Follow-up to #2181 / PR #2546 (which removed the legacy monolith detail renderer and the renderViaSchema kill-switch). PR #2546 was merged on green CI (type-check, unit 3012, DOM 2904, console production build, and a Playwright console smoke all passed). Two items from the original #2181 checklist could not be completed in the CI/agent environment and are tracked here.
1. Real-backend browser verification of detail pages (main)
The console smoke in CI runs against a mocked build with no @objectstack backend, so the four representative detail-page shapes were never exercised against real data after the legacy branch was deleted. Now that schema-driven is the only path, confirm each renders correctly end-to-end:
For each: no blank page, no fatal console errors, highlight strip present, audit meta footer at the bottom, and ?tab= survives a refresh. As a direct check that the kill-switch is gone, append ?renderViaSchema=0 to a detail URL and confirm it has no effect (still the schema page).
Repro (per-agent isolated stack, from AGENTS.md)
# backend — showcase, fresh temp DB, seeds admin@objectos.ai / admin123
cd framework/examples/app-showcase
pnpm exec objectstack dev --seed-admin --fresh -p 4010 # first run on a clean checkout: `pnpm setup` in framework root
# console pointed at it
cd objectui/apps/console
DEV_PROXY_TARGET=http://localhost:4010 pnpm exec vite --port 5190 --strictPort
# self-check → should return Northwind/Contoso:
curl 'http://localhost:5190/api/v1/data/showcase_account?$top=2'
Then log in at http://localhost:5190 (admin@objectos.ai / admin123), enter the showcase app via the launchpad (route by the app's name), and open a record for each shape above.
2. framework: drop the stale renderViaSchema source comment (minor)
@objectstack/spec no longer types renderViaSchema / the per-object detail.* block (verified in the pinned @objectstack/spec@14.6.0 — types are already clean). Only a source comment remains, noting the field "retires together with the legacy monolith render path." Now that the render path is gone (PR #2546), that comment can be removed in the framework repo.
Filed as the deferred remainder of #2181 after PR #2546 merged.
Follow-up to #2181 / PR #2546 (which removed the legacy monolith detail renderer and the
renderViaSchemakill-switch). PR #2546 was merged on green CI (type-check, unit 3012, DOM 2904, console production build, and a Playwright console smoke all passed). Two items from the original #2181 checklist could not be completed in the CI/agent environment and are tracked here.1. Real-backend browser verification of detail pages (main)
The console smoke in CI runs against a mocked build with no
@objectstackbackend, so the four representative detail-page shapes were never exercised against real data after the legacy branch was deleted. Now that schema-driven is the only path, confirm each renders correctly end-to-end:fieldGroups: groups render as titled cards in declared order; collapsible groups expand/collapse.record:pathstepper is emitted.showcase_account: Related tab / reference rail render, and eachrecord:related_listself-fetches lazily when its tab is shown (no eager fan-out on page open — the removed legacy path used to preload all of them).For each: no blank page, no fatal console errors, highlight strip present, audit meta footer at the bottom, and
?tab=survives a refresh. As a direct check that the kill-switch is gone, append?renderViaSchema=0to a detail URL and confirm it has no effect (still the schema page).Repro (per-agent isolated stack, from AGENTS.md)
Then log in at http://localhost:5190 (
admin@objectos.ai/admin123), enter the showcase app via the launchpad (route by the app'sname), and open a record for each shape above.2. framework: drop the stale
renderViaSchemasource comment (minor)@objectstack/specno longer typesrenderViaSchema/ the per-objectdetail.*block (verified in the pinned@objectstack/spec@14.6.0— types are already clean). Only a source comment remains, noting the field "retires together with the legacy monolith render path." Now that the render path is gone (PR #2546), that comment can be removed in the framework repo.Filed as the deferred remainder of #2181 after PR #2546 merged.