- Ya
- 6 Agosti 2026, 22:12 UTC
- Mwandishi
- Kamo
- Ahadi ya
- d7b20b9
Four sites: delivery (none -> PENDING), void (open -> VOIDED), redelivery (open -> SUPERSEDED) and finish (open -> EXECUTED). Each passes the status it read BEFORE the write and the one it wrote, and LegalPendingCountTransition decides — so the rule lives in one testable place instead of four. Two status writes deliberately do NOT publish. PENDING -> IN_PROGRESS is stamped every time a member opens a package and both states are OPEN, so it would put a STOMP frame (logged at INFO with full payload) on the wire per page view. The filer's EXECUTED <-> EXECUTION_INCOMPLETE cannot cross the boundary at all: its FILEABLE set is exactly those two. Finish into AWAITING_OTHERS publishes nothing by the same rule. The member signed, the instrument is not complete, and clearing the badge would say otherwise.