Struct RemoteStageStore 

pub struct RemoteStageStore { /* private fields */ }

A StageStore backed by a remote noether-registry over HTTP.

All stages are prefetched into a local MemoryStore on construction, making subsequent reads (get, list) instant and allocation-free.

Implementations

impl RemoteStageStore

pub fn connect( base_url: &str, api_key: Option<&str>, ) -> Result<Self, StoreError>

Connect to a registry at base_url and prefetch all active stages.

api_key is sent as X-API-Key header; pass None if the registry runs without auth (local dev with NOETHER_API_KEY="").

pub fn from_env() -> Option<Result<Self, StoreError>>

Build a store from the NOETHER_REGISTRY environment variable. Also reads NOETHER_API_KEY if set. Returns None if NOETHER_REGISTRY is not set.

pub fn refresh(&mut self) -> Result<usize, StoreError>

Re-fetch all stages from the registry using offset pagination and rebuild the local cache. Each page is 200 stages; iteration stops when the server returns an empty page or offset >= total.

Call this if you know the registry was mutated externally.

pub fn base_url(&self) -> &str

The base URL this store is connected to.

pub fn get_live(&mut self, id: &StageId) -> Result<Option<&Stage>, StoreError>

Fetch a single stage directly from the registry, bypassing the cache. On success the stage is inserted into the local cache so subsequent get() calls will find it without another HTTP round-trip.

Returns Ok(None) when the server returns 404.

Trait Implementations

impl StageStore for RemoteStageStore

fn remove(&mut self, id: &StageId) -> Result<(), StoreError>

Removes the stage from the remote registry (DELETE /stages/:id) and then from the local cache.

fn get(&self, id: &StageId) -> Result<Option<&Stage>, StoreError>

Returns the stage from the local cache.

For a guaranteed-fresh read use RemoteStageStore::get_live (which takes &mut self so it can update the cache).

fn put(&mut self, stage: Stage) -> Result<StageId, StoreError>

fn upsert(&mut self, stage: Stage) -> Result<StageId, StoreError>

Insert a stage, replacing any existing stage with the same ID. Used to upgrade unsigned stdlib stages after signing is added.

fn contains(&self, id: &StageId) -> bool

fn list(&self, lifecycle: Option<&StageLifecycle>) -> Vec<&Stage>

fn update_lifecycle( &mut self, id: &StageId, lifecycle: StageLifecycle, ) -> Result<(), StoreError>

fn stats(&self) -> StoreStats

fn get_owned(&self, id: &StageId) -> Result<Option<Stage>, StoreError>

Return an owned clone of the stage. Useful for async contexts where holding a borrow across lock boundaries is not permitted.

fn list_owned(&self, lifecycle: Option<&StageLifecycle>) -> Vec<Stage>

Return owned clones of all matching stages.

fn find_by_name(&self, name: &str) -> Vec<&Stage>

Find all stages whose metadata name field matches exactly. Used by graph loaders so composition files can reference stages by their human-authored name instead of their 8-char content-hash prefix. Returns every match across all lifecycles; callers typically filter for Active.

fn get_by_signature(&self, signature_id: &SignatureId) -> Option<&Stage>

Look up the Active stage for a given SignatureId. This is the M2 “resolve signature to latest implementation” pathway: a graph that pins a stage by signature_id gets whichever implementation is Active today.

fn active_stages_with_signature( &self, signature_id: &SignatureId, ) -> Vec<&Stage>

Return every Active stage whose signature_id matches. Ordered lexicographically by implementation ID so iteration is stable across HashMap-backed stores.