- Shipped
- August 26, 2026 at 4:43 AM UTC
- Author
- Kamo
- Commit
- df51ca5
MediaController.getChatSessions hand-rolled the pre-extraction version of the "was this written by somebody else?" rule (rowVisitor != null && callerId **************** which only ever covered the public web-chat case. A SOCIAL session's visitor messages are stored under the connection owner for the same reason a web-chat visitor's are stored under the org's system member, so that branch never tripped for SOCIAL rows and the "not written by me" count returned 0 for the connection owner's own Chats list rows — while the nav badge, already routed through OwnAuthorshipAmbiguity one layer up in **************** counted correctly. Same conversations, two disagreeing numbers, in the same feature. Resolves each SOCIAL guid's owner id per page (socialOwnerByGuid, alongside the existing publicChatOwnerByGuid) and delegates the per-row decision to a new rowOwnAuthorshipIsAmbiguous helper that calls OwnAuthorshipAmbiguity .isAmbiguous directly, so the rule is stated once and this call site can no longer drift from it. Adds **************** unit-testing the new helper directly (same pattern as OutboundKindTest) — including regression coverage for the two behaviors that must not change: an in-app support ticket filed by its own viewer stays not-ambiguous, and a public web-chat session viewed by its owner stays ambiguous.