:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.
Python ab-testingai-analyticsanalyticscdpdata-warehouseexperimentsfeature-flagsjavascriptproduct-analyticspythonreactsession-replaysurveystypescriptweb-analytics
fix(aio): stop distinct_id alias breaking evaluation backfill queries (#121191)
posthog[bot]711a24ef73cb
diff --git a/products/ai_observability/backend/backfill_candidates.py b/products/ai_observability/backend/backfill_candidates.pyindex ecc7b089d6c2..9c2c862b423b 100644--- a/products/ai_observability/backend/backfill_candidates.py+++ b/products/ai_observability/backend/backfill_candidates.py@@ -66,7 +66,7 @@ _UNITS_SQL = """SELECT{unit_key} AS unit_id,min(timestamp) AS unit_timestamp,- argMin(distinct_id, timestamp) AS distinct_id,+ argMin(distinct_id, timestamp) AS unit_distinct_id,argMin(properties.$session_id, timestamp) AS web_session_id,argMin(properties.$ai_trace_id, timestamp) AS unit_trace_idFROM eventsdiff --git a/products/ai_observability/backend/test/test_backfill_candidates.py b/products/ai_observability/backend/test/test_backfill_candidates.pyindex d50690b053f7..b2a35ec3e1b3 100644--- a/products/ai_observability/backend/test/test_backfill_candidates.py+++ b/products/ai_observability/backend/test/test_backfill_candidates.py@@ -160,14 +160,28 @@ class TestBackfillCandidates(ClickhouseTestMixin, APIBaseTest):assert self._count(target="generation", window_end=BASE + timedelta(minutes=1)) == 1assert self._count(target="generation", window_start=BASE + timedelta(hours=1)) == 2- def test_property_condition_filters_units_and_full_rollout_keeps_them_all(self) -> None:- conditions = [- {- "properties": [{"key": "$ai_model", "value": "gpt-4o", "operator": "exact", "type": "event"}],- "rollout_percentage": 100,- }+ @parameterized.expand(+ [+ (+ "event_property",+ {"key": "$ai_model", "value": "gpt-4o", "operator": "exact", "type": "event"},+ ["t1", "t3"],+ ),+ # `distinct_id` is an events column, so a select alias of the same name would capture it in WHERE.+ (+ "distinct_id_column",+ {"key": "distinct_id", "value": "d-t1", "operator": "exact", "type": "person"},+ ["t1"],+ ),]- assert self._count(target="trace", conditions=conditions) == 2+ )+ def test_property_condition_filters_units_and_full_rollout_keeps_them_all(+ self, _case: str, property_filter: dict[str, Any], expected: list[str]+ ) -> None:+ conditions = [{"properties": [property_filter], "rollout_percentage": 100}]+ assert self._count(target="trace", conditions=conditions) == len(expected)+ page = self._fetch(target="trace", conditions=conditions, limit=10)+ assert sorted(candidate.unit_id for candidate in page.candidates) == expected@parameterized.expand([("a tenth", 10), ("two fifths", 40), ("just under the first bucket", 8)])def test_sampling_picks_the_units_the_live_scheduler_would_pick(self, _case: str, rollout: int) -> None: