https://sonarly.com/issue/40978?type=bug
After a frontend deploy, some iOS Safari/Chrome/PWA clients keep an old app shell and request removed JS chunks, then get stuck on a blank screen instead of recovering to login or a usable fallback.
Fix: I implemented a frontend-side loop breaker for stale Vite chunk failures in the same code path that currently hard-reloads the page.
What changed:
1) Added `getViteStaleChunkRecoveryState` utility to decide whether stale-chunk auto-recovery should reload again, based on current navigation type:
- first stale-chunk failure on normal navigation => `shouldReload: true`
- stale-chunk failure after a reload navigation => `shouldReload: false` (stop loop)
2) Updated `AppErrorBoundary` stale-chunk branch to:
- compute recovery state before generic exception capture
- send dedicated stale-chunk telemetry (attempt count + whether reload is attempted)
- reload only when `shouldReload` is true
- return early so stale-chunk failures do not flow through generic exception handling
This preserves the original “try reload once” recovery behavior while preventing repeated automatic reloads that can hard-lock iOS Safari/PWA users on a blank screen when cache remains stale.
I also validated this is in the correct layer (frontend runtime error recovery), checked blame on the exact reload lines to avoid accidentally reverting unrelated intent, and verified usage blast radius is limited to this module.
Authored by Sonarly by autonomous analysis (run 46706).
sonarly-bot
b2c18e34ad