feat(events): rename Programme* models to Event*/Template*#267
Merged
Conversation
mindsers
force-pushed
the
feat/events/rename-programme-to-event
branch
from
July 20, 2026 06:57
c69532e to
7a52a99
Compare
Prisma model rename map: - ProgrammeTemplate → EventTemplate - ProgrammeTemplatePart → TemplatePart - ProgrammeTemplateServiceRole → TemplateServiceRole - ProgrammePartAssignment → EventPart - ProgrammeServiceRoleAssignment → EventServiceRole - ProgrammeTemplateResponsible → TemplateResponsible - ProgrammeTemplatePartAllowedRole → TemplatePartAllowedRole - ProgrammePartAssignmentAllowedRole → EventPartAllowedRole - ProgrammeTemplateServiceRoleAllowedRole → TemplateServiceRoleAllowedRole - ProgrammeServiceRoleAssignmentAllowedRole → EventServiceRoleAllowedRole Field-name adjustments: - Event.partAssignments → Event.parts - Event.serviceRoleAssignments → Event.serviceRoles - Member.partAssignments → Member.eventPartsAsAssignee - Member.partAssistantAssignments → Member.eventPartsAsAssistant - Member.serviceRoleAssignments → Member.eventServiceRolesAsAssignee - ExternalSpeaker.partAssignments → ExternalSpeaker.eventParts - Role.allowedForPartAssignments → Role.allowedForEventParts - Role.allowedForServiceRoleAssignments → Role.allowedForEventServiceRoles Migration 20260720300000_rename_programme_to_event does pure ALTER TABLE ... RENAME TO for the tables, renames every PK/FK/index to Prisma's expected convention, and rewrites historical entityType strings on AuditLog and NotificationEvent. RLS policies survive the rename automatically (bound to table OID, not name — same as the 20260507 user_role_to_permission precedent). Deliberately kept: - Programme in display-board — the domain vocabulary (the meeting programme rendered on the board). - programme_* i18n keys — values reference the same domain concept; NDJSON archive names stay for backup-import compatibility. - File renames (programme-*.server.ts → event-*.server.ts) — deferred to a follow-up PR to keep this focused and reduce merge conflicts.
…ile renames Applies the PR review findings on top of the initial model rename: - migration `20260720400000_finish_programme_to_event_cleanup` renames the `assignmentId` FK columns on the two allowed-role join tables to `eventPartId` / `eventServiceRoleId`, and rewrites still-pending `NotificationEvent.debounceKey` rows so the post-rename debounce lookup keeps matching pre-deploy pending entries - schema back-relations align with the new model names (`TemplatePart.eventParts`, `TemplateServiceRole.eventServiceRoles`, `Event.eventParts`, `Event.eventServiceRoles`) - helper files renamed from `programme-*` / `Programme*` to `event-*` / `template-*` / `Event*` via `git mv`, with imports swept across the codebase and the guard scripts / biome overrides updated to match - `ARCHIVE_VERSION` bumped 2.0 → 2.1 with a `rewriteLegacyEntityType` shim in the AuditLog importer so historical `Programme*` entityType strings become `Event*` / `Template*` on read; NDJSON archive filenames stay frozen for backward compatibility - stale "programme assignments" comments and doc references updated to the new domain vocabulary, with a Programme-vs-Event section added to the architecture conventions doc
…dels Rebase carry: the viewer-role tagging (#266) and part-role-labels (#268) PRs shipped tests that reference `db.programme*` and `event.partAssignments`, which are the pre-rename names. Retarget them at `db.eventPart` / `db.templatePart` / `event.eventParts` so the suite stays green on top of the Programme→Event rename.
…column Two runtime callers still passed the old `assignmentId` name to Prisma after the column rename to `eventPartId` / `eventServiceRoleId`: - edit.tsx loader — read fails with `Unknown argument 'assignmentId'` when opening the event edit page, because the join-table where/select used the pre-rename column name. - export-congregation.server.ts — writes the two `*-allowed-roles` NDJSON files. Selects `eventPartId` / `eventServiceRoleId` from the DB now and renames the field to `assignmentId` on serialize so the archive format stays byte-compatible with pre-2.1 importers. Typecheck passed because both sites had no `select` narrowing that tripped on the removed column at compile time — the failure only surfaced at query planning time.
