Historical research candidates and release decisions¶

Historical technical and governance decision records¶

The records are presented in their contemporary sequence, with their original scope and scientific qualifications intact:

• 1.0 stabilization decision — why the post-0.12 stabilization audit did not justify adding another estimator.
• 1.2 release-qualification audit, 23 October — source/package evidence and limitations before the eventual publication.
• Proposed narrow 1.2 scope, 23 October — initial proposal for ten W1–W4 orchestration contracts.
• RC2 E5 content decision request, 23 October — experimental E1–E4 scope considered separately from stable workflow support.
• RC2 publication review checklist, 23 October — qualifications and safeguards requested at that date.
• RC2 publisher handoff, 23 October — production authorization and the required human/environment controls.
• Owner-directed RC2 authorization, 24 October — owner request, not evidence of independent review at the time.

For the eventual completed 1.2.0 release, use the published 1.2.0 notes, separate stable-publication governance, and release readiness with GitHub and PyPI evidence.

Earlier development-candidate context (retained)¶

The prior development source 1.2.0rc2.dev0 was unpublished at the time of its original reviews; the literal 1.2.0rc2 then underwent its own separate prerelease qualification. These are past development states, not assertions about today's stable distribution.

Historical 1.2.0rc1 workflow qualification is preserved under the archived RC1 page. Its qualification records remain immutable: F1–F6 experimental science cannot be retroactively attributed to that candidate.

The F1–F6 development draft PR #227 had 27/27 GitHub workflow groups green on exact head d48c939b01fbfa19f0e060902b6a61ebddbe7aa6. That is historical engineering evidence, not calibration of new hypothesis tests.

The later draft PR #235 introduced early study-design and quality utilities for separate review. Its contemporaneous pending tests and publication posture must not be presented as the current 1.2.0 release status.