Arranger

Minimal, functional, and offline-friendly todo lists for personal daily use.

About the Project (WIP)

A self-hosted productivity web app inspired by Things for macOS, focused on smooth motion and a clean interface. Built from scratch to learn modern full-stack development; not affiliated with anyone.

A deployed demo reflects current progress.

Recording.mp4

Tech Stack

Docker · Bun · Typescript · SvelteKit + Svelte 5 · PostgreSQL

• tailwindcss for styling
• drizzle-orm for type-safe database access
• zod for validation
• @internationalized/date for date utils
• iconify for icons

Features / Goals

• Chrome & Safari — cross-browser support
• Projects — create, rename, archive, and trash; contains todos and groups (visual dividers)
• Todos — title, note, planned date, completion status, and nested checks
• Date picker — "infinite"-scroll calendar and natural language parser (e.g. "2nd fri in may", "2 days before oct 11")
• Drag-and-drop — intuitive way to move things around
• Functional views — Inbox, Planned, Archive, Trash, and Projects list
• Search — full-text search across todos and projects, with filters
• Panels — resizable multi-panel layout for placing views side by side; collapsible side bar, popup sub-panel, and switcher for quick navigation
• Offline-friendly sync — sequence-stamped server cache for delta fetches, with local mutation overlay for optimistic UI and push composition
• User account — sign up, sign in, email verification, Oauths, and account management
• Concurrent-write robust — server-side conflict resolution and client-side reconciliation for concurrent writes

Current Task

A well-rounded sync system.

Dev Setup

Requires only Docker and VSCode (with the Dev Containers extension) on your machine — Bun, Postgres, and all tooling run inside the containers.

1. Create your env file: cp .env.example .env — the defaults work out of the box. (Do this first: reopening starts the Postgres container, which reads .env.)
2. Open the repo in VSCode and run Dev Containers: Reopen in Container.
3. Create the database tables from the schema: bunx drizzle-kit push.
4. Start the dev server: bun run dev --host.
5. Or serve a production build with the bun adapter: bun run build && bun run serve:prod-local.
6. Visit http://localhost:5173.

Deploy to your VPS

A single VPS running Docker, where the whole stack runs as Docker Compose containers — Caddy (TLS), the SvelteKit app, and Postgres.

Internet → (DNS / optional Cloudflare) → Caddy (TLS, :80/:443) → web (Bun + SvelteKit) → db (Postgres)

You build and deploy entirely from your laptop over an SSH Docker context — no git repo is checked out on the server. For that, create the context on your laptop beforehand: docker context create myserver --docker "host=ssh://user@your-vps".

1. Point your domain at the VPS

Add a DNS A record your-domain.com → <VPS public IP>. Caddy provisions a Let's Encrypt cert automatically the first time the stack starts.

Behind Cloudflare's proxy? Set SSL/TLS mode to Full (strict). Otherwise Cloudflare talks HTTP to the origin, Caddy keeps redirecting it to HTTPS, and you get a redirect loop.

2. Set your domain in two files

• Caddyfile — replace arranger.gwo.me with your domain.
• docker-compose.prod.yml — set ORIGIN: https://your-domain.com on the web service.

3. Create the prod env file

Run cp .env.prod.example .env.prod and fill in the necessary env variables.

4. First deploy

# build images and start the stack (run from your laptop)
docker --context myserver compose -f docker-compose.prod.yml up -d --build

# verify TLS — issuance is async (a few seconds, up to ~30s after startup);
# follow the log and wait for "certificate obtained successfully" (Ctrl-C to stop)
docker --context myserver compose -f docker-compose.prod.yml logs -f caddy

# create/update the database tables — first deploy, and whenever the schema changes
docker --context myserver compose -f docker-compose.prod.yml --profile tools run --build --rm migrate

Then open https://your-domain.com — the app should be running, served over a valid cert.

5. Operate & update

# ship updates: re-run the deploy command — Compose rebuilds only what changed
docker --context myserver compose -f docker-compose.prod.yml up -d --build

# tail logs (swap web for caddy or db)
docker --context myserver compose -f docker-compose.prod.yml logs -f web

# reclaim disk — build cache accumulates and can fill the VPS
docker --context myserver system prune -f

# stop and remove the containers (named volumes — DB data — persist)
docker --context myserver compose -f docker-compose.prod.yml down