Repository navigation
Release 0.6.0: offline repository analysis and executable assurance - #75
abogun-product wants to merge 47 commits into
Conversation
The Hungry Crab run of 2026-09-25 against pypa/build left its configuration (.crab.yml, all defaults) and its ledger (.crab/ledger.json: one meal, eighteen nutrients, two of them served as drevendev#36 and drevendev#37) in the working tree, from where a roadmap commit of the 0.1.9 release branch swept them in by accident. They were taken back out of that branch; this commit offers them on their own, byte for byte, because the configuration says the ledger is committed (ledger: repo) and without it the next run would propose the same eighteen nutrients again. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…versions Direction becomes direction.bands.v2 and carries both accepted repairs in one rule move. Linkage is state-neutral (PV-REV-DIRECTION-CLOSED-TARGET-001): a change request linked to a declared target stays linked when the target closes, which repairs calibration finding 9, where closing A1 and B6 took the linked count from 2 of 6 to 0. The open-target count is still derived, for presentation only. Every active change request is LINKED, UNLINKED or UNRESOLVED (PV-DIRECTION-INCOMPLETE-001): a reference a complete register lacks is a broken reference and unlinked, with DIRECTION_TARGET_REFERENCE_MISSING; a reference nobody could resolve, or a title, body or milestone of the wrong type, is unresolved. Over L + U + R = N the completion set k in L..L+R is mapped through the unchanged predicates: one reachable band is DEGRADED / band / EXACT, several are UNKNOWN. Unreadable linkage evidence therefore no longer fails the whole project; LinkageEvidenceError is gone, and the record names what it could not read in linkage_unresolved. Horizon becomes horizon.bands.v2 (PV-HORIZON-PARTIAL-001). A partial target enumeration that returned a target proves the capability; its counts are observed-subset lower bounds, and only PH3, an open target already beyond the frame, forces a band: DEGRADED / EXTENDED / EXACT. PH0 to PH2 stay UNKNOWN, because Horizon declares no order to pick a representative by. Debt becomes debt.bands.v2 (PV-DEBT-PARTIAL-001). A partial register with a confirmed open item is DEGRADED / PRESENT / EXACT, one without is UNKNOWN, and, as issue drevendev#39 asked for Debt, a PARTIAL count is read as a lower bound only with its OBSERVED_SUBSET_COUNT proof. The target marker is a standalone token now (PV-AUDIT-TARGET-MARKER-SYNTAX-001): SubTarget: B1 and PreTarget:B1 link nothing. Counts that contradict each other (L + U + R != N, beyond > future > open) are InadmissibleEvidence, refused before classification and reported by the command line as an input error. The vector format learns two things the new families need: given.derive runs the derivation over stated inventories, so a case about how a change request becomes linked is executed where that happens, and expect.rejected states a refusal (HOR-PARTIAL-15). DIR-CLOSED-01..09, DIR-INCOMPLETE-01..16, HOR-PARTIAL-01..16 and DEBT-PARTIAL-01..16 are executable vectors. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Integrity history was reconstructed from whatever the provider still showed, so a failure hidden behind a passing re-run on a parent-level surface, or dropped by retention, left the counts as if it had never happened. The accepted contract, PV-HIST-002 (PV-REV-HIST-002), makes history durable. The carrier is the union of every attempt observed per parent per revision. A build now reads its immediate predecessor, the bundle the comparison already resolved, and adds that union as ci.revision_history_carried before evaluating; Integrity merges it with the provider's current parents and emits the result as derived.revision_history in the snapshot, on every path, UNKNOWN ones included, so history also crosses a bundle whose verification evidence could not be read. A missing, unverifiable, foreign or later predecessor is a gap that says so (REVISION_HISTORY_GAP); nothing older is ever carried in its place. verify checks that the source a bundle names is the bundle its manifest follows, and a build refuses observations that already carry history from somewhere else. The state of a revision is a function of its union: an observed failure is FAILURE_OBSERVED on any surface and stays; favorable evidence proves PASS_ONLY_OBSERVED only when every attempt of every parent is observed, so a check-suite pass, or a run whose earlier attempts are not all known, is UNKNOWN_HISTORY (R52), which makes the Vital UNKNOWN, because T9 puts unknown history above every conservative branch. A current verdict is never carried. The first 0.2.0 bundle of a project replays the revision records of its 0.1.9 predecessor (lineage devostasis.ci-history.v1) instead of starting empty. On the stored corpus every active revision is attempt-complete, so no fleet project loses its band to R52; all 326 stored bundle directories still verify. The rule moves to integrity.bands.v1+ci-unit-004+hist-002; the band table and the 0.1.9 judgements are unchanged. HIST-01..07, -09..11, -13, -15, -16 and -19 are vectors, HIST-08, -12, -17, -18 and -20 store tests; HIST-14 has no accepted migration to execute. