diff options
Diffstat (limited to 'CHANGELOG.md')
| -rw-r--r-- | CHANGELOG.md | 43 |
1 files changed, 43 insertions, 0 deletions
diff --git a/CHANGELOG.md b/CHANGELOG.md index 7147906..5fb7650 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -11,6 +11,49 @@ point at which they are stable. ## [Unreleased] +## [0.10.0] - 2026-08-06 + +Tag edits no longer stall the window when a background sync is running, and +Sync is one control instead of two that disagreed. + +### Changed + +- **A tag edit made while a background sync runs is held and sent when the sync + finishes**, instead of being sent straight into a database open that blocks. + The open never failed, it blocked and then succeeded, and because the worker + is a single thread everything queued behind it waited too: the message pane + froze on whichever thread was selected first and replayed the queue on + release. The row keeps its tag in the meantime and the edit still counts as + unsynced, so the quit prompt cannot let work leave silently. +- **One Sync control.** The button beside the query bar is gone; Sync lives on + the toolbar, the File menu and its shortcut. The two used to behave + differently, and only the button showed the sync log, disabled itself, or + reported that a sync was already running. +- **The saved-query buttons moved onto the query row**, after the query field, + so the bar is framed by the account selector on one side and the saved + queries on the other. The row they occupied is gone and the thread list has + that space. + +### Added + +- **A clear button in the query field**, Qt's own, drawn inside the field and + shown only when there is something to clear. + +### Fixed + +- **Sync stayed clickable from the toolbar and the menu during a background + sync.** 0.9.0 disabled the button beside the query bar and nothing else, so + every other route to Sync could still start a run that could only be skipped. +- **A rejected tag write no longer clears the whole undo stack**, only the edit + that was rejected. + +### Internal + +- Two tests depended on the machine they ran on: one read the live + `/proc/locks` and failed whenever a real sync happened to be running, the + other asserted a window wider than the offscreen platform's screen. Neither + indicated a fault in the application. + ## [0.9.0] - 2026-08-04 Small corrections from using 0.8.0, most of them things the application was |