"Role" was overloaded — it named both permission roles (Role model) and event assignment slots (ServiceRole model). Renaming the second one to ServicePart aligns the family of join tables into a clean, symmetric pattern: TemplatePart / EventPart → meeting parts (talks, readings) TemplateServicePart / EventServicePart → service parts (sound, ushering) <Parent>AllowedRole → which Roles are allowed on that slot Now "Role" unambiguously means permission role, and every AllowedRole join table reads the same way. Changes: - Schema rename: TemplateServiceRole → TemplateServicePart, EventServiceRole → EventServicePart, and their two AllowedRole join tables. FK columns renamed to match (serviceRoleId → servicePartId, eventServiceRoleId → eventServicePartId). - Field renames on Event, EventTemplate, Congregation, Member, Role. - Migration `20260720500000_rename_service_role_to_service_part`: mirror of the Programme→Event one — table rename via ALTER preserves rows, FKs, and RLS policies (bound to OID). Also rewrites AuditLog / NotificationEvent entityType strings from `EventServiceRole` to `EventServicePart` and rebuilds pending debounceKeys so post-deploy notification cancellation keeps matching. - Code sweep across ~70 files: Prisma model access, function names (updateServiceRoleAssignment → updateServicePartAssignment), types (ServiceRoleReleaseAssignment, ServiceRoleRow), local vars, and audit action enum (ServiceRoleAllowedRolesChanged → ServicePartAllowedRolesChanged). - i18n keys stay `_service_role_*` — the user-facing "service role" copy in French/English is unchanged, and renaming the JSON keys is out of scope for this PR. - NDJSON archive filenames stay frozen (`programme-service-role-assignments` etc.) for pre-2.1 archive backward compatibility, same principle as the Programme→Event rename.
Review flagged that the `rewriteLegacyEntityType` shim had wrong map keys. My earlier ServiceRole→ServicePart sweep rewrote them (`ProgrammeTemplateServiceRole` → `ProgrammeTemplateServicePart`, `ProgrammeServiceRoleAssignment` → `ProgrammeServicePartAssignment`), but real pre-2.1 archives store the original `ProgrammeTemplateServiceRole` / `ProgrammeServiceRoleAssignment` strings — never the interim names. The colocated test mirrored the same wrong strings, so it locked in the bug: every import of a pre-2.1 archive silently passed those two audit strings through untouched and wrote dead entity names back into `AuditLog.entityType`. Fixes: - Restore the original `Programme*ServiceRole*` keys in the map, so pre-2.1 archives map straight to the current `ServicePart` names. - Add `EventServiceRole` → `EventServicePart` and `TemplateServiceRole` → `TemplateServicePart` for 2.1-era archives (exported between the two renames in this same PR). - Rewrite the test to assert against the actual pre-2.1 strings and cover the 2.1 fallback too, with a comment referencing the old schema so future maintainers see why the keys look the way they do. Also flush comment/doc rot the review surfaced: - `architecture-conventions.md`: correct migration timestamp (`20260720300000` → `20260720350000`), rename stale `programmeAssignmentPolicy` / `assignServiceRole` references, mention the ServicePart follow-up migration and its rationale, extend the frozen-NDJSON note to list `programme-service-role-assignments.ndjson`. - `notifications.md`: "service-role assignee" → "service-part assignee" in the release-fire paragraph. - Migration `20260720400000` header: bump the sibling-migration timestamp and spell out the two actual pre-rename join-table names. - `template-role.ts` header: drop "programme part or service-role assignment" for the current "event part or event service part" vocabulary. - `days-off.tsx` inline comment: "service roles" → "event service parts".
Adds an integration test that seeds a synthetic archive with one row per legacy `entityType` (pre-2.1 Programme* + 2.1-era ServiceRole*, plus a current-name control), runs it through the real `importAuditLogs` path against the live test DB, and asserts every persisted row carries the current model name. The colocated unit test for `rewriteLegacyEntityType` already covers the mapping — but the review flagged that a same-authored unit test can lock in a wrong mapping. This end-to-end round-trip proves the shim actually rewrites entityTypes on the way into `AuditLog.entityType`, exercising Prisma serialization + write path in one shot. It would have caught the `ProgrammeTemplateServicePart` map-key bug at the branch level.
mindsers
force-pushed
the
feat/events/rename-programme-to-event
branch
from
July 20, 2026 08:43
0ed318a to
52ac70e
Compare
Second-pass review flagged that the new integration test seeded every row
with `actorId: null`, leaving the `idMap.getOptional('user-accounts', ...)`
branch unexercised. If a future rename tweaks the FK remap, a legacy
archive that carries a real `actorId` would silently null out its actor
without any test failure to surface it.
