You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Schema narrower than spec after #364: reproducibility and appraisal.method/re_execution rejected #366
Once #364 merges, schema/trace-claim.json (and its packaged copy) is narrower than the spec: a conformant record carrying the section 3.1.4 members is rejected.
The 11 September ruling on #247 gives this condition a severity: "A schema that is narrower than the spec rejects conformant records. That is a defect with a severity, not a documentation nit, and it should be filed as one." This is that filing.
appraisal.method (closed set, one value re-execution)
appraisal.re_execution {outcome in reproduced | diverged | not-attempted, observed_digest (required when diverged), reason (required when not-attempted), optional verifier_code_identity}, present iff method is re-execution
Plus producer vectors carrying the claim and an appraisal that re-ran it. Sponsor-free under CONTRIBUTING (schema changes tracking a merged spec change). I take it, sequenced immediately after #364.
#370 carries all of it: both schema copies, the models with the same three rules as validators, the four docs/schema.md additions, and examples/reproducibility-claim/ with 21 signed vectors (five accepted, sixteen rejected, two per rule) held to their verdicts, loci, signatures and recomputed digests.
Two notes in the PR body: the three SECTION_OF_PATH entries #365's guard will need once both are in, verified on a local merge; and that the section binds re_execution to method but does not bind method: re-execution to the presence of reproducibility, so the schema does not either.
Activity
github-actions commentedon Sep 16, 2026
🟡 Contributor Check: MEDIUM
Automated check by AgenTrust Contributor Check.
chernistry commentedon Sep 18, 2026
#370 carries all of it: both schema copies, the models with the same three rules as validators, the four
docs/schema.mdadditions, andexamples/reproducibility-claim/with 21 signed vectors (five accepted, sixteen rejected, two per rule) held to their verdicts, loci, signatures and recomputed digests.Two notes in the PR body: the three
SECTION_OF_PATHentries #365's guard will need once both are in, verified on a local merge; and that the section bindsre_executiontomethodbut does not bindmethod: re-executionto the presence ofreproducibility, so the schema does not either.schema(claims): carry the reproducibility claim and its re-execution …