In some circumstances the library can show stale data for a track in the time between a TrackPointer expiring and being saved to the database. Here's the rough series of steps that leads to this:
BaseTrackCache contains stale data, TrackPointer has modified data
1) Track N+cacheSize Loaded
2) TrackDAO QCache expires TrackPointer N -- TrackPointer reference count drops to zero
3) TrackPointer destructor called -- calls QObject::deleteLater (or whatever we use)
From this point on we are serving stale results for the track out of BaseTrackCache since TrackDAO will not serve Track N out of the cache (the TrackPointer has already been deleted and it is pending a database write). TrackDAO's weak reference cache (m_sTracks) will not convert to a TrackPointer since its strong reference is gone.
4) At some point, a library update re-draws the cell for the row for track N. RENDERING STALE DATA
5) QObject::deleteLater runs, deletes the TIO. ~TrackInfoObject runs, signals to TrackDAO to save.
6) TrackDAO saves the track, sets the track clean.
7) TrackN signals clean, BaseTrackCache runs updateIndexForTrackId(track n)
8) BaseTrackCache trackChanged signal hits BaseSqlTableModel, which emits row-redraw calls for affected rows
9) Library table redraws affected cells, drawing up to date data
did we fix this?