Offline Recovery for Browser Uploads Without False Promises

Short summary

The article outlines honest design patterns for offline-resilient browser uploads, emphasizing that 'works offline' is not a single capability but several distinct behaviors that should be tested and documented separately. It covers service worker caching for static assets, IndexedDB state persistence, file re-selection verification via metadata, and per-file recovery UX instead of global error banners. The core principle is never implying bytes were sent when they weren't, and telling users exactly what remains to do.

• •Treat offline as multiple distinct capabilities—cache shell, persist metadata, retry chunks, survive reload—not one badge
• •Verify reselected files by name, size, modification time, and partial fingerprint; never trust filename alone
• •Show per-file recovery states and test on real mobile devices with airplane mode, captive portals, and browser restarts

Generated with AI, which can make mistakes.