Elamur Groupantsparse-cache-identity
The groupant beside world/src/parsecache.rs in the Mova repository: what the code computes, what it never does, how to go back from its outputs, its invariants, cost and checks.
Facts
Implementation material. A Mova code model describes this code at one source revision: partial models, finite checks and their stated scope. It is not a reviewed general exposition, and its labels are shown as written, never raised.
Components
| component | label | where |
|---|---|---|
| approximant | partial | Forward |
| ast | partial | ../syntax.json |
| composition | partial | Composition |
| cost | partial | Cost |
| descriptant | present | model.md |
| invariant | partial | Invariant |
| tests | bounded execution | tests |
Obernants
The typed relations around the forward model: the behaviour complement (what the code never does), the reverse reconstruction (every input compatible with an output) and specialised views.
| id | type | label | where |
|---|---|---|---|
nonbehavior | behavior_complement | partial source derived model | nonbehavior |
antecedents | reverse_reconstruction | partial source derived model | antecedents |
streaming-hash | chunk_invariance | partial source derived model | streaming-hash |
absence-boundary | metadata_partition | partial source derived model | absence-boundary |
Model text
descriptant
parse-cache-identity
Forward
Construct a sorted dependency map and hash its JSON; optional file absence is an explicit sentinel, while required file/read failures propagate.
Invariant
Mode and dependency names determine projection membership; file paths do not themselves enter optional file digests. The identity does not contain per-text parse results.
Cost
Linear in bytes read; file_hash uses a fixed 65536-byte buffer, while driver bytes and serialized dependency map are materialized.
Composition
identity calls file_hash, optional_hash, execution mode and model path providers. Selected node envelope includes hash, file_hash, data_root, identity and optional_hash. External execution-mode and model-path providers remain explicit unmodeled dependencies.
Epoch memory
2026-10-29, source SHA-256 da30cef6ff680f3c2853c158dada7d661307d0bbf0464bf058624bdd63d2c384. First source-bound instance; runtime source unchanged. Inspection exposed two important non-rollback effects: duplicate missing occurrences are sent separately and the final occurrence wins; hit/miss counters advance before parsing succeeds. Refresh is a persistent flag rather than a one-call invalidation. The source-bound partial AST is not semantic proof: no macro expansion, name resolution, cfg evaluation or type inference. General real identity/finish checks and concurrent-writer behavior remain untested. No speedup or complete parser-dependency guarantee is claimed.
2026-10-30: whole-source identity refreshed after appending native tests; this fragment's source range and selected syntax nodes are unchanged. See native transition domains and current evidence. Identity computation and finish remain unexecuted; fake open_key identities exercise only the cache transition/envelope boundary.
catalog
parse-cache-identity typed models
Subject: world/src/parsecache.rs at SHA-256 6545414761854d33b7b41f8a864a7f4c59eebeaee721eef68772bb4a7a652479; epoch 2026-10-30. Scope: source bytes [384,3551). Transition horizon is one invocation unless an entry names a call history. Successful ordinary allocation; no usize counter overflow; a stable filesystem during each modeled read; no claim of parser linguistic correctness. Source inspection is evidence of a conditional specification, not a semantic proof. Evidence status is bounded_execution only where tests.md names the exact checked observation; all other oracles are prospective.
nonbehavior
Type: behavior_complement. Status: partial_source_derived_model.
Domain and observation: U is identity input-observation/output tuples including mandatory dependency read errors; R is identity().
Operator and approximation: C subset U minus R contains successful identity after a failing executable/model hash read, and adding a CPU-only gpu/* part without the GPU branch. U also includes arbitrary maps; an unsorted insertion-history-dependent JSON key order is excluded by BTreeMap.
Assumptions: Stable exec mode/env; ignore hash collisions when inferring changed digest from changed parts.
Evidence and falsifiable oracle: Source-derived only; proposed dependency error injection and two insertion-order serializations. See tests.md for executed bounds versus prospective work.
antecedents
Type: reverse_reconstruction. Status: partial_source_derived_model.
Domain and observation: Observe K; hold current executable and schema fixed; vary remaining dependency readings.
Operator and approximation: All dependency observations d with H(JSON(parts(d)))=K remain compatible. Equal optional file bytes and absent path choices need not be distinguished. Arbitrary omitted environment entries vary freely; no SHA inversion algorithm or unique recovery is supplied.
Assumptions: Whitelist exactly as source at bound revision; hash compression necessarily has many preimages.
Evidence and falsifiable oracle: Probe optional-file fibers only; full-key preimage enumeration prospective. See tests.md for executed bounds versus prospective work.
streaming-hash
Type: chunk_invariance. Status: partial_source_derived_model.
Domain and observation: file_hash reads one byte stream through a 65536-byte buffer; hash sees the same bytes in memory.
Operator and approximation: Provided all reads succeed and EOF represents the end of a stable stream, file_hash(p)=hash(bytes(p)), independent of short read partition. Time Theta(B), hash state and buffer O(65536), excluding returned digest.
Assumptions: No concurrent write; Read does not fail midstream.
Evidence and falsifiable oracle: Probe lengths 0,1,65535,65536,65537,131073 compare direct and file hashes. See tests.md for executed bounds versus prospective work.
absence-boundary
Type: metadata_partition. Status: partial_source_derived_model.
Domain and observation: metadata(path) outcome followed by optional_hash.
Operator and approximation: NotFound -> Ok(absent); other metadata error -> Err; metadata Ok -> file_hash, including its read failures. A directory is not equivalent to absence. Absence is represented as a literal string distinct from hexadecimal digests.
Assumptions: Platform metadata error classes; ordinary digest formatting.
Evidence and falsifiable oracle: Probe two absent paths, equal existing files and directory rejection. See tests.md for executed bounds versus prospective work.
Native 2026-10-30 refinements: transition relations; source scopes remain unchanged because tests were appended.
tests
Checks
See ../tests.md and ../evidence.md for executed bounds. Models are conditional source-derived specifications; only the named probe observations are executed. The identity fragment has streaming/optional-hash evidence but no actual identity call. Batch and envelope fragments are exercised by finite probe oracles. Syntax ranges establish snapshot anchoring, not semantic correspondence.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/world/src/parsecache.rs.groupant/identity.groupant/