Added three new documents with nine Mermaid diagrams to make the strangler
fig strategy visible:
- README: container topology diagram at the start, with the proxy entry point
and three seams labelled
- docs/architecture.md: five diagrams tracing the exact implementation:
- The four seams and who holds authority at each boundary
- How by-id read goes through the resolver, but list-read bypasses it
- Case lifecycle state machine (the strategy in one picture)
- Take-ownership sequence with failure windows annotated
- Write-through error round-trip showing zero validation logic crossed
- docs/playbook.md: how to apply this to a production system:
- Write-path decision tree (five read/write patterns)
- Cutover ordering diagram (side-effects-free first, least recoverable last)
- Seven transferable rules with pointers to the files that demonstrate them
- Scope diagram of what's proven vs. left as your decisions
Resolved all 13 dangling § citations (to an absent spec doc) by linking to
the actual files or dropping them. Replaced portal-frontend/README.md
boilerplate with accurate content. All diagrams parse and link-check clean.
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Enables dry-run checking before committing to case adoption. The preflight
shares the same side-effect-free checks (steps 1–3) as the real take-ownership
handler, so it cannot drift from what will actually succeed. Returns the same
status codes and error shapes as the real endpoint (200 with wouldSucceed:true,
or 409/404/422 if it would fail).
Portal renders a "Vooraf controleren" button for legacy cases, surfaced through
the existing actions block pattern. Confirmed in smoke.sh with two cases: one
where preflight predicts success (and writes nothing), one where it predicts
a named invariant failure (matching what the real call reproduces).
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Would have caught the previous commit's bug: checks that GET /portal
(no trailing slash) 301s and that following the redirect actually
lands on a 200 on the same host:port, rather than only checking
/portal/ directly.
`return 301 /portal/` let nginx build the Location header from its
own internal `listen 80`, dropping the host's published port (e.g.
:8080) entirely - a browser hitting bare "http://host:8080/portal"
got redirected to "http://host/portal/" (port 80) and failed to
connect. Using $http_host keeps the port the client actually used.
Notes the /portal path, that it's kept alongside new-frontend rather
than replacing it, and updates the container count and click-through
intro accordingly.
Container count is now 10; adds checks that /portal/ and a direct
deep link both serve the Angular app (200) through the proxy,
proving both the routing wiring and the container's SPA fallback.
Full smoke run verified green on a fresh stack.
Adds portal-frontend as a new service (no published port, only
reachable via the proxy, same as new-frontend) and two nginx location
blocks for /portal - a bare-path redirect plus the prefix-stripping
proxy_pass. The existing /, /legacy, and /api/ blocks are unchanged.
Verified end to end: /portal/, a deep link, and / all return 200
through the single published proxy port, and /api/worklist still
returns the full 17-item worklist.
Multi-stage: node build with --base-href=/portal/, then nginx:alpine
serving the compiled dist with try_files SPA fallback for client-side
routes (e.g. a direct hit on /legacy/1001). Verified standalone: root
and a deep link both 200, base href correct.
Take-ownership keeps the placeholder's confirm() dialog before this
one-way per-case migration step, then navigates straight to the new
owned detail on success (201). Release-ownership navigates back to
the worklist on success (204), no confirm dialog. Both surface 422
InvariantViolationResponse / 409 MessageResponse verbatim in a banner
rather than a curated client-side message.
Owned-mode form embedded in case-actions where the redirect link
isn't used. verifiedItems is comma-split client-side for convenience
only; the server re-validates everything regardless. closurePending
renders as a neutral success note (ADR-001), never as an error. 422
InvariantViolationResponse surfaces invariant+message verbatim in a
banner, matching the edit-details form's approach.
Shared form for writeThrough and owned modes. Shows the ADR-002
disclaimer only in writeThrough mode (no client-side validation
stricter than legacy provides). Maps 400 ErrorsResponse to per-field
inline errors, 422/409 to a banner with the server's own text -
never a client-side switch over invariant names. Embedded in
case-detail with reload() on save so the actions block refreshes.
Resolves legacy/:id vs owned/:id off the matched route's static path
segment and fetches accordingly. Renders applicant, diploma, and
assessment fields, a seams panel (which backend serves each section),
and embeds app-case-actions. Read-only - write forms land next.
Table of WorklistService results with bucket/origin/search filters
(FormsModule ngModel); row click navigates to legacy/:id or owned/:id
based on origin, matching the placeholder's key scheme.
Renders record-assessment as a link only in redirect mode (otherwise
a form-trigger slot, filled in once record-assessment-form exists),
and take-ownership/release-ownership buttons purely off presence of
their optional action key. Emits requests upward rather than calling
the API itself - the actual take/release-ownership handling lands in
a later commit.
Red: case-actions has no template yet. Encodes the legacy-vs-owned
actions contract from CaseDetailResponseFactory - record-assessment
renders as a link only in redirect mode, take/release-ownership
buttons follow presence of their optional action key.
Empty placeholder components for worklist-list and case-detail,
wired into app.routes.ts (legacy/:id and owned/:id share one page
component since the rendered shape is identical). Replaces the CLI
welcome template with a minimal shell rendering <router-outlet>.
Builds the four-seam, three-write-path reference demo backend: case-framework
(seam D stand-in), legacy-backend/frontend (SQL Server, seams A/B/C targets),
and new-backend (Domain/Application/Infrastructure.*/Api implementing the
source resolver, take/release-ownership, write-through translator, and owned
assessment flow), wired together via docker-compose with a plain placeholder
frontend standing in for the Angular portal until Session 2.
All 11 Architecture.Tests pass and scripts/smoke.sh passes end-to-end against
a fresh `docker compose up`, covering acceptance criteria 1-3 and 7-22.
Fixes two real domain bugs found only once the stack ran for real: the BSN
eleven-proof checksum trivially passes all-zero digits, and the adoption
mapper silently treated a partial legacy address as absent instead of failing
loudly. Also fixes several environment-specific integration issues (rootless
Podman/SELinux bind-mount permissions, a buildah NuGet layer-caching bug,
SqlClient's invariant-globalization incompatibility, and an nginx path-prefix
mismatch for the legacy frontend).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Documents TDD/BDD/DDD methodology and Conventional Commits in CLAUDE.md,
and adds Claude Code model-switching setup (haiku/opus subagents, sonnet default).
ehoandClaude Haiku 4.5