mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-11-16 11:59:26 +01:00
A returning hard load of Mine restored its rows from the saved query cache but still waited for the first approvals, join-request and dismissal reads, because those items sort in among task rows by time. Those reads were not saved, so the cache gave no benefit on a hard load. Mine now saves the approvals, pending join requests and dismissals it showed as one record per user and company. On a returning visit with a restored Mine snapshot, that record stands in until the reads answer (and after a transient failure), so rows reveal at once and every read refreshes in the background. A record for another user or company, or one too large to save, is ignored, and the first-visit wait is unchanged. Selection now records its row key only when the selection moves. Recording on every list change stored the row now at the old index before the re-anchor ran, so a row inserted above the selection moved it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>