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ligation PV-REV-TEST-004 accepted the final T9 reconciliation (PV-TEST-004, generator PV-T9-GEN-003) with zero CONTRACT_GAP terminals: each of the nine gap identifiers PV-T9-GEN-002 exposed maps to one deterministic obligation, and higher-precedence fail-closed conditions stay higher. With the Direction, Horizon and Debt repairs of this release every obligation T9 consumes is implemented, so T9 can be adopted. The implementation side of it is a finite generator: the acquisition tokens (CF, POS_ABSENT, PARTIAL, UNAVAILABLE, FORBIDDEN, UNKNOWN, ERROR, STALE) crossed with the regions each family names, about 1,270 cells, each evaluated and held to the accepted table: Horizon PH0..PH3, Clutter's confirmed floors, Direction's completion set and per-reference resolution, Integrity's unresolved, partial and unconfigured-recent regions, Debt's PD0 and PD1, and unknown history above every conservative branch. The oracles restate the accepted table rather than calling the evaluators. The T9v2 cell-key serialization remains the research process's to deliver as vectors. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The live run against a copy of the fleet store showed two costs the tests could not. The first 0.2.0 bundle of trade_simulation carried its 0.1.9 predecessor's whole revision series, about a megabyte, although every record in it had aged out of the 14-day window and would be dropped on arrival. And its snapshot grew from 12 KB to 393 KB, because 834 workflow_dispatch runs in two weeks attach to a handful of revisions and the carrier wrote each run as its own object. The carried observation now holds only in-window records, and a replayed 0.1.x record only the fields replay reads (revision, time and every parent's identity, kind and attempts). The carrier stores parents grouped by shape (kind, greatest attempt, observed attempts) with their ids, which is lossless: every parent keeps its identity and its own attempts, and a parent listed in two groups is corrupt history. On the same evidence the snapshot is 91 KB. Which of those runs should count as verification at all is the question issue drevendev#40 already asks the research side. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Version 0.2.0, the pins that name it, the example bundle regenerated under the four new rule versions (its change-request records now in the shape the adapter writes, one merged change linked to the closed release it delivered, so the example shows the state-neutral linkage: 6 of 10 linked where the open-target count says 5), and the changelog section. The roadmap moves B3 to done, the Direction, Horizon, Debt and T9 judgements to adopted, finding 9 to judged, and the count of named cases without a test from 63 to 62. It also says plainly that 0.2.0 ships before the condition it had set for 0.2 (Phase B closed, an outside consumer, no unadopted research): the owner asked for a large release (drevendev#47), and this is the part of it that rests on accepted contracts, so B7 and the Phase C contracts become the bar for 0.3. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… on it Durable history is keyed by immutable revision identity, and the new union code read the revision of every record directly, so a hand-written record without one ended the evaluation with a KeyError. A record that names no revision can be neither counted nor carried: it is now REVISION_RECORD_INCONSISTENT, the same defect as a contribution that contradicts its own history, and the Vital is UNKNOWN. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
B3, durable revision history, is delivered by this release. v0.2 described the condition the roadmap had set for 0.2 (phases A and B closed, an outside consumer, no unadopted research); 0.2.0 ships before it at the owner's request, so the release target closes and the condition moves to a new target, v0.3, as the roadmap now says. Ids are never reused. The pull requests that advanced B3 stay linked after it closes: that is the state-neutral linkage of direction.bands.v2, and this repository is its first beneficiary. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
drevendev
left a comment
drevendev
left a comment
There was a problem hiding this comment.
Manual research handoff: two inherited Integrity admission gaps
Unit: PV-MANUAL-REPOSITORY-HANDOFF-20261014-001 (RUN-433).
Inspected exact head: 09d497f16520b6dc20f5dd42077ff76774e1f5e1.
This is an advisory COMMENT review, not approval, a blocking review event, or acceptance of a new semantic contract.
