Files
ims/docs/CHANGELOG.md
adminandClaude Sonnet 5.5 2fed177fdc chore(release): 2.0.0, merge roles into SSA
Bump version to 2.0.0 (breaking: reporter, supervisor and capa_owner merged
into ssa, routes moved under /ssa). Add 2.0.0 release notes to docs/CHANGELOG.md.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-06 22:16:01 +08:00

153 lines
6.0 KiB
Markdown

# Changelog
All notable changes to the **Setia Corporation HSE Incident Management System (IMS)**
will be documented in this file.
The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
and this project follows [Semantic Versioning](https://semver.org/)
(`MAJOR.MINOR.PATCH`).
> **How to version going forward**
> - **PATCH** (`1.1.0 → 1.1.1`): small bug fixes, no new behaviour.
> - **MINOR** (`1.1.0 → 1.2.0`): new features that don't break existing ones.
> - **MAJOR** (`1.x → 2.0.0`): changes that break existing behaviour.
> Add each new release as a new `## [version] - date` block **at the top**, under `[Unreleased]`.
---
## [Unreleased]
_Nothing yet._
---
## [2.0.0] - 2026-11-06
**Breaking:** the Reporter, Supervisor and CAPA Owner roles are merged into one
role, **Site Safety Assistant (SSA)**. Existing sessions carry the old role, so
those users must log out and in again after the deploy.
### Changed
- **Roles — `reporter`, `supervisor` and `capa_owner` become `ssa`.**
The role set is now `ssa`, `hse`, `management`, `admin`.
- SSA sees incidents for their own site and their own reports, can update
incidents, and works their assigned CAPAs.
- Role checks in the API, the nav, notifications (e.g. new-incident and CAPA
escalation emails) and the user manager now use `ssa`.
- Unknown roles fall back to the SSA navigation.
- **Pages moved under `/ssa`:** `/reporter` → `/ssa/my-reports`,
`/supervisor` → `/ssa`, `/supervisor/incidents` → `/ssa/incidents`,
`/capa-owner` → `/ssa/capas`. The header search for SSA targets `/ssa/incidents`.
### Removed
- Legacy Supabase code and dependencies: `lib/supabase/*`, `@supabase/ssr`,
`@supabase/supabase-js` and the `supabase/` migrations folder. The app already
ran on local PostgreSQL; `db/schema.sql` and `db/rls.sql` are the schema source
of truth.
### Affected modules
Auth and roles (`lib/auth/roles.ts`), navigation and shell (`components/layout/*`),
SSA pages (`app/(protected)/ssa/*`), incident, CAPA and addenda API routes,
admin user manager, email and CAPA-escalation notifications, RLS policies.
### Database migrations
- **`db/ops/2026-11-06-merge-roles-into-ssa.sql`**, applied by hand to production
on 2026-11-06 as the `postgres` superuser (the app DB user cannot bypass RLS).
One transaction: converts `reporter`/`supervisor`/`capa_owner` users to `ssa`,
recreates the `user_role` enum as `ssa, hse, management, admin`, recreates
`auth_user_role()`, then drops and re-creates all 34 RLS policies.
Take a `pg_dump` first. Users with the old role in their session cookie must log
out and in again.
- No migration is needed for the Supabase removal.
### API changes
- No endpoint was added or removed. The role value in the session and in
`/api/admin/users` is now `ssa` instead of `reporter`, `supervisor` or
`capa_owner`. Creating or updating a user with any role outside the four new
ones returns `422 Invalid role`; the default role for a new user is `ssa`.
- Page routes changed as listed under **Changed**.
### Testing
1. Log in as an `ssa` user. Check `/ssa`, `/ssa/my-reports`, `/ssa/incidents`
(site incidents only) and `/ssa/capas`.
2. Submit a new incident at `/report` and confirm it shows under **My Reports**.
3. Open an assigned CAPA and submit proof (note and photos).
4. As `hse` and `admin`, confirm dashboards, incident lists and the user manager
still work, and that the user manager offers only the four new roles.
5. Run `npm run test` and `npm run build`. Seven tests fail on `main` as well
(AI routes, bulk-import, email); no new failures were introduced.
6. In Postgres: `SELECT DISTINCT role FROM users;` returns only
`ssa`, `hse`, `management`, `admin`.
### Also in this version since 1.2.2 (summarised from git history)
- IMS Admin UI refresh: new tokens, Geist, `TableCard`, new shells; Incidents and
CAPA lists migrated.
- CAPA owners submit proof with a note and photos; HSE sees a photo gallery.
- Trucks are linked to a site; the report form lists all trucks.
- AI provider switched to a local model.
---
## [1.1.0] - 2026-07-16
### Fixed
- **Auth — expired login link no longer dead-ends the user.**
Clicking an old Supabase magic link / email OTP redirected to a raw error URL
(`error=access_denied&error_code=otp_expired&error_description=Email link is invalid or has expired`).
- Added graceful handling: expired or invalid links now show a friendly
"This link has expired — request a new one" screen with a resend action,
instead of leaving the error in the address bar.
- Increased the auth email-link / OTP expiry window in Supabase Auth settings
for the testing phase.
- Verified the Supabase **Redirect URL** configuration points to
`http://localhost:3000/...` for local development.
- **CAPA — "Assigned To" owner dropdown now populates.**
On the *Add CAPA Action* screen (e.g. incident `SC1-202607-0002`), the
**Assigned To \*** dropdown showed only the `Select owner…` placeholder and no
selectable users, blocking CAPA creation.
- Ensured the user/owner source table (`profiles` / `users`) is seeded with at
least one valid user.
- Added / verified a Supabase **RLS `SELECT` policy** so authenticated users can
read the owner list (missing policies were returning zero rows silently).
- Confirmed the fetched users are mapped into `<option>` elements
(`value` = user id, label = user name) so the list renders.
### Notes / Known operational items
- **Dev server must be running** for `localhost:3000` to respond. If you see
`ERR_CONNECTION_REFUSED` ("localhost refused to connect"), start it with
`npm run dev` and keep that terminal open before opening the app or clicking
any email link.
- `localhost:3000` only works on your own machine during development. Production
will use the deployed domain, and Supabase redirect URLs must be updated
accordingly at that time.
---
<!--
TEMPLATE — copy this block for each new release and fill it in:
## [X.Y.Z] - YYYY-MM-DD
### Added
- ...
### Changed
- ...
### Fixed
- ...
### Removed
- ...
-->