

This pull request refactors the run history and status commands to utilize a new event-driven metadata backend and repository pattern, replacing direct DuckDB queries. The code review highlights a critical filter-after-limit bug in the history command that can result in empty lists when filtering. Additionally, several feedback items suggest robustness improvements to safely handle potentially missing or null attributes in events and JSON results, preventing potential runtime errors and type mismatches.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

…nsive guards Fix filter-after-limit bug (comment #3553429137/#3553429151): list_runs() now accepts limit=None and returns the full sorted list; _list_history() passes limit=None and slices after all filters are applied. Eliminates the case where tool/source/layer filtering on a pre-truncated list returns an empty result when there are matching runs beyond limit*3. Defensive or-guards (comments #3553429151/#3553429156/#3553429163/#3553429171): - core/history.py: e.rows_loaded or {} in list_runs() and get_run() - core/history.py: event.rows_loaded or {}, event.tables_created or [] in RunDetail construction - status.py: dict(e.rows_loaded or {}) in _sources_from_backend() Safer run_results.json extraction (comment #3553429176): - transform.py: float(... or 0.0), run_results.get('args') or {}, run_results.get('results') or [] prevents TypeError if any field is None rather than absent

…mpleted _build_run_completed was iterating pkg.jobs as a list, but dlt's LoadPackageInfo.jobs is a dict[TPackageJobState, list[LoadJobInfo]]. Even with that fixed, LoadJobInfo has no rows_count field, so counts would always be zero. Switch to pipeline.last_trace.last_normalize_info.row_counts which gives exact per-table item counts directly from dlt's normalize step.

JesuFemi-O · 2026-07-09T18:33:58Z

How to test this branch locally

1. Pull and install

git fetch origin feat/m1-wire-run-events
git checkout feat/m1-wire-run-events
uv sync

2. Create a test project

mkdir /tmp/chess-test && cd /tmp/chess-test

cat > tycoon.yml << 'YAML'
name: chess-test
stack:
  ingestion: dlt
  transformation: none

sources:
  chess:
    type: rest_api
    schema: raw_chess
    config:
      client:
        base_url: https://api.chess.com/pub/
      resources:
        - name: player_profile
          endpoint:
            path: player/hikaru
        - name: player_stats
          endpoint:
            path: player/hikaru/stats
YAML

No API key needed — chess.com is a public API.

3. Run a source

alias tycoon=/path/to/tycoon-cli/.venv/bin/tycoon   # adjust to your checkout path

tycoon data sources run chess

Expected: dlt loads two endpoints into .tycoon/raw_chess.duckdb and writes a RunCompleted event to .tycoon/metadata.duckdb.

4. Check history

tycoon data history

Expected: one row — chess/<8-char-id> ✓ 2 rows N.Ns

tycoon data history show <8-char-prefix>   # copy from Ref column

Expected: drilldown showing player_profile: 1, player_stats: 1.

5. Run it a second time to see multiple rows

tycoon data sources run chess
tycoon data history

Expected: two ✓ rows in the list, newest first.

6. Check status

tycoon data status

Expected: Sources panel shows chess row with last sync timestamp and row counts pulled from the events backend.

dbt testing (requires an existing dbt project wired into tycoon.yml):

After a tycoon data transform run or tycoon data transform build, a DbtRunCompleted event is written. Then:

tycoon data history --tool dbt
tycoon data history show <8-char-event-id>

The drilldown won't show per-table rows (that's an M1 limitation — DbtRunCompleted doesn't carry per-node detail yet), but status, duration, and model counts are captured.

The filesystem read_csv() transformer names its output table _read_csv, which was being dropped by the startswith('_') filter. Switch to an explicit denylist of known dlt internal tables (_dlt_pipeline_state, _dlt_loads, _dlt_version) so user tables with _ prefixes are preserved.

db-tycoon-stephen

Good shape overall. I verified all six Gemini findings are fixed in the current diff (filter-after-limit, limit=None support, and the four defensive defaults), CI is green, and the event emission is properly fail-safe on every path — RunStarted best-effort, RunFailed in the outer except with re-raise, everything through _emit_event_safe. That fully addresses the single-writer degradation concern carried over from the #136 review.

