Expand description
The project’s state tree on disk: where it is anchored (paths), how a source
asset is found inside it (source), how the compiled blob is read out of it
(blob), how the runtime’s regenerable artifacts are kept in it (cache),
and how a regenerable container is published without a reader ever seeing it
part-written (atomic).
This is the engine’s only home for filesystem knowledge below the engine and
cook crates. concinnity_core::blob owns the pure bytes <-> metadata
transforms and performs no I/O; concinnity-core owns the runtime blob types
and the PayloadStore seam and knows neither paths nor files. The reads that
join the two live here.
Modules§
- atomic
- Replacing a file whole, for the state tree’s regenerable containers.
- blob
- Runtime blob access: the state root’s
data/path layout, the payload residency store, and all blob file I/O. The concinnity-blob crate owns the format contract (schema, header, version, bytes <-> metadata) and is deliberately I/O-free, so every read below isfshere plus a pure parse there. Blob data is read-only at runtime; concinnity-cook writes whatconcinnity_core::blob::encode_cnbreturns. - cache
- The runtime cache segment: regenerable artifacts the running application
writes for its own later launches, all of them in
cache/0. - paths
- The project state tree: where the engine’s state is anchored, and the names of the directories hanging off it.
- source
- Finding an authored source asset on disk.