siderust-archive

Scientific dataset manifests, provenance, checksums, and Rust accessors for the Siderust ecosystem.

siderust-archive is the small Rust crate and repository layout used to keep Siderust scientific data out of downstream application code. It stores dataset metadata in TOML, keeps raw payloads in their native formats, and exposes only the Rust bindings needed to parse, verify, fetch, or bundle those datasets.

Pre-1.0 archive: manifest shape and feature-gated APIs may still change. Release notes live in CHANGELOG.md.

Scope

• TOML archive and family manifests.
• SHA-256 checksum helpers and provenance records.
• IERS/USNO time-data parsing, runtime fetch, and bundled offline snapshots.
• Build-time generated accessors for committed coefficient tables.
• Dataset-family manifests for VSOP87, ELP2000, nutation, JPL kernels, Lagrange kernels, gravity, atmosphere, frames, constants, and Pluto.

Non-goals

• Hiding upstream data provenance.
• Fetching large or mutable datasets from downstream build scripts.
• Silently replacing version-pinned datasets.
• Re-implementing Siderust scientific algorithms in the archive crate.

Installation

[dependencies]
siderust-archive = "0.1"

IERS time-data parsers and the bundled offline snapshot:

[dependencies]
siderust-archive = { version = "0.1", features = ["time"] }

Runtime download/cache support for current IERS/USNO time data and JPL kernel metadata:

[dependencies]
siderust-archive = { version = "0.1", features = ["fetch"] }

Planetary and geophysical tables are opt-in:

[dependencies]
siderust-archive = { version = "0.1", features = ["vsop", "elp", "nutation"] }

The crate declares its own [workspace], so it is not absorbed into a parent workspace when used through a local path dependency.

Feature Flags

Archive Layout

archive/
├── CHANGELOG.md
├── Cargo.toml
├── LICENSE
├── MANIFEST.toml                # top-level dataset registry
├── README.md
├── schema/                      # manifest and binary-format specs
├── src/                         # crate sources and committed datasets
├── tools/validate/              # structural manifest validator
└── reports/validation/          # validation output, when present

Start with MANIFEST.toml. Each [[family]] entry points to a src/<family>/manifest.toml file that records source, generator, validity, units, references, payload files, byte counts, and SHA-256 checksums.

Manifest Conventions

• Metadata is TOML. JSON is not used for archive manifests or provenance.
• Payload files keep their authoritative upstream format where practical.
• Generated Rust tables are build artifacts under Cargo OUT_DIR; do not commit them.
• Dataset generators must record source, generator version, generated time, references, byte counts, and checksums.

Validate the registry and all referenced family manifests:

cargo run -p archive-validate -- MANIFEST.toml

Adding Data

1. Put upstream payloads under src/<family>/raw/ or the family-specific payload directory.
2. Update src/<family>/manifest.toml.
3. Register new families in MANIFEST.toml.
4. Run cargo run -p archive-validate -- MANIFEST.toml.
5. Build with the relevant feature enabled, for example cargo check --features nutation.

Large mutable datasets need a reviewed process before they become automated.

Time-Data Maintenance

IERS/USNO time data are operational data. The weekly Update Time Data workflow may refresh only:

• src/time/eop/raw/
• src/time/eop/manifest.toml
• src/time/bundled/snapshot.rs

The snapshot carries TOML provenance in time_data.provenance.toml. After refreshing raw files, the updater rewrites the managed [[files]] block in src/time/eop/manifest.toml with current SHA-256 checksums and byte counts so archive-validate can verify every committed payload. The workflow validates manifests before committing or publishing. Runtime fetch is still available behind features = ["fetch"] for users who want current upstream data without waiting for the next committed snapshot.

Other archive families are manual/version-pinned. The time-data workflow must not replace JPL kernels, VSOP/ELP, nutation, gravity, atmosphere, frames, constants, or derived kernels.

Releases

Manual releases should update CHANGELOG.md before tagging. Automated IERS/USNO data-refresh patch releases add a generated changelog entry for the new patch version and only publish when validation, tests, the WIP guard, and crates.io credentials all pass.

License

BSD-3-Clause — see LICENSE.

Each upstream dataset keeps the license and attribution declared by its original source. Check the relevant family manifest before redistributing data payloads outside this repository.