One substantive behavior change that deserves a deliberate decision (Sources panel row semantics), one stale help text, and two smaller cleanups — inline below.

Small observations, no action needed:

_DLT_INTERNAL_TABLES is an exact-match set of three names; a t.startswith("_dlt") prefix filter would match the PR description's stated intent ("filters _dlt_*") and survive new dlt internals.
The #136 read-only-guard deferral rationale ("Task 2 only opens the backend for writes") turned out not to hold — this PR opens read-only in both status and history. No harm since both call sites guard for missing files, but it's a good sign that guard eventually wants to live in the backend itself.
Stacking is handled correctly; just remember the rebase after #136's uv.lock restore lands.

- Use metadata_db_path() in runner.py instead of hand-building the path - Switch _dlt internal table filter to startswith('_dlt') prefix so new dlt internals are covered without updating an explicit denylist - Narrow bare except on read-side history/show to emit a dim warning instead of silently swallowing genuine backend errors - Rename status Sources panel column to 'Last Sync Rows' — the events backend returns rows moved in the most recent sync, not current table totals; the new name makes that semantic explicit - Update --layer help text to state the M1 limitation (all dbt runs shown; per-model filtering comes in a later milestone) - Rename test_layer_filter_restricts_to_invocations_touching_layer to test_layer_filter_shows_all_dbt_runs_m1_limitation so it doesn't imply layer filtering is actually implemented

db-tycoon-stephen · 2026-07-11T09:15:37Z

/gemini review

gemini-code-assist

Code Review

This pull request refactors the run history and status tracking systems to use a new event-based metadata backend instead of querying DuckDB tables directly. It updates the ingestion runner, transform commands, and CLI commands to emit and query structured events, and migrates the test suite to use this new event-seeding mechanism. The review feedback highlights three key improvements: explicitly excluding 'skipped' dbt statuses from being counted as errors, validating and rejecting conflicting CLI filter combinations upfront to prevent confusing empty states, and refining prefix resolution to distinguish between non-existent and ambiguous run IDs.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

- Escape raw exception strings with rich.markup.escape before interpolating into dim warning messages — OSError messages like '[Errno 13] Permission denied' would otherwise be parsed as Rich markup tags and raise MarkupError on the exact error path we just made visible - Add 'skipped' to dbt non-error statuses so skipped models don't inflate models_errored and incorrectly mark a run as failed - get_run() now raises ValueError for ambiguous prefixes (listing the matching run IDs) instead of silently returning None — CLI surfaces this as an actionable error message distinct from 'not found' - Reject --tool dlt + --layer and --tool dbt + --source upfront with a clear error instead of silently returning an empty list

JesuFemi-O · 2026-07-15T10:08:06Z

/gemini review

gemini-code-assist

Code Review

This pull request refactors the history and status commands to use a new event-driven metadata backend and repository pattern instead of direct DuckDB queries, updating the ingestion runner and transform commands to emit structured run events. The review feedback highlights several key improvements: safely wrapping the construction and emission of RunCompleted events in the ingestion runner to ensure observability code never fails a successful pipeline run, raising explicit errors instead of swallowing database exceptions in the history commands to avoid misleading user feedback, and specifying UTF-8 encoding when reading dbt run results to prevent platform-specific decoding issues.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

