Version: 2.0.2 (using KDE 4.2.1) OS: Linux Installed from: Fedora RPMs I have the following Random Playlist configured: Proportional Bias Proportion: 100% Match: Play Count With: less than 3 So I would expect that all tracks that are added to the playlist must have a Play Count 0, 1 or 2. For most tracks that is done correctly, but sometimes a track with Play Count 3 or higher ends up on the list. This is reproducible.
My situation is slightly different. I have a dynamic playlist with multiple 0% biases to match specific Tags (music I really don't want to end up in the playlist). It works most of the time, but sometimes I a track with those specific Tags end up in the playlist anyway.
I have monitored the behaviour for some time now. My conclusions: the playlist filter (Playcount < 3) seems to work correctly BUT only if the item added has not been played before in the same session. All the wrong entries that shouldn't have been added had been played before in the same Amarok session. My theory: Amarok keeps the state of a song cached in memory. When a song is played its statistics in the database are updated correctly (playcount = N + 1) BUT the state in memory is not (playcount = N). The playlist filter (playcount < N + 1) is looking at the cached state and not the database and therefore re-adds the song to the playlist.
Confirmed: bug still exists in release version of Amarok 2.1
Any news on this?
Changing version and target *sigh*
*** Bug 217483 has been marked as a duplicate of this bug. ***
Still happens with 2.3.1. It is very strange, using the same criteria in bug description it always adds 10 songs *randomly*. Everytime I run the 10 songs are different and all of them with more than three execution. Just as to know, my Amarok is localized into Brazilian Portuguese, does the localition interfere in the search criteria?
Lamarque, please do not confirm bugs labeled for Git with a stable version, git is the latest development version and is ahead of the released versions..
Why not? This is around for more than one year since 2.0.2 at least. I am just inform you that there is people here complaning about a bug that should have been fixed months ago. Just saying to wait is not gonna fix it.
What is the state of this bug in Amarok 2.4 or 2.4. beta 1?
The major problems have been fixed, e.g. that the dynamic play list definitions weren't stored at all. So currently such a dynamic play list behaves correctly most of the time. But once in a while it still adds a song incorrectly: - comment #2 is still valid. - Amarok can't handle exhaustive dynamic play list definitions. "playcount < N" is an exhaustive definition, i.e. after each song has reached playcount = N, such a dynamic play list SHOULD NOT add any songs any more and Amarok should stop playing when the end of the play list is reached. The test case for the scenario is pretty simple: create a collection with 10 titles, a play list length of 5 and a dynamic play list definition "playcount < 1".
Thank you for the feedback. Setting current version to 2.4
The new dynamic bias implementation improves this behaviour. However it still does not treat track statistics that will change during a playback correctly if the track is in the playlist several times. e.g. this playlist: Bad: playcount 1, last played 1 week ago Thriller: playcount 0, last played never Bad: And now the track should be handled as playcount 2, last played 5 minutes ago