Production Checklist

HyperDjango spans template rendering, runtime JS, and caching layers. Small config mismatches across these layers are a common source of release regressions.

Runtime and App Settings

• Set HYPER_FRONTEND_DIR to your deployed frontend source directory.
• Set HYPER_VITE_OUTPUT_DIR to built static asset output.
• Set HYPER_DEV = False in production.
• Ensure collectstatic includes Vite output and hyperdjango/static.

See the Assets and Vite page for the main asset build and manifest flow.

Caching and Action Responses

Action responses include no-store/no-cache and Vary headers for Hyper request metadata.

• keep reverse proxies from overriding these headers on action endpoints
• avoid caching action JSON/partial responses in CDN edge caches
• preserve X-Hyper-Request-ID, Last-Event-ID, and X-Hyper-Switch-Depth

For command-to-query handoffs, verify the originating mutation uses retry: false, commits durable state before returning action.switch_to(...), and can be recovered by refresh if the response is lost. Audit destination watchers as genuinely side-effect-free and test their ordered replay or idempotent replacement-patch contract under reconnection. Set HYPER_SWITCH_ACTION_MAX_DEPTH only if the default four-switch loop bound is too small.

Security

• keep Django CSRF middleware enabled
• send CSRF cookie or render {% csrf_token %} in base layout
• if using CSP, use Django's CSP middleware/context processor or otherwise expose request._csp_nonce so HyperDjango can nonce rendered asset tags, runtime scripts, and dynamically activated scripts
• if HYPER_DEBUG_TOOLBAR=True, confirm that exposing sanitized traces, SQL, request metadata, replay, pause, and clear controls is intentional and protected by the deployment's access policy; otherwise disable it

Client Contracts

• use stable DOM IDs/selectors for server-targeted HTML and delete patches
• enable strict targets (hyper-strict-targets) in QA to catch selector drift
• define fallback behavior for missing JS (full-page paths should still work)

Performance

• use sync: "replace" or explicit keys for rapid interactions (search, typeahead)
• use hyper-loading-delay to avoid flicker on fast requests
• prefer block rendering (render_block) for hot action paths

Testing

• run routing checks in CI: python manage.py hyper_routes
• add tests for route conflict cases and action response contracts
• test back/forward navigation with enhanced links/forms
• verify 422 validation flows for form-driven $action(..., {}, { form }) submits
• for every command-to-query handoff, interrupt the watcher in an E2E test and assert: the command was sent once, the watcher reconnected with a distinct command/watcher request ID pair, only the watcher inherited its own Last-Event-ID, and keyed loading remained active until the final watcher completed
• test configured switch-depth rejection and external abort/replacement of the complete chain when an application uses multi-switch workflows

Deployment Validation

• start app with production settings and run key routes manually
• verify assets resolve from manifest (no Vite dev server URLs)
• verify toasts/signals/swaps on at least one action-heavy page
• verify target-not-found errors are absent in browser logs