- transform: record started_at before subprocess; skip stale run_results.json (dbt crash leaves previous file on disk — mtime check prevents false event) - runner: call raise_on_failed_jobs() on legacy and native/dynamic paths so partial dlt job failures flip to RunFailed instead of silently succeeding - status: derive Runs column from events backend instead of legacy dlt_runs table; hint gate changed to check metadata DB existence; fix connection leaks in _query_run_counts and _query_layer_last_build (context managers) - history: escape source_id, run_id, command, and error in _show_run drilldown to prevent Rich markup injection from user-controlled values - core/history: guard get_run against empty prefix (startswith("") matches all)

JesuFemi-O · 2026-07-15T10:41:48Z

Second round of fixes in cf6f190, addressing 7 bugs surfaced by an independent code review pass:

Critical

Stale run_results.json on dbt crash (transform.py) — _capture_dbt_and_refresh_safe now records started_at = time.time() before the subprocess and skips emitting a DbtRunCompleted event if run_results.json's mtime is older than the subprocess start. Previously, a killed/crashed dbt run would emit a ghost event from the previous run's file.
raise_on_failed_jobs() missing on native and legacy paths (runner.py) — Added after _run_legacy() and after pipeline.run() in the native/dynamic branch. Partial dlt job failures now correctly surface as RunFailed rather than silently emitting RunCompleted.
Runs column broken for new-system-only projects (status.py) — _render_sources_panel was reading run counts from the legacy dlt_runs table via _query_run_counts. Moved that responsibility into _sources_from_backend, which now counts total RunCompleted events per source alongside tracking the most-recent sync. The "Drill in with tycoon data history" hint is now gated on _meta_db.exists() rather than if run_counts.

Significant

4–5. Rich markup injection in _show_run (commands/history.py) — detail.error, s.source_id, s.run_id, and s.command are now all wrapped with escape() before interpolation into Rich console prints. An error message or source name containing [bold] markup would previously crash the renderer.

get_run("") matched every event (core/history.py) — "".startswith("") is always True, so an empty prefix selected every run in the DB and raised a spurious "ambiguous prefix" error. Added an early if not run_id_prefix: return None guard.
Connection leaks in status.py — _query_run_counts and _query_layer_last_build both opened bare duckdb.connect() handles without close() on exception paths. Converted to with duckdb.connect(...) as con: context managers.

All 29 tests in test_history.py + test_metadata_contract.py pass. test_runs_column_reflects_captured_loads was updated to seed RunCompleted events (the new source of truth) instead of legacy dlt_runs rows.

- runner: add _emit_run_completed_safe wrapper so exceptions in _build_run_completed (e.g. unexpected load_info types) never propagate into the outer except block and misclassify a successful run as RunFailed - history: replace silent exception swallowing in list and show paths with error() + Exit(1); a corrupted/locked DB now surfaces a clear message instead of "No runs captured yet" - transform: explicitly open run_results.json with encoding="utf-8" for cross-platform safety

JesuFemi-O · 2026-07-15T10:46:52Z

Addressed all three items from the latest Gemini review in dc5bd7d:

High — _emit_run_completed_safe: Added a combined wrapper that calls _build_run_completed then _emit_event_safe inside its own try/except. All three _emit_event_safe(_metadata_db, _build_run_completed(...)) call sites now use _emit_run_completed_safe(...) instead. Any exception during event construction (unexpected dlt types, Pydantic validation, etc.) is now fully contained and cannot flip a successful run into RunFailed.

Medium — history exception handling: Both _list_history and _show_run now call error() + raise typer.Exit(1) on DB failure instead of swallowing the exception and continuing with empty state. A locked or corrupted metadata.duckdb now surfaces a clear message rather than claiming "No runs captured yet."