Extends the test with three actor shapes on the same run:
- Row 0: mapped source actor → asserts target actor id round-trips.
- Middle rows: `actorId: null` → asserts null passes through untouched
(the pre-existing case, still covered).
- Last row: source actorId not in `idMap` → asserts `getOptional`'s
null fallback is what actually lands in the row (documenting the
observed behavior).
Also refine two comments the reviewers surfaced:
- Drop the misleading "sort both by the LEGACY string" note — rows are
ordered by `entityId` and `cases[i]` maps to `rows[i]` because the
seed uses monotonic `id: 1000 + i`. State that up front instead.
- Tighten the shim's `TemplateServiceRole` / `EventServiceRole` entries
comment: no released 2.1 archive can carry those strings (both
renames ship in the same PR); they cover pre-merge / dev / preview
archives produced between the two rename commits.
… rename Six stale technical references caught in dev docs post-rename: - architecture.md model catalog listed `ProgrammeTemplate` / `EventKind` / `ProgrammePartAssignment` / `ProgrammeServiceRoleAssignment` etc; now `EventTemplate` / `TemplatePart` / `TemplateServicePart` / `EventPart` / `EventServicePart` / `TemplateResponsible`. EventKind is gone (merged into EventTemplate in PR #265). - architecture.md FK rule example used the old `ProgrammePartAssignment` / `ProgrammeServiceRoleAssignment` — now `EventPart` / `EventServicePart`. - architecture-conventions.md import example imported the deleted `EventKind` symbol; now imports `EventTemplateKey`. - coding-conventions.md `Setting and EventKind have compound unique keys` gotcha still cited the deleted `EventKind`. Real current set is `Setting`, `Role`, and `EventTemplate` (grep of schema.prisma confirmed). - coding-conventions.md aggregate-doctrine reference to `ProgrammeAssignment` — now `EventPart` (the policy file's name). - row-level-security.md tenant-scoped table list had every renamed model under its old name; refreshed to the current 12 Event/Template identifiers. - data-transfer.md `ARCHIVE_VERSION` still said 2.0 — bumped narrative to 2.1 with the shim rationale. `1.0` archive description switched from "programme parts/service-roles" to "template parts / template service parts". Conflict-resolution table row for "Event kinds" renamed to "Event templates". Product docs (`docs/product/*.md`) and self-hosting/managed docs use "programme" as the domain word — those stay unchanged per the user-facing vocabulary policy.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
PR-B of the events data model refactor — pure symbol rename. No behavioural change. Follows PR #265 (EventKind → Template merge).
Programmeprefix and the misleadingAssignmentsuffix. Tables now read as "EventTemplate describes the shape, EventPart is a materialised instance".Event.partsreads better thanEvent.partAssignments;Member.eventPartsAsAssigneepreserves the role hint the oldpartAssignmentscarried).20260720300000_rename_programme_to_event) — pureALTER TABLE … RENAME TO, rewritesentityTypestrings onAuditLog+NotificationEventso audit queries don't have to reason about two names forever. RLS policies survive automatically (bound to table OID, not name).Deliberately kept
Programmeindisplay-board— that's the domain vocabulary (the meeting programme rendered on the board), not a Prisma model reference.programme_*i18n keys — the values reference the same domain concept in French/English, and theprogramme-templates.ndjsonetc. archive names stay for backup-import compatibility.programme-*.server.ts→event-*.server.ts) — deferred to a follow-up PR to keep this one focused on the model rename and reduce merge conflicts for branches open against main.Test plan
pnpm test:typecheck— cleanpnpm test:unit— 2176 passpnpm test:integration— 265 pass, 1 skippedpnpm test:lint— 0 errorspnpm test:file-sizes,pnpm test:boundaries,pnpm test:server-barrel-exports,pnpm test:service-test-coverage— all passSELECT tablename FROM pg_tables WHERE tablename LIKE 'Programme%'returns 0 rowsSELECT DISTINCT \"entityType\" FROM \"AuditLog\" WHERE \"entityType\" LIKE 'Programme%'returns 0 rowsSELECT tablename, policyname FROM pg_policies WHERE tablename IN ('EventTemplate','TemplatePart','TemplateServiceRole','EventPart','EventServiceRole','TemplateResponsible', ...)returnstenant_isolationon each renamed tableFollow-ups (not in scope)
app/features/events/server/programme-*.server.ts→event-*.server.ts,ProgrammeDocument.tsx→EventDocument.tsx, etc.data_transfer_entity_programme_*i18n keys + entity file names once we're comfortable breaking backup compatibility.