Elamur Groupantsschool-question-label
The groupant beside world/src/knowledge.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 source model | forward |
| ast | partial | ../../syntax.json |
| cost | partial source model | cost |
| descriptant | present | epoch-memory |
| invariant | partial source model | invariant |
| tests | existing scoped checks | checks |
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 |
|---|---|---|---|
school-question-label-behavior-complement | behavior_complement | partial source model | behavior-complement |
school-question-label-reverse-reconstruction | reverse_reconstruction | partial source model | reverse-reconstruction |
school-question-label-marked-position | label_projection | partial source model | marked-position |
school-question-label-delimiter-boundary | grammar_boundary | partial source model | delimiter-boundary |
Model text
Source-bound partial models, epoch 2026-10-29. Subject: world/src/knowledge.rs at the manifest SHA-256. Observation: Question separator and marked-answer decoding. Every relation below concerns this frozen source/configuration. Structural validation does not establish semantic correctness.
forward
Split on literal space-pipe-space, trim parts, require at least question plus two options; choose sole starred option or default first; preserve remaining option order.
invariant
Successful question output contains at least one wrong option; no uniqueness or nonempty string guarantee exists.
cost
O(input bytes + output strings) for split/trim and copied options.
checks
knowledge::tail_tests::the_starred_option_is_the_right_one. See ../../tests.md for executed scope. A proposed oracle below is not a completed check.
epoch-memory
2026-10-29: inspected the existing implementation before modeling. Preserved current behavior, including lossy decoding, index-sensitive ties and conditional updates. Existing witness tests distinguish active-scope learning from a wrong global witness. Counts represent distinct observations, not independent proofs. No source changes, performance gain, full-course result or autonomous combiner is claimed. AST remains explicitly absent; ranges bind source bytes only. Parser resources, panic/allocation behavior and complete inference semantics remain gaps.
behavior-complement
Type: behavior_complement. Domain and observation: Question separator and marked-answer decoding, as narrowed below. Operator/target: U is question(value,file) Result transitions; R is its delimiter/marker decoder. Inner-bound U\R excludes success with fewer than three split parts and success with two or more marked option positions. A star in the question position is not an answer marker.
Approximation: source-derived scoped relation, conditional on valid internal state and fixed flags/resources; unmodeled behavior remains unknown. Evidence/oracle: Existing two-star failure is a negative control; short-line cases are proposed.
reverse-reconstruction
Type: reverse_reconstruction. Domain and observation: Question separator and marked-answer decoding, as narrowed below. Operator/target: For (q,r,wrongs), compatible encodings include unmarked r-first and every insertion position of starred r among ordered wrongs, with whitespace and repeated leading stars normalized. Restrict encodings so other wrong options do not start with star. Do not infer original answer position.
Approximation: source-derived scoped relation, conditional on valid internal state and fixed flags/resources; unmodeled behavior remains unknown. Evidence/oracle: Proposed exhaustive insertion-position oracle.
marked-position
Type: label_projection. Domain and observation: Question separator and marked-answer decoding, as narrowed below. Operator/target: Let M={i>=1 | part_i starts with ''}. |M|=0 selects part1 unchanged; |M|=1 selects trim(trim_start_matches('',part_i)); |M|>1 errors. Wrong option order is the original index order with chosen index deleted.
Approximation: source-derived scoped relation, conditional on valid internal state and fixed flags/resources; unmodeled behavior remains unknown. Evidence/oracle: Existing test covers sole marker away from first and legacy fallback.
delimiter-boundary
Type: grammar_boundary. Domain and observation: Question separator and marked-answer decoding, as narrowed below. Operator/target: Only literal ' | ' splits fields; bare pipe or tab-pipe-tab remains within a part before trimming. Empty trimmed answer strings and duplicate strings are admitted if cardinality/marker guards pass. Format validation therefore does not imply semantically distinct alternatives.
Approximation: source-derived scoped relation, conditional on valid internal state and fixed flags/resources; unmodeled behavior remains unknown. Evidence/oracle: Prospective separator/empty/duplicate fixture; no new executed claim.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/world/src/knowledge.rs.groupant/format.groupant/question.groupant/