Skip to content

CHANGELOG.md

Latest commit

 

History

History
48 lines (39 loc) · 7.6 KB

CHANGELOG.md

File metadata and controls

48 lines (39 loc) · 7.6 KB
Raw

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]

Added

  • Trivy vulnerability scanning for Docker images (api, app, processing) in CI; scans run on Dockerfile changes and weekly against published ghcr.io images, with SARIF reports uploaded to GitHub Code Scanning and the workflow failing on CRITICAL findings. Closes #372.
  • Dependabot configuration in .github/dependabot.yml covering all five package.json locations (root, api, app, xstreamroll-sdk, xstreamroll-processing), with grouped updates (@nestjs/*, @opentelemetry/*, React/Next, @radix-ui/*), weekly schedules, and auto-merge for patch and security updates via .github/workflows/dependabot-auto-merge.yml. Closes #368.
  • .trivyignore file at the repo root documenting the suppression format and serving as the canonical place for triaged false-positive findings.
  • @xstreamroll/contract-tests workspace package (tests/contracts) with a custom consumer/provider contract test suite covering the streams and auth endpoints: api/src/contract-provider.spec.ts verifies the API satisfies each contract, xstreamroll-sdk/__tests__/contract.consumer.test.ts verifies the SDK sends/parses what each contract describes. CI fails if either suite fails (#399).
  • @xstreamroll/types workspace package (packages/types) as the single source of truth for domain types (User, Stream, StreamEvent, pagination, and API error shapes) shared by api, app, and xstreamroll-sdk (#376).
  • Kubernetes deployment manifests for API, App, Processing Worker, and PostgreSQL (#217).
  • Health check endpoints (/api/health, /livez, /healthz) to support container orchestrator probes.
  • SECURITY.md policy covering supported versions, SLAs, and private vulnerability reporting.
  • Comprehensive CONTRIBUTING.md guide covering monorepo setup, conventional commits, and PR expectations.
  • PostgreSQL-backed TagsDbRepository and StreamsDbRepository for secure, parameterized data access.
  • User registration page (/auth/register) in the Next.js frontend with Zod schema validation.
  • Stream visibility (#393): additive visibility: "public" | "private" field on Stream, with "private" as the safe default. New visibility query filter on GET /streams, opt-in visibility on create/update payloads, and migration 2026080101_add_streams_visibility.up.sql to backfill existing rows.
  • Stream event replay API (#396): new GET /streams/:id/events endpoint returning paginated historical events for a stream, backed by a new StreamEvents refactored into the existing repository pattern so swapping the in-memory store for a Postgres-backed one is a drop-in.
  • SDK streaming pagination helper (#390): paginateAll() async iterator on StreamingClient walks every page of a paginated list endpoint without forcing callers to re-implement cursor logic. Computes hasMore from total / page / limit so the on-wire PaginatedResponse<T> type stays unchanged.
  • Browser-compatible SDK build (#387): the SDK now ships a dual CJS + ESM build (dist/ and dist-esm/), with a module, browser, and exports field on xstreamroll-sdk/package.json. Bundler entry points (Vite, Webpack, Rollup, esbuild) pick the ESM bundle automatically.
  • Pre-commit and pre-push hooks (#385): husky and lint-staged are wired (npm run prepare installs the hooks) so staged TS/JS files get eslint --max-warnings=0 + prettier --write; pre-push runs the package-wide tsc --noEmit suite in parallel for the api, sdk, and processing workspaces.
  • Mutation testing scaffold (#389): a stryker.config.json for the SDK plus an npm run test:mutation script that runs Stryker against the SDK retry / HttpClient / webhook signature surfaces. Opt-in (not in default CI) so existing CI workflows stay green until the team is ready to expand coverage.
  • ADR-0008 evaluating a message-queue replacement (#404): documents the candidate architectures, recommendation (Redis Streams), and phased migration path with explicit exit criteria.
  • PostgresLockManager coverage expansion (#401): added four edge-case tests in xstreamroll-processing/__tests__/leader-election.test.ts so the SQL-contract branches the existing in-memory suite could not exercise are now covered: (1) renew returns false when the same row has been taken over by a different worker_id; (2) release is a no-op for a stream that was never claimed; (3) releaseAll binds the current worker_id (not a wildcard); (4) acquire returns null when the DB hands back a row owned by a foreign worker.

Changed

  • Breaking (api): Stream and User id/userId fields are now serialized as strings in JSON responses instead of numbers, resolving the Stream.id: number (API) vs. Stream.id: string (SDK) contract mismatch. Internal storage (Postgres SERIAL columns) and request-side route params are unaffected — the change is at the response boundary only.
  • Breaking (xstreamroll-sdk): User, Stream, CreateStreamDto, UpdateStreamDto, StreamEvent, StreamEventRecord, and the pagination/error types are now re-exported from @xstreamroll/types instead of being defined locally, and no longer include fields the API never actually implemented (User.role, User.displayName, Stream.visibility). CreateUserDto now matches the real POST /auth/register contract (username instead of displayName) — previously any register() call would fail server-side validation.
  • app's server-side stream cache (lib/cache/streams.ts) now uses the shared Stream / PaginatedResponse types instead of a locally-defined StreamSummary / StreamListResult whose field names (items) had drifted from the real API response shape (data).
  • StreamOwnershipService in API backend now safely queries the PostgreSQL database via parameterized queries instead of relying on demo environment variables.
  • Prepared AdminStatsService for database integration to aggregate platform-wide stats.
  • Breaking (xstreamroll-sdk): StreamingClient now uses the fetch-based HttpClient (with shared withRetry) instead of axios. The axios dependency has been removed from @stellar/streaming-sdk. Callers that relied on axios-specific error shapes (AxiosError, error.isAxiosError, axios interceptors on the client instance) must switch to ApiError / HttpRequestError. The public StreamingClient method surface is unchanged.
  • ADR-0003 (docs/adr/0003-polling-based-processing.md) is enriched with an explicit recovery model, signals to revisit the decision, and a forward link to ADR-0008 (#391).
  • ESLint configuration now enforces consistent import ordering via eslint-plugin-import's import/order rule (groups: builtin / external / internal / parent / sibling / index) — autofix with npm run lint:fix. The lint script no longer fails on warnings so a gradual rollout is safe across the existing codebase (#403).

Fixed

  • UI ConfirmDialog component now properly handles async states and prevents dialog dismissal while action promises are pending.
  • xstreamroll-sdk README replaces the obsolete axios wording with the current fetch-based HTTP transport, removes the "track issue #34" placeholder for the upcoming WebSocket support, and corrects two stale status URLs (#395).

Security

  • Scheduled re-scan of published ghcr.io/<owner>/xstreamroll-{api,app,processing}:latest images so newly disclosed CVEs are surfaced outside of release windows. (#372)