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
test_combat_boundaries_and_private_knowledge_filter starts two same-modifier combatants without explicit initiative, then assumes combat_start always returns an active combat. Since #93 correctly requires an explicit ruling when initiative totals tie, the test intermittently receives a pending ruling and crashes while indexing a missing combat status.
This exact failure has occurred in unrelated PR #95 and #103 Python 3.12 full MCP jobs, while all feature-focused and Python 3.11 tests passed.
Acceptance criteria
Make this test deterministic with distinct explicit initiative values (or explicitly resolve the tie contract if tie behavior is what it intends to test).
Audit adjacent tests for the same “random initiative but unconditional active combat” assumption and fix only confirmed susceptible cases.
Summary
test_combat_boundaries_and_private_knowledge_filterstarts two same-modifier combatants without explicit initiative, then assumescombat_startalways returns an active combat. Since #93 correctly requires an explicit ruling when initiative totals tie, the test intermittently receives a pending ruling and crashes while indexing a missing combat status.This exact failure has occurred in unrelated PR #95 and #103 Python 3.12 full MCP jobs, while all feature-focused and Python 3.11 tests passed.
Acceptance criteria
Reference
packages/mcp/tests/test_runtime_integrity_mcp.py:2100-214033414864406and33415794987