Skip to content

Release Notes

Version history for Azothedge Pulse. Versions track repository SemVer.

Versioning policy

  • Snapshots are published at minor versions — 0.1, 0.2, 1.0. Patch-level doc fixes update the current minor in place.
  • Critical fixes are backported to the previous minor (N-1) only. Older versions are frozen.
  • The version selector (top of every page) lets you read the docs for any published version; latest always points at the newest.

v0.1 — Ingestion plane

The first public version. Phase 1 — the end-to-end ingestion plane — is complete.

Schema registry

  • Multi-tenant PDM registry: classes, field groups, datatypes, schemas, descriptors.
  • Composition by URI reference ($class, $fieldGroups); strict JSON Schema (Draft 2020-12) validation.
  • Definition lifecycle: draft → published → archived (forward-only); published definitions are immutable; soft delete; archive is blocked while referenced (409 with referenced_by).
  • Endpoints: GET/POST/PATCH/DELETE across /v1/classes, /v1/field-groups, /v1/datatypes, /v1/schemas, /v1/descriptors.

Ingestion & lake

  • POST /v1/collect — PDM-validated event acceptance, published to per-tenant Kafka topics, 202 Accepted.
  • Datasets — POST/GET /v1/datasets[/{slug}] materialize per-tenant Iceberg tables, partitioned by day(timestamp).
  • Async worker — consumes every tenant's raw topic, micro-batches (30s / 10k rows), appends to Iceberg.
  • Exactly-once into the lake — Postgres dedup gate; a re-delivered event never duplicates a row.
  • Dead-letter routing — invalid/unresolvable events go to the tenant's events.dlq.
  • GET /v1/datasets/{slug}/preview — DuckDB query over the lake, RLS-gated.

Platform

  • Tenant isolation — Postgres row-level security with a non-superuser runtime role; per-tenant Kafka topics (pre-created on tenant creation) and Iceberg namespaces.
  • Observability — Prometheus metrics (API + worker), OpenTelemetry traces to Tempo, Grafana dashboards, API /ready and worker /healthz.
  • Admin portal — Next.js UI for tenants, schemas/classes/field-groups (with the publish + "create dataset from schema" flow), and dataset preview.

Documentation

  • This documentation site: a structured information architecture, built-in search, dark mode, Mermaid diagrams, a Redoc API reference generated from the live OpenAPI 3.1 spec, and a single-page PDF export.

See the Roadmap for what's coming next (identity, profiles, segments, journeys).