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
Establish the authoritative inventory and Python 3.15 waiting policy #29
Refresh all public issues, pull requests, default-branch checks, releases, and local sidecar issue
records in the scoped repositories. Record a reviewed ecosystem decision that makes Python 3.12
through 3.14 blocking, treats Python 3.15 as a visible non-blocking canary for dependency-blocked
libraries, retains truthful package upper bounds, and replaces the September 30 expiry with objective
interpreter and dependency triggers. Update ecosystem issue 2 plus Data issue 37 and Diagnostic issue
45 so their labels, evidence, owner, and recheck condition match that decision.
Verify with fresh GitHub API queries, current PyPI file inventories for the named blockers, ecosystem
decision validation, and the ecosystem repository's complete documented quality gates. Done means
every scoped item has one current classification and no normal release is blocked only by Python 3.15.
Plan reference: .workspace/work/stable-libraries-and-skills-closure/plan.md - Issue 1 in milestone 2026-09-20 - Stable libraries and Skills closure.
The authenticated ecosystem audit passes Data 0.1.7, Engineer 0.1.4, Backtest 0.1.7, Specs 0.1.5,
Diagnostic 0.1.6.post2, and Models 0.1.4 at their current default-branch commits.
Live 0.1.2 remains the only failed ecosystem report. Its source-to-PyPI metadata mismatch and
missing canonical project URL are part of the pending 0.1.3 release tracked by Live #92 and
ecosystem Complete Live qualification and reconcile its sidecar #36.
Decision 0013 in pull request #41 makes Python 3.12 through 3.14 the blocking stable matrix for the
affected 0.1 release lines. Data and Engineer wait on Python 3.15 final or a Polars release after
1.44.2. Diagnostic waits on Python 3.15 final or PyArrow 26. Each affected library must keep a
truthful <3.15 upper bound and run a visible non-blocking canary at least monthly. A review trigger
requires fresh three-platform evidence but does not block an otherwise qualified stable release.
Pydantic-core 2.49.0 now publishes CPython 3.15 wheels, so it is no longer listed as a Diagnostic
blocker. Ecosystem #2, Data #37, and Diagnostic #45 now have normal priority because they track
upstream readiness rather than a current stable-release failure.
Activity
stefan-jansen commentedon Sep 21, 2026
Inventory refreshed at 2026-09-21T00:43:44Z.
Public queue
Models have no open issues. None of the seven libraries has an open pull request.
now carry the shared type, priority, status, compatibility, and ecosystem labels. Draft pull
request Standardize stable libraries and record closure evidence #41 contains the milestone commits.
Current evidence
Diagnostic 0.1.6.post2, and Models 0.1.4 at their current default-branch commits.
missing canonical project URL are part of the pending 0.1.3 release tracked by Live #92 and
ecosystem Complete Live qualification and reconcile its sidecar #36.
private and are assigned to the owning reconciliation issues Reconcile and remediate Data #32 through Reconcile and remediate Models #38.
Python 3.15 classification
Decision 0013 in pull request #41 makes Python 3.12 through 3.14 the blocking stable matrix for the
affected 0.1 release lines. Data and Engineer wait on Python 3.15 final or a Polars release after
1.44.2. Diagnostic waits on Python 3.15 final or PyArrow 26. Each affected library must keep a
truthful
<3.15upper bound and run a visible non-blocking canary at least monthly. A review triggerrequires fresh three-platform evidence but does not block an otherwise qualified stable release.
Pydantic-core 2.49.0 now publishes CPython 3.15 wheels, so it is no longer listed as a Diagnostic
blocker. Ecosystem #2, Data #37, and Diagnostic #45 now have normal priority because they track
upstream readiness rather than a current stable-release failure.
Standardize stable libraries and record closure evidence