1. Extra attempt numbers can establish complete history
In src/devostasis/revision_history.py, parent_from_current() admits a parent declaring current_attempt=2 with attempts 1, 2 and 3. parent_complete() tests only set(range(1, latest + 1)) <= numbers, so union_state() returns ('PASS_ONLY_OBSERVED', True) for three passing attempts despite the contradictory declared latest attempt.
Minimal package-level reproduction for the implementer:
from devostasis.revision_history import parent_from_current, union_state
p = parent_from_current({
'parent_id': 'github:actions:run:781',
'kind': 'github_actions_workflow_run',
'current_attempt': 2,
'attempts_observed': [
{'attempt': n, 'state': 'VERIFY_PASS'} for n in (1, 2, 3)
],
}, 'synthetic')
print(union_state([p], []))
# Current inspected decision: ('PASS_ONLY_OBSERVED', True)Required repair boundary: validate attempt-number consistency before asserting completeness. Preserve legitimate successive observations of one attempt (VERIFY_UNRESOLVED then VERIFY_PASS) and failure-sticky history; do not reject every differing same-attempt state as corruption. Include missing-middle, unexpected-higher, both input orders, valid transition and retained-failure regressions. Avoid allocating a set proportional to an untrusted declared attempt number merely to check contiguity.
2. Current revision multiplicity is not protected by the carried-history repair
_carried_records() now rejects duplicate carried revisions. That is useful, but _series_records() still returns duplicate current rows and reconcile() constructs current_by_rev using last-record-wins dictionary assignment. vitals/integrity.py::_bearing() subsequently iterates the original list and can retrieve that surviving record repeatedly.
At the inspected counting/classification boundary:
- Three unique passing revisions: 3 contributions,
SPARSE. - Same rows with one identical duplicate: 4 contributions,
CLEAN. - Conflicting duplicate failure then pass: 4 contributions, 0 failures.
- Reverse order: 4 contributions, 2 failures.
This is not a claim that the normal collector emits duplicate records. It is a saved/current observation admission and reduction-consistency gap, related to #38 and the scope of #57. Please reject contradictory or duplicate current identities before reconciliation, or use an explicitly specified canonical reduction consistently in both history and counting. Duplicating a revision must never improve sample strength. Add a full ObservationSet -> evaluator -> bundle replay regression, not only a _carried_records() test.
Verification and release boundaries
Ten isolated tests of source-extracted admission/reduction/counting functions passed, including negative controls and legitimate state transitions. These are NOT full Devostasis integration tests; no repaired implementation or complete-suite result is claimed. The package-level reproduction above is supplied for that next integration check.
Both findings are present in this cumulative release head, so fixing an earlier branch alone is insufficient unless the repair reaches this head. devostasis.ci-history.v2 is already used here; do not reuse that identifier for unrelated new external-Check-Run semantics. Preserve old bundle replay and make any required compatibility decision explicit.
Publication receipt: DV-MANUAL-20261014-PR75-a438e23c.
Manual research handoff — three independently reviewed admission gapsInspected head: 1. CI attempt vocabulary is not equivalent to known historyPrior independent judgement: In revision_history.py,
Minimal package reproduction: from devostasis.revision_history import parent_from_current, union_state
p = parent_from_current({
'parent_id': 'github:actions:run:781',
'kind': 'github_actions_workflow_run', 'current_attempt': 2,
'attempts_observed': [
{'attempt': 1, 'state': 'UNKNOWN'},
{'attempt': 2, 'state': 'VERIFY_PASS'},
],
}, 'synthetic')
print(union_state([p], []))
# Inspected behavior: ('PASS_ONLY_OBSERVED', True)
Verification: six freshly executed isolated tests of the inspected admission/reduction functions reproduced these three cases and three controls (valid pass, retained failure with UNKNOWN, unresolved-to-pass for the same attempt). These were not installed-package tests. 2. Current timestamps can contaminate durable CI historyPrior independent judgement:
The missing regression must exercise the complete chain, not only malformed-carrier rejection:
This four-bundle end-to-end sequence is source-traced, not executed here. Its immediate malformed-carrier result is UNKNOWN; the concern is a later falsely favorable result. Normal GitHub collection has not been demonstrated to emit malformed timestamps. Repair boundary: parse/admit timestamps before durable reconciliation; preserve trusted prior facts and durable uncertainty; do not skip malformed inventory rows while continuing to claim complete inventory. A repaired current carrier, true aging-out, and independently proven branch removal must remain distinguishable. Cover both current and replayable historical lineages. Do not invent acceptance of a new migration. 3. Recognized source-unavailability reasons are not evidence of their causePrior independent judgement: In codeanalysis/bundle.py, A safe regular Python file that fits the unchanged policy can therefore be replaced by Important limit: coverage remains PARTIAL, not falsely COMPLETE. The problem is unjustified omission and suppression of findings, not a claim of cryptographic forgery or an observed incident. Executable package-level probe for the implementer, using an ordinary temporary Git repository containing from copy import deepcopy
from devostasis.codeanalysis.engine import default_policy
from devostasis.codeanalysis.git import collect
from devostasis.codeanalysis.bundle import build, verify_content
config = default_policy()
raw = collect(repo, 'HEAD', config, '2026-10-16T12:00:00Z')
changed = deepcopy(raw)
entry = next(e for e in changed['entries'] if e['path'] == 'safe.py')
assert entry['status'] == 'ADMITTED'
entry.update(status='UNAVAILABLE', reason='UNSAFE_PATH')
changed['sources'].pop('safe.py')
_, members = build(changed, config)
verify_content(members) # Must not substantiate this contradictory exclusion.This complete probe is supplied for execution, not reported as executed. A size-based repair needs another distinction: a Git tree authenticates mode/path/object ID, not a separately supplied size for a blob whose bytes are omitted. In a fresh Git 2.47.3 fixture, a 71-byte Python source had blob Repair boundary: substantiate tree-backed path/mode/scope predicates and count/total-budget decisions only from trustworthy premises. Keep unavailable size claims explicitly unproven until a suitable evidence/compatibility contract exists; do not request unbounded blob reads merely to make the check pass. Preserve legitimate exclusions and deterministic budget priority. Add negative cases for all five reason classes, forged size, valid exclusions and old-format replay. Handoff and integrationThese are three independently reviewed findings being made visible at the cumulative release head. Their proposed Please implement and test the finite repairs in the release chain rather than treating another unchanged watch pass as new work. Existing #35/#38 cover related admission families; #73 remains the separate ancestry/root/change-count proof issue. Neither a green unrelated matrix nor this comment establishes that these regressions passed. No code, branch, tag, workflow or repository setting was changed by this handoff. Publication receipt: |
Owner-requested cleanup: current CI, #63 repair, and remaining release gatesInspected head: CI status correctionUpstream now has the protected #63 now has a concrete two-line configuration repairThe current Proposed diff, preserving schedule, grouping and cooldown: @@
- target-branch: "release/0.3.0"
+ target-branch: "release/0.6.0"
@@
- target-branch: "release/0.3.0"
+ target-branch: "release/0.6.0"The patch was checked and applied against the COMPLETE original file after verifying its Git blob SHA. Both target substitutions and absence of unrelated changes passed. This is prepared code, not an upstream commit. Actual scheduled Dependabot execution is NOT_RUN; #63 explicitly requires that operational confirmation after integration, so do not close it from a text edit alone. Integrity integration and roadmapThe A new regression suite imports six COMPLETE original modules, verified by Git blob identity, and exercises real ObservationSet/canonical JSON/history serialization. Baseline: 23 failed, 16 passed; repaired candidate: 39 passed on Python 3.13.5. This is module integration at #57's dependency set, NOT the complete 0.6 package/evaluator/HistoryStore/CLI/conformance suite. The source-file identity match permits preparing propagation; it does not substitute for full testing on this branch. The October 16 UNKNOWN-attempt/vocabulary and Source Admission handoffs remain separate unresolved work. The size-proof boundary must not be silently decided by an implementation patch. The four-bundle full evaluator/store regression is still required. None of these findings is closed by the current candidate. The owner-requested roadmap priorities are published as an applicable suggestion in review 5562151796. Preserve the existing release dependency order unless the owner deliberately changes publication policy; carry repairs into every final release head and retire #36/#37/#50 only after verifying their delivered content. Avoid reopening parallel implementations for issues already addressed by #64/#69/#75. The connected |
abogun-product
changed the title
andy-zen-dev
left a comment
andy-zen-dev
left a comment
There was a problem hiding this comment.
Fresh bounded release/admission review at a4761da. This cumulative PR and feature #81 now use the same source head; the older reviews at 09d497f do not establish acceptance of this 224-file candidate.