Medium — encoding="utf-8": run_results.json is now opened with encoding="utf-8" explicitly.

db-tycoon-stephen · 2026-07-15T10:53:48Z

/gemini review

gemini-code-assist

Code Review

This pull request refactors the history and status commands to use a new event-based metadata backend, migrating direct DuckDB queries to the HistoryRepository and DuckDBFileBackend. It also introduces event emission (RunStarted, RunCompleted, RunFailed, and DbtRunCompleted) during ingestion and transformation processes, and updates the test suite accordingly. Feedback on the changes includes addressing a potential filesystem timestamp resolution issue when checking dbt run results to prevent fast runs from being discarded, and removing the now-unused _query_run_counts helper function in the status command module.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

db-tycoon-stephen · 2026-07-15T12:11:20Z

@JesuFemi-O — one flag on dc5bd7d, which I think crossed in flight with my review replies this morning:

The history read-error change reverses a settled decision. dc5bd7d switched _list_history / _show_run from warn-and-continue to error() + Exit(1) — adopting the Gemini suggestion that was marked won't-fix earlier today, citing the decision from the 2026-07-10/11 review round: a locked metadata DB during an active ingest is an expected transient state, and tycoon data history exiting non-zero on it is worse than a visible warning + empty state. DuckDB file locks are per-process, so "ingest running + user checks history" is a routine collision, not an error condition.

Please revert those two hunks to the d7bf39c behavior (dim warning with the escaped error, then continue). Everything else in cf6f190/dc5bd7d verifies clean — the _emit_run_completed_safe wrapper, utf-8, the escape() additions in the drilldown, the empty-prefix guard, and raise_on_failed_jobs() on all three dispatch paths all look right.

db-tycoon-stephen · 2026-07-15T12:11:23Z

/gemini review

gemini-code-assist

Code Review

This pull request refactors the run history and status commands to use a new event-based metadata backend instead of querying legacy DuckDB tables directly. Key feedback includes ensuring that observability capture runs even when ingestion jobs fail by swapping the execution order of raise_on_failed_jobs() and _capture_and_refresh_safe(). Additionally, it is recommended to add a small tolerance to the dbt run results timestamp check to account for coarse filesystem resolutions, wrap event queries in try-except blocks to gracefully handle missing tables on new or legacy databases, and clean up the unused _query_run_counts helper function.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

…view) Reverts the error()+Exit(1) shape from dc5bd7d back to the d7bf39c behavior. A locked metadata DB during an active ingest is an expected transient state (DuckDB file locks are per-process), so history exiting non-zero on it is worse than a visible dim warning + empty state. This also degrades gracefully when the metadata DB exists but the events table doesn't yet. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

db-tycoon-stephen · 2026-07-15T12:36:40Z

Update: I pushed 7ddf7fa restoring the warn-and-continue behavior on the two history read paths (per the flag above), and merged the base branch in so this PR inherits the uv.lock restore (a41bf0a on #136). @JesuFemi-O no action needed on either — the remaining open threads (mtime buffer, dead _query_run_counts/_seed_metadata, raise_on_failed_jobs ordering, skip-visibility) are being triaged as follow-up issues.

JesuFemi-O · 2026-07-15T15:13:47Z

sounds good @db-tycoon-stephen thanks!

db-tycoon-stephen

All review asks verified landed at 7ddf7fa: _emit_run_completed_safe wrapper, escape() on all Rich interpolations, warn-and-continue restored on history read paths, filter-conflict rejection, ambiguous-prefix handling, utf-8, raise_on_failed_jobs on all three dispatch paths. Tests 84 passed (one known pre-existing failure, fails on main too), CI 7/7 green. Non-blocking items tracked in #146. Approved — retarget to main and merge after #136.

#137)

gemini-code-assist Bot reviewed Jul 9, 2026

View reviewed changes

JesuFemi-O added 3 commits July 9, 2026 18:06

fix(lint): remove unused imports flagged by ruff (F401)

