Independent · Designer and developer · 22 Nov 2026
Dog Diary
An offline Android app I am building to help my mum care for my epileptic dog, by logging food, symptoms and seizures and looking for possible triggers.
Overview
Dog Diary is an Android app for logging what a dog ate, how they were, and when they had a seizure. It is for my mum, who looks after my epileptic dog, and it is in active development and not released. It is a diary and not a diagnosis tool: it shows possible patterns to take to the vet, and its wording never says a food or event is the cause.
The problem
Seizures, upset tummies and itching can follow a trigger by hours or days, and a pen-and-paper log makes patterns hard to see. The person using it is not technical and has an older phone, so it has to be fast, use big buttons, need little typing, and never lose her records.
What I did
- Built a seizure logger with a one-tap timer ring, a safety banner for long seizures and clusters, a one-tap vet call button and a follow-up form for the details a vet asks about.
- Built meal logging with mixes of several foods, a food library with ingredients and proteins, and a starter list of common UK brands. Pasting an ingredients list ticks the proteins automatically.
- Added symptom logging with stool picture cards, an itch slider and body areas, plus stress and context events, and medicine dose logging that records missed doses.
- Wrote a correlation engine that compares how often a symptom follows a food or event against days without it, over time windows that suit each kind of symptom. It labels weak or thin data as "not enough data yet" and flags days where food and stress overlap.
- Added a vet report as a PDF and CSV, with a guard against spreadsheet formula injection in the export.
- Made data safety a first-class feature: SQLite in WAL mode with versioned migrations, weekly automatic copies, manual backup and restore with a preview, and optional Google Drive backup limited to the hidden app folder.
- Supported more than one dog, with profile photos, and added demo datasets that can be loaded from Settings to try the pattern screens.
- Designed it around a written DESIGN.md: a warm paper and deep teal look in light and dark, bundled fonts, large touch targets, calm motion and a reduce-animations setting.
- Set up the Android project with release signing, with the key kept outside the repo.
Stack
- Tauri 2 for the Android shell, with a Rust core for the database and the logic
- SQLite through rusqlite, with the SQL kept in Rust and never in the front end
- Svelte 5, Vite and TypeScript for the UI, with plain CSS design tokens and no UI kit
- Vitest for the front-end logic and the correlation engine, and cargo tests for the database layer
- jsPDF for the vet report, loaded only when it is needed
Outcome / impact
- 33 commits on 22 Nov 2026, covering every screen in the original plan from the Today view to the vet report.
- Not yet in daily use. Still to do: testing on the real phone and handing it over.
- It works offline and needs no account. Google Drive backup is optional.