Two concrete blockers were revalidated in the current source and are tracked separately:
- #82: both current and historical Git-tree walkers lack per-tree duplicate-name and canonical-order rejection. Hash-valid malformed trees can give an ambiguous path inventory.
- #83: v1/v2 carried-history arrays are still read through
or [], converting explicitly malformed fields into empty evidence.
The old source-selection defect is repaired in this head (source_selection is recomputed and compared). The prior bounded-attempt repair is present as well. Do not reopen or blindly reapply those obsolete local patches. This review covers the actual admission paths/diff examined on the identical #81 head and release dependencies, not independent acceptance of all semantic changes or all conformance fixtures.
#81 has a newly published REQUEST_CHANGES for these defects. Complete the fixes in the existing feature → release/0.6.0 path, retain valid historical canonical bytes, run full tests and sample/v1/v2/work replay on the resulting exact head, then obtain independent final-head review before this master release. Keep the earlier prepared release gates, B1 fixture acceptance/T9, B7 and owner-deferred #71 explicit.
Upstream feature run 39617246533 is confirmed action_required, not a passed job matrix. Local full pytest/replay is NOT_RUN in this environment. No merge, tag, publication or issue closure is claimed.
3 participants
Cumulative 0.6.0 release candidate. Feature PR: #81. This PR uses the same final source head and retains the stacked independent review/release gates below.
Pinned source-tree replay previously trusted recorded roots, ancestry and change counts. This 0.6 increment adds bounded authenticated Git history proof and re-derives those claims offline while preserving historical bundle/packet bytes. It also materializes the remaining R11–R53 and the earlier operation cases, with actual governed evaluation, durable history and persisted bundle verification.
Target: E4
Target: B1
Final source head:
a4761da.Delivered behavior
devostasis.git-history-proof.v1; separate exact v1 replay. Hashed raw objects establish all parent edges, roots, merge ancestry, committer-time window and declared byte-line edits. Missing/capped proof stays PARTIAL; rehashed false roots/counts/COMPLETE claims fail.devostasis.ci-normalization.v1produces raw native evidence, normalized records and provenance. Governed Integrity recomputes and compares all four emitted envelopes. Altered provider/project/time/version/digest/records, contradictory immutable identities, unknown fields and broken pagination fail closed. Partial acquisition cannot prove favorable complete history. The new rule has an explicit per-Vital comparison boundary; absent new inputs, historical evaluation and bundle replay remain unchanged.The case-to-proof table, native CI contract and literal
tests/fixtures/ci-revisions.jsonstate the detailed R11–R53 acceptance. Exact fixture fidelity remains an independent review obligation, especially the current/history distinctions under later PV-CI-UNIT-004/PV-HIST-002 rules.Validation
examples/codeand newexamples/code-v2regenerated byte for byte.a4761daa9fa87026ea106a3d722e7299b3de3b8b.action_required; maintainer approval is still required.Independent acceptance and release boundary
PV-REV-CONFORMANCE-SURFACE-001 accepts the capability contract; PV-REV-TEST-002 accepts the permanent R-case expansion. Neither independently accepts these local fixtures or this concrete implementation. The missing-implementation-fixture debt D-1 is closed; B1 stays open for independent concrete-surface/whole-case acceptance and separate T9 cell-key delivery. B7 retains sustained outside-consumer evidence. Independent reviews are the next owner-directed phase and are not performed or claimed in this PR.
This remains 0.6.0. Supersedes readiness-only #80, includes its repairs and the cumulative 0.2–0.5 ancestry represented by #57/#59/#64/#69/#75, and targets
release/0.6.0. The cumulative #75 uses the same head. Earlier release artifacts and their independent gates remain separate. External scheduling is owner-deferred (#71). No merge or release tag is performed.Acceptance ledger for every prepared version
Checked implementation items below are execution evidence. Unchecked independent review, upstream authorization and maintainer release gates remain required.
Release acceptance through 0.6
This is the cumulative acceptance ledger for the prepared 0.2–0.6 increments.
They remain unreleased until maintainer merges/tags under CONTRIBUTING.md.
The latest published tag observed in the October 21 audit was v0.1.9.
Implementation evidence and independent acceptance are separate gates.
The assurance increment is part of 0.6.0, not a new release number. It
includes the October 21 readiness repairs and preserves older canonical bundles.
The pull request records final commit/run links; the commands below reproduce
the acceptance from a source checkout without network or private calibration.
