fix(web,api): storage settings post-save cleanup + activation reliability (#788) - #790
Conversation
#785 shipped the dev-only UI kit page reading env.ENVIRONMENT, but apps/web never declared the var — typecheck failed and the production 404 gate never tripped (undefined !== "production"), leaving the page live on prod.
…, no more ghost switch failures (#788) - Lane card is the page once a bucket is saved: connect pitch + walkthrough hidden, state badge (Active / Not in use yet), context line for where uploads currently go, Replace bucket reopens the wizard, and the stray 'Saved.' status line is gone. - window.confirm replaced with an in-page confirm box in both views. - 'Couldn't switch' ghost fixed three ways: storage verify/save/activate/ detach client calls get a 30s timeout (the pipeline legitimately includes a 5s public-URL probe and jurisdiction probing), the activate/detach usage reconcile moves off the response path via waitUntil, and on a failed- looking activation the client re-checks status before reporting failure — a promoted lane means the switch actually landed.
|
Deploying with
|
| Status | Name | Latest Commit | Preview URL | Updated (UTC) |
|---|---|---|---|---|
| ✅ Deployment successful! View logs |
uploads-web | 1a201b1 | Commit Preview URL Branch Preview URL |
Aug 23 2026, 01:05 AM |
Deploying with
|
| Status | Name | Latest Commit | Preview URL | Updated (UTC) |
|---|---|---|---|---|
| ✅ Deployment successful! View logs |
uploads-api | e9d1f3f | Commit Preview URL Branch Preview URL |
Aug 23 2026, 12:56 AM |
|
Important Review skippedAuto reviews are limited based on label configuration. 🏷️ Required labels (at least one) (2)
🚫 Excluded labels (none allowed) (1)
Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
…LaneId Review findings on the first cut: openWizard left the lane card (and its live Use-this-bucket/Switch-back buttons) stacked above the wizard, and the activation self-heal inferred success from the lane's absence in `lanes` — a stale-id 404 would also read as absent. The wizard now owns the page while open (closeWizard restores from the last-loaded status), and the self-heal requires `activeLaneId === laneId`.
…orage-settings-788
1 participant
Closes #788.
Two-state storage page
The post-save screen stacked the connect pitch, the 3-step walkthrough, the CTA, and three separate status strings on top of the saved-bucket card. Now the page renders one state at a time:
In-page switch confirmation
window.confirm(a native "uploads.sh says…" browser alert) replaced with an in-page confirm box in both views, covering all three confirmation sites (use bucket, switch back, forced switch back)."Couldn't switch. Try again." for a switch that succeeded
A lane activated >10 min after verification triggers a server-side re-verify (which legitimately includes a 5s public-URL probe and jurisdiction probing) plus an awaited usage reconcile — together they could outlive the browser's default 8s request timeout. The worker kept going and committed the switch; the client reported failure. Fixed three ways:
waitUntil(unit-test runtimes without an execution context fall back to awaiting).Drive-by fix
#785's dev-only
/dev/ui-kitpage readsenv.ENVIRONMENT, but apps/web never declared the var —pnpm typecheckfailed on main, and the production 404 gate never tripped (undefined ≠ "production"), leaving the dev page live on prod. DeclaredENVIRONMENT: "production"in apps/web's wrangler vars, matching the apps/api / apps/auth convention.Testing
pnpm test(4907 tests) green;pnpm typecheck(Node 24) clean including the previously-failing apps/web; lint + format clean.requireElementselector cross-checked against the markup (1:1); no nativeconfirm(remains.