v0.2.0 final release validation

Disposition and candidate

Validation passed; the candidate is ready for the owner’s release decision. This is verification evidence, not a commit, tag or publication. Before external distribution, the owner must still resolve the recorded project-license decision.

Final execution: 0.2.0-candidate-5 execution. UTC interval: 2026-10-18T20:40:04.137930+00:00 to 2026-10-18T20:46:49.932074+00:00. Baseline Git revision: bee5a15fed8e0b89a21b9cced09d7a1e67bb3c40, plus the recorded uncommitted layout/test/runner corrections. Source/test identity: 3129fab28bbb196691acd6326a03a689f9a5c25997bf03d61950f29e9570eb29. Wheel: sphinx_traceability_view-0.2.0-py3-none-any.whl; SHA-256: ac1a23625c104ca26768856b1c00675c042fe62a0685bac7967a57f52f0ef6e5.

All required executable suites passed: 76 passed, zero failed, zero errors, zero skipped (58 Python/browser cases and 18 JavaScript cases). The run includes Windows Firefox 158.0, file and nested static-HTTP workflows with external requests blocked, Python 3.10.9, pinned Sphinx/Needs, Node v24.21.0, isolated offline wheel installation with Node absent, coverage, Sphinx warnings-as-errors, traceability and published diagram links. Exact commands, environment, per-case requirement mappings, logs, tested wheel and screenshots remain in the candidate execution.

The final wheel’s 11 package files were compared byte-for-byte with the tested source/assets; no mismatch was found. Wheel/source check (retained in the engineering checkout). Candidate 5 reused only candidate 4’s host-built capacity fixture after proving identical wheel, fixture inputs, Python, dependency hashes and runtime JavaScript. Every browser capacity test was rerun; no previous browser result substituted for a final test. The reuse proof and benchmark path are retained in candidate 5.

Criteria, corrections and history

Implementation criterion review and operating limits records the criterion assessment, with Executable requirement coverage providing executable mappings for every delivered system/ software requirement. Ten system and 53 software requirements are now verified passed. Offer occurrence mode later (STV_SWR_019), Retain state across later m... (STV_SWR_021), Key later expansions by occ... (STV_SWR_022) and Project later shared expans... (STV_SWR_023) remain not_assessed and outside the delivered scope. Stakeholder/element verification and all acceptance statuses are preserved. Verification-only source changes after execution are identified separately; they do not alter the tested product or tests.

The layout correction pools only conflicting desired positions and minimizes their displacement. An unrelated parent previously shifted 109.33px now stays exactly over its child; shared competing parents retain their separation and local balance. Regression coverage also checks cycles, isolated nodes, variable heights, test lanes and deterministic geometry. No dependency, component boundary, state transition or accepted stakeholder intent changed.

Expanded file/HTTP tests cover pointer and keyboard seed confirmation, hidden/ invalid/exact IDs, empty catalogs, repeat replacement, independent views, overview reset and failed/stale layout preservation. A diagnostic test initially used WebDriver clear(), which left hidden suggestion DOM unrefreshed; real Ctrl+A/Backspace events corrected the test. Its failed report and six-case retest remain retained. Four isolated HTTP diagnostics passed before the full failure traceback identified the new-tab race; they were never counted as a release pass.

  • Initial seed diagnostic failure (retained in the engineering checkout)

  • Corrected seed diagnostic: six passed (retained in the engineering checkout)

  • HTTP diagnostic (retained in the engineering checkout)

  • HTTP repetition 1 (retained in the engineering checkout)

  • HTTP repetition 2 (retained in the engineering checkout)

  • HTTP repetition 3 (retained in the engineering checkout)

Capacity and operating scope

The real 30,000-Need/90,000-edge host build passed in 309.590 seconds. It produced 34897944 graph bytes, 147393890 detail bytes and 415474937 total published bytes. Browser measurements repeated three times per transport, expanded to exactly 200 visible nodes and verified atomic rejection beyond the hard cap, collapse and first/cached detail loading.

Final repeated browser measurements

Transport

Startup median / max (ms)

Expansion median / max (ms)

Working set median / max (MiB)

file

552.3 / 559.2

87.0 / 232.0

1749.5 / 1769.2

http

591.9 / 612.9

76.5 / 200.0

1613.2 / 1642.3

Memory is the sum of Firefox process-tree working sets; shared pages may be counted more than once. It is neither private memory nor peak allocation. Fixture copying/cleanup is test preparation, outside the interaction timings. No owner-agreed numeric latency/memory budget exists, so measurements are reported without claiming a threshold pass. The 200-node cap is independently verified. Occurrence mode, touch, other browsers/platforms and widget-specific JavaScript parity remain deferred; remote resources retain static fallbacks.

Publication closure

After the verification-only edits, the full warnings-as-errors Sphinx rebuild, shared traceability validation, all source/published diagram links and strict coverage checks passed. Coverage reports 67 requirements and 76 executable cases, with no unmapped active IDs or unknown references. A fresh Firefox publication check passed (one case, 16 deselected); this repeats the publication case after documentation edits and is not added to the 76-case qualification total.

Only canonical verification metadata differs from the candidate’s frozen source/test inventory. All tested product, test and runner files remain unchanged; the tested wheel checksum still matches. No acceptance status was changed.

  • Closure checks and verification identities (retained in the engineering checkout)

  • Final publication JUnit report (retained in the engineering checkout)