0.2.0 — durable core history and incomplete evidence
or multiply immutable revision samples; timestamps and attempt identities are
admitted before reconciliation. UNKNOWN attempt evidence remains unknown.
immutable revision time, expiry and exact current/historical replay.
T9 generator execute; unavailable evidence cannot become favorable zero.
Evidence:
tests/test_ci_history_admission.py,tests/test_revision_history.py,tests/conformance/, core exampleverification. The #57 fork matrix passed on Python 3.12/3.13/3.14 at
b849923.The broader B1/B7 milestone is still open; this ledger does not silently mark
that historical target achieved.
0.3.0 — Evidence to Action
acceptance; observed exact-head evidence gates merge admission.
Evidence:
tests/test_workscope.py,tests/test_work_handoff.py,examples/work. The #59 fork matrix passed ata41fac6on Python 3.12/3.13/3.14.0.4.0 — reproducible consumer handoff
acquisition distinctions remain identity bearing.
mixed, ambiguous or unauthorized nonidentity edges fail closed.
membership before/after source reads; packets bind exact verified bytes.
Evidence:
tests/test_receipt_identity.py,tests/test_compatibility.py,tests/test_work_handoff.py, historical examples. The #64 fork matrix passedat
c98e7edon Python 3.12/3.13/3.14.0.5.0 — Consumer Operations
collection or canonical publication; disabled invocations retain negative control.
survive private history export, verification and restore.
Evidence:
tests/test_work_operations.py,examples/work/adoption.json,CI archive round trip. The #69 fork matrix passed at
aa5e0e1onPython 3.12/3.13/3.14. Actual scheduled outside adoption (#71), collection
qualification (#62), long-term incremental archive scale (#67) and B7 remain
explicit owner/external gates.
0.6.0 — offline analysis and executable assurance
conservative finding comparison, SARIF and bounded verified source packets.
tree/blob objects; roots, window selection, edit counts and COMPLETE claims
are derived again without Git, a checkout, network or source execution.
caps and unsafe paths retain explicit partial states; incomplete roots cannot
establish comparable repository identity.
kinds and unused proof objects fail. An independent LCS oracle checks the
declared byte-line edit algorithm, including repeated/unterminated lines.
examples/coderegenerates byte for byte. Newexamples/code-v2and separate schemas replay independently; cross-lineage comparison is refused.
consumed RAW_RUNS provenance, one shared multi-Vital graph, all T8 substitutions,
canonical bundle operations and stateful HistoryStore operations.
native parent identities, exact overlap, proved reruns, unresolved precedence,
commit-time order/window and failure-sticky real-store lifecycles. Raw child
fallback and parent-level history limits remain explicit.
altered bindings; unknown lineage, contradictory identity, broken pagination
and incomplete favorable history fail closed. New behavior is opt in under
a separate rule; old bundle examples and legacy paths retain exact replay.
candidates, reports conflict and rebuilds from latest. Permission failures
remain explicit HISTORY_GAP.
capability-level research acceptance alone does not accept this implementation.
the October findings on Release 0.6.0: offline repository analysis and executable assurance #75 and the included readiness fixes.
cumulative Release 0.6.0: offline repository analysis and executable assurance #75 release PR, then merges/tags v0.6.0 under CONTRIBUTING.md.
Reproduce:
python -m pytest devostasis vectors --quiet python scripts/check_conformance_surface.py devostasis verify --bundle examples/sample-bundle python scripts/make_code_examples.py python scripts/make_code_examples.py --v2 git diff --exit-code -- examples/code examples/code-v2 devostasis code verify --bundle examples/code/bundle devostasis code verify --bundle examples/code-v2/bundle devostasis code verify-packet --bundle examples/code-v2/bundle --packet examples/code-v2/packet.json devostasis work verify --bundle examples/code-v2/work-bundle devostasis work replay --bundle examples/code-v2/work-bundle python -m pip wheel --no-deps .CI additionally verifies both work/core examples, deterministic repeated builds
and the operations archive restore on Python 3.12/3.13/3.14. Green checks establish
those executions at the recorded head; they do not complete the unchecked gates.
The former 62-case implementation gap is materialized; B1 retains independent
concrete-surface/whole-case exact-fixture acceptance and separate T9 cell-key
delivery. B7 retains sustained
outside-consumer evidence. Legacy v1 history and omitted blob sizes keep their
documented proof limits. No compatibility migration, core GitLab collector,
threshold retuning, aggregate health score or publication is implied by this PR.