release: 0.10.0

[Vhivi]

Objective

Publish version 0.10.0 with existing-schedule optimization, configurable split shifts, and the solver corrections validated while integrating the feature.

Breaking Change Notice

No configuration-schema or endpoint compatibility break.

Existing 0.9.3 configuration files remain valid because every field introduced in this release is optional and has a runtime default. No migration is required.

Solver behavior is intentionally stricter or more accurate in several areas: manual shifts are locked by default, day/night rest windows are enforced, and worked hours and free weekends account for boundary-spanning assignments. Schedules relying on the previous calculations may now produce a different result or become infeasible.

Scope

• Add POST /optimize-existing-planning and the corresponding frontend mode.
• Support manual shifts and non-working statuses, with strict locking by default and optional soft preservation through existing_assignments_strict: false.
• Return warnings, suggestions, blocking reasons, impacted cells, and modified existing assignments.
• Add configurable half-vacations, segment coverage, labels, colors, penalties, and night/rest metadata.
• Add agents[].include_in_balance and the configurable weekend/Monday-night soft penalty.
• Correct day/night rest, weekly-hour, continuity, restriction, and free-weekend accounting.
• Publish the frontend package and changelog as 0.10.0.

Validation

• backend/.venv/Scripts/python.exe -m pytest
• npm test -- --runInBand
• npm run lint
• npm run build
• git diff --check origin/main...HEAD

Results

• Backend suite passes: 171/171.
• Frontend suite passes: 13/13 across 3/3 suites.
• Frontend lint passes.
• Production build completes successfully.
• GitHub Actions passes: backend and frontend jobs are green.
• A regression test confirms that a 0.9.3-shaped runtime configuration is still accepted.

Risks and Mitigations

• Risk: corrected hard constraints can reject schedules that were accepted through incomplete hour, rest, or weekend accounting.
• Mitigation: diagnostics identify blocking constraints and manual cells, while documented solver settings remain configurable.
• Risk: strict manual locks can make an edited schedule infeasible.
• Mitigation: users can disable strict locking to preserve assignments through objective weighting instead.

Deferred

• Optimization scope filters for selected agents or slots.
• Local draft save/load and optimization-result persistence.
• Existing Dependabot findings, which are unrelated to this release and require dedicated maintenance work.