c3ab253

JesuFemi-O requested a review from db-tycoon-stephen July 9, 2026 19:47

db-tycoon-stephen mentioned this pull request Jul 10, 2026

feat(core): MetadataBackend protocol + DuckDBFileBackend [M1 Task 1/2] #136

Merged

3 tasks

db-tycoon-stephen reviewed Jul 10, 2026

View reviewed changes

Comment thread src/tycoon/commands/status.py

Comment thread src/tycoon/commands/history.py

Comment thread src/tycoon/commands/history.py Outdated

Comment thread src/tycoon/ingestion/runner.py Outdated

JesuFemi-O requested a review from db-tycoon-stephen July 10, 2026 23:47

gemini-code-assist Bot reviewed Jul 11, 2026

View reviewed changes

Comment thread src/tycoon/commands/transform.py

Comment thread src/tycoon/commands/history.py

Comment thread src/tycoon/core/history.py Outdated

db-tycoon-stephen requested changes Jul 13, 2026

View reviewed changes

JesuFemi-O requested a review from db-tycoon-stephen July 15, 2026 10:00

gemini-code-assist Bot reviewed Jul 15, 2026

View reviewed changes

Comment thread src/tycoon/commands/transform.py

Comment thread src/tycoon/commands/status.py

gemini-code-assist Bot reviewed Jul 15, 2026

View reviewed changes

Comment thread src/tycoon/ingestion/runner.py

Comment thread src/tycoon/ingestion/runner.py

Comment thread src/tycoon/commands/transform.py

Comment thread src/tycoon/core/history.py

Comment thread src/tycoon/core/history.py

Comment thread src/tycoon/commands/status.py

db-tycoon-stephen and others added 2 commits July 15, 2026 08:35

chore: merge base branch (uv.lock restore) into stacked PR

92c11ba

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

db-tycoon-stephen mentioned this pull request Jul 15, 2026

M1 follow-ups: non-blocking review items from PRs #136/#137 #146

Open

5 tasks

db-tycoon-stephen approved these changes Jul 15, 2026

View reviewed changes

Base automatically changed from feat/m1-metadata-backend-protocol to main July 15, 2026 16:26

JesuFemi-O merged commit 3ebd5e9 into main Jul 15, 2026
7 checks passed

db-tycoon-stephen added a commit that referenced this pull request Jul 15, 2026

chore(v0.1.10): sync release branch with main (M1 metadata backend #136/

bf50752

#137)

This was referenced Jul 21, 2026

Release v0.1.10 — metadata backend + a slimmer install #159

Merged

Properly implement Codesmith (Blacksmith) as the repo's AI PR reviewer #173

Open

db-tycoon-stephen deleted the feat/m1-wire-run-events branch August 2, 2026 21:43

Uh oh!

Conversation

JesuFemi-O commented Jul 9, 2026 • edited by blacksmith-sh Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What this PR does

Issue breakdown

#95 — Emit RunStarted / RunCompleted / RunFailed from run_source()

#97 — Emit DbtRunCompleted from _capture_dbt_and_refresh_safe()

#101 — Replace raw-DB queries in status.py with events backend

#103 — Rewrite history.py using HistoryRepository; add core/history.py

Tests

Stacking note

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

JesuFemi-O commented Jul 9, 2026

How to test this branch locally

Uh oh!

db-tycoon-stephen left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

db-tycoon-stephen commented Jul 11, 2026

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

JesuFemi-O commented Jul 15, 2026

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

JesuFemi-O commented Jul 15, 2026

Uh oh!

JesuFemi-O commented Jul 15, 2026

Uh oh!

db-tycoon-stephen commented Jul 15, 2026

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

Uh oh!

Uh oh!

db-tycoon-stephen commented Jul 15, 2026

Uh oh!

db-tycoon-stephen commented Jul 15, 2026

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

JesuFemi-O commented Jul 9, 2026 •

edited by blacksmith-sh Bot

Loading

#95 — Emit RunStarted / RunCompleted / RunFailed from `run_source()`

#97 — Emit DbtRunCompleted from `_capture_dbt_and_refresh_safe()`

#101 — Replace raw-DB queries in `status.py` with events backend

#103 — Rewrite `history.py` using HistoryRepository; add `core/history.py`