Watertown development manual

Watertown is a small data lake for querying, transforming, and publishing industrial telemetry. This is the current development manual, not a versioned stable-release contract. Public interfaces have compiled reference coverage, but cloud/API workflows and full browser interactions are not all end-to-end verified. Migration cleanup remains in D11.

Start here

Create and query a timeseries with a deterministic, offline example. Learn the exact interfaces in the synthetic-timeseries factory and CSV format references.

Build a timeseries site, publish a blog/RSS feed, or practice backup/recovery. These packages create fresh workspaces and assert results in numbered tests.

Reference

CLI commands and global options come from the production Clap definitions. Read shared concepts before combining URLs, credentials, imports, and temporal operations. Export is not replication. Factory transforms, ingestion, storage, and format providers state their prerequisites and evidence boundaries.

Site generation

The shortcode reference explains literal examples and source inclusion; the configuration reference covers YAML, routes, reports, feeds, and nested settings. The manual itself is a host-mode sitegen build: no pond, live services, external fonts, or visualization downloads are needed.

Contributing

Begin with architecture for developers: storage authority, directory locality, query/cache services, transaction boundaries, and how to extend the platform without duplicating its infrastructure.

See the authoring workflow and migration inventory. Only docs/manual/ is selected as page content; archives and design notes are not published. Generated fragments under _generated/ are excluded from page discovery.

Download compiled metadata. The footer records the renderer/documentation version and publication source revision; local previews are explicitly labeled working-tree.