BiblePath

Storing reads by book and chapter, independent of canon

The Protestant canon has 66 books and 1,189 chapters, the Catholic canon has 73 books and 1,334 chapters. If I stored progress against the active canon, switching canon would silently destroy data. Instead reads are stored as readChapters keyed by book id and chapter, with no idea which canon is active, and every stat is recomputed against whichever canon you are in. Chapters that fall outside the active canon stay in storage and are simply left out of the totals. The lesson is to store the raw facts and derive every view from them, so a settings change can never corrupt the underlying data.

A store that survives React Native remounts

Reads kept resetting, because the state lived inside a provider that React Native was remounting. I moved all persisted state into a module level store that lives outside the component tree and survives remounts, exposed through a useApp hook, and I gate onboarding with Stack.Protected route guards rather than a Redirect, because a Redirect remounts the whole navigator and wipes the in memory state. Persistence is a debounced save with a flush when the app backgrounds, never a direct storage write from a screen. If an app loses state on navigation, look for a remount, not a missing save.

Shipping three whole Bibles as offline JSON

The reader ships the King James, the Berean Standard, and the World English Bible as 77 per book JSON files generated from open source USFM text, so the whole thing works with no network. The Catholic Deuterocanonical books fall back to the World English Bible through a single effectiveTranslation function, so the reader never has a gap even where a translation does not cover a book. Precomputing content into the exact shape the app reads, instead of parsing at runtime, keeps a text heavy app fast and fully offline.

Hardening a restore that accepts a file from outside

Backup restore is the one place the app takes in a file a stranger could have edited, so two adversarial reviews went at it. The importer validates every entry, for example an isValidJournalEntry check, and refuses a malformed file with a reason instead of writing partial or bad state into storage. Treat any imported or synced blob as hostile input, and validate it at the boundary the same way you would a network response.

Changing a default without overriding people who chose

When I moved the default reader font from 17pt sans to 23pt serif, I could not just change the default, because that would override everyone who had already picked their own. So a coerceStoredReader step detects only the users still sitting on the exact old default and moves just them, and leaves anyone who customised untouched. A default change is a migration, and the people who already made a choice are the ones you must not step on.

What the version 1.2 rejection actually was

Apple rejected version 1.2 over two concrete things, a background audio mode entitlement the app did not need, and price wording in the copy. I dropped the background audio, set enableBackgroundPlayback to false on the audio module, and removed the price language, and it was approved. Rejections are usually this specific, a named capability or a phrase, not a vague judgement.

Reading plans that can never disagree with free reading

A plan pre generates its daily readings when you create it, but completion is read back from the same readChapters map that free reading writes to, so a chapter you read on your own counts toward the plan automatically, and the two can never drift apart. Only one plan is active at a time, enforced in the store. When two features touch the same idea, give them one source of truth rather than two states you have to keep in sync.

No account, no analytics, everything local

There is no account, no ads, no analytics, and no backend. Everything lives on the device, the daily reminder is a local notification, and the App Store privacy answer is honestly no data collected. Local first is not only a privacy stance, it removes a whole class of sync bugs and outages you would otherwise have to build for.