- Rust 97.4%
- Nix 2.6%
JMARyA
5c2a271ffd
Itemdb sources are now structured: items live under <source>/items/ and flows under <source>/flows/. Flow definitions become markdown (frontmatter + body), are drawn from the same itemdb sources, and layer by id the same way items do (later source wins, deep merge). The body becomes a FlowDef description, surfaced by GET /flows. - itemdb loader scoped to <source>/items/ so flow files aren't read as items - flowdefs reads markdown, layered across config.itemdb.sources - drop the obsolete paths.flows config - migrate example/ and deploy/ to the new layout Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> |
||
|---|---|---|
| .woodpecker | ||
| deploy/kubernetes | ||
| docs | ||
| example | ||
| migrations | ||
| src | ||
| .gitignore | ||
| Cargo.lock | ||
| Cargo.toml | ||
| devenv.lock | ||
| devenv.nix | ||
| devenv.yaml | ||
| flake.lock | ||
| flake.nix | ||
| README.md | ||
| renovate.json | ||
| zensical.toml | ||
CDB
CDB is a primitive economic database. It tracks the flow of things — what exists, where it is, where it came from, where it went.
What does the C stand for? Some say economic. Some say commodity. Some say community. Some say commune. All of them are right.
What it is
- Small. Five primitives — items, states, transactions, locations, flows — that compose into whatever inventory situation you have.
- Scale-invariant. One person's pantry, a shared household, a maker space, a community.
- API-first. CDB is infrastructure. Automation and integrations live outside as operators and talk to it over HTTP.
- Yours. Self-hosted, no cloud, no analytics. The data about what you own and consume belongs to you.
- Federated (eventually). Sovereign instances exchanging items and demand signals, peer to peer.
The full documentation lives in docs/ — concepts, architecture, data model.
Running
CDB is a single binary plus PostgreSQL. An instance is configured by a config.toml and a set of definition directories (itemdb sources, locations, flows) — see Storage. Users and tokens live in the database, not in files.
Development
devenv up # starts postgres (and provides nightly rust)
cargo run -- example/config.toml # migrate + serve on 127.0.0.1:8080
Create the first admin (also works as lockout recovery):
cargo run -- example/config.toml bootstrap <name>
Log in with the printed one-time password via POST /auth/login, then manage users and operator tokens through the API — see Users.
Deployment
The container image is built with nix — no Dockerfile:
nix build .#containerImage
skopeo copy docker-archive:result docker://registry.example.com/cdb:latest
deploy/kubernetes/ is a deployment template, one manifest per resource (apply with kubectl apply -f deploy/kubernetes/): itemdb sources are cloned by git init containers (fetching and pinning repos is the deployment's job — core only reads directories), config lives in a Secret, and a minimal postgres is included for small instances.