Elamur Groupantsmova-seaspider-english-tests
The groupant beside tools/seaspider/tests/english.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 approximant |
| ast | partial | syntax.json |
| composition | partial | Composition |
| cost | partial | Cost |
| descriptant | present | Descriptant |
| invariant | partial | Invariant |
| tests | finite evidence | 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 |
|---|---|---|---|
behavior_complement | behavior_complement | partial source derived model | behavior_complement |
reverse_reconstruction | reverse_reconstruction | partial source derived model | reverse_reconstruction |
Model text
model
mova-seaspider-english-tests
Descriptant
English tests of Sea Spider with the native model: prompt-to-capability/argument table, refusal stages, and a served end-to-end run against a local HTTP fixture. Source: tools/seaspider/tests/english.rs.
Forward approximant
Planner table: 20 positive prompts with expected capabilities and arguments, 12 refusals with expected stages and one review candidate each, strict policy refusing loopback and .test. Served: seaspider::animal::serve in a thread on a private socket; English answers for stats, status, robots (fetched now, then from the store without a request), queued fetches; status answered within 1.5 s during a 3 s fetch; page order /slow, /u1 (urgent), /n1 with /private/x never requested and /u1 starting after /slow ended; past-tense status with the robots rule; page summary; per-host stats; six refusals sending no request; repeated id returns the journaled answer; STOP ends the server.
Invariant
No assertion relies on a host other than 127.0.0.1; every negative assertion is a path the server must not have seen.
Cost
About 12 s wall time (2 s per-IP spacing and a 3 s stalled response); model load once per test.
Composition
Parent: seaspider. Skipped with a message when MOVA_OCTOPUS_TEST_MODEL is unset; network only to 127.0.0.1 with a.test pinned there and private hosts allowed in the test config only.
behavior_complement
Not covered: TLS, real DNS, IPv6, angel restarts after a crash, journals after a restart, many concurrent clients, a deputy without the model. Counterexample role: each of the three negative controls turns one of these tests red. Evidence status: source-derived model with finite checks, not a proof.
reverse_reconstruction
A pass shows these prompts and fixture behaviours held on that run with that model; it does not reconstruct timings or prove behaviour for other phrasings. No unique inverse is claimed.
Epoch 2026-11-04 (Whales in English)
Epoch 2026-11-04: 'What is the status of URL?' is now read through its subject 'status' and refused at argument binding (argument_unbound) instead of unsupported_capability.
Epoch 2026-11-04 (Sea Spider live: singleton and callers)
Epoch 2026-11-04: The served test configures service.lock and callers octopus, crab; it checks a refused second in-process serve, Crab answered through mova_octopus::relay, a Whale refused with nothing queued, a Crab-style animal_ask deputy relaying «Is the queue empty?», and requests.jsonl rows.
tests
Checks
Command: cargo test --manifest-path tools/seaspider/Cargo.toml --offline --locked with MOVA_OCTOPUS_TEST_MODEL=$MOVA_DATA/download/0/0.4/en.bin (31 unit + 9 end-to-end + 2 English tests on a local 127.0.0.1 server; two identical runs on 2026-11-04).
Negative controls 2026-11-04: binding arguments to the first action instead of through the parse, allowing every robots decision, and ordering jobs by arrival only each turn the English tests red; restored afterwards. These are finite fixtures and one live smoke against the project's own site (2026-11-04); not a proof.
2026-11-04: cargo test --manifest-path tools/seaspider/Cargo.toml --offline with the 0.4 model: 2 English tests pass with the updated expectation.
2026-11-04: cargo test --manifest-path tools/seaspider/Cargo.toml --offline --locked with MOVA_OCTOPUS_TEST_MODEL=$MOVA_DATA/download/0/0.4/en.bin (34 unit + 9 end-to-end + 1 singleton CLI + 2 English tests; two identical runs on 2026-11-04) Negative controls 2026-11-04: a lock that always succeeds turns the singleton unit and CLI tests red; admitting every caller turns the served English test red (the Whale's fetch is queued); restored afterwards.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/tools/seaspider/tests/english.rs.groupant/