I'm giving Amarok 2 heavy use (~12K MP3s on random) for the first time. I was a big fan of 1.4, which I found easy and natural to use. Version 2... not so much. Here are the things I have noticed:

* You cannot skip to the next track with the default (Main) toolbar. Needless to say, this is vital when using a random playlist. I know the Slim toolbar has this, but why doesn't the Main one? Every jukebox needs a Skip button!
EDIT: I didn't realize that the prev/next track names doubled as navigation buttons. This is nice, and invalidates my point, but it doesn't look like a button!

* There is no way to edit the track details for the current track when using a random playlist. The rack is removed from the playlist when it starts to play, so fixing artist/album info requires searching for the track. This really should be a right-click option on the Current Track widget.

* Getting tags is hard. I cannot fetch tags for the current track (as with the above, I have to find the current track, then select it, then pick the Discogs from the very crowded (after a few scripts are installed) Tools menu. Even then, the Discogs script returns blank dialogs instead of actual results or 'not found' (MusicBrainz worked much better). I know this is a script problem, not an Amarok problem, but tagging is something I would expect to be integral to Amarok as it is crucial to managing a digital music collection. I would expect tags to be handled like lyrics: a subsystem with a dedicated Context Pane widget that is handled by a script category.

* Cover fetching fails about 10% of the time. Some album names that are found with last.fm's search page do not work in the Fetch Cover tool. The Fetch Cover tool does not seem to work with broad queries, e.g. sending a partial album name and displaying a list of possible matches. EDIT: well it does work, sometimes; maybe the query just takes awhile. Busy cursor?

* Moodbar doesn't "just work", even with the script installed. Again, a script problem and not an Amarok problem, but this kind of thing (along with the previous two items) was simple and automatic in 1.4. I read the instructions for generating Moodbars and decided eh, too much work. I'd rather spend a fraction of that time writing a complaint!

* Various Artists acts weird. Some tracks with Unknown album are listed under Various Artists, some are not. There seems to be no way to move these tracks from Various Artists, as the context menu 'Do not show under Various Artists' is not shown for the Unknown album. Perhaps albums should have a dedicated details dialog instead of repurposing the Track Details dialog.

* There is a lot of wasted space in the Context Pane widgets (the Info widget is a stellar offender here). It would be nice if some of these could have a configurable a size limitation (e.g. 25% of Context Pane) so that larger widgets could be stacked on smaller widgets. Of course getting them full size would never happen... maybe a double-click on the Context Pane would open an item in 'dedicated' mode. Or maybe there should be layout/container/view widgets in the context pane. OK, let's face it, the Context Pane acts weird.

On the subject of the Context Pane, how are new widgets added? Are they .so's, or does is JS supported as in KDE's plasma widgets?

Finally, I don't want to give the impression that I had a uniformly bad experience with Amarok2. In fact, Amarok2 has largely been awesome, and that always makes the annoyances stand out more (raised expectations and so forth). Fixing the tags in my rather motley collection should make things go more smoothly.

I will try to answer some of the questions you raise:

semper_non_sequitur wrote:* There is no way to edit the track details for the current track when using a random playlist. The rack is removed from the playlist when it starts to play, so fixing artist/album info requires searching for the track. This really should be a right-click option on the Current Track widget.

Please file a wish on http://bugs.kde.org

semper_non_sequitur wrote:* Getting tags is hard. I cannot fetch tags for the current track (as with the above, I have to find the current track, then select it, then pick the Discogs from the very crowded (after a few scripts are installed) Tools menu. Even then, the Discogs script returns blank dialogs instead of actual results or 'not found' (MusicBrainz worked much better). I know this is a script problem, not an Amarok problem, but tagging is something I would expect to be integral to Amarok as it is crucial to managing a digital music collection. I would expect tags to be handled like lyrics: a subsystem with a dedicated Context Pane widget that is handled by a script category.

This is a deliberate choice of the Amarok team, as Amarok is a music player, not tagging software. You can tag your tracks with a lot of specialized software like Easytag, kid3 or Picard, Amarok does expect the tracks to be correctly tagged to have a reliable collection. So please use an external tagger, this is not something that should be solved with scripts either.

semper_non_sequitur wrote:* Cover fetching fails about 10% of the time. Some album names that are found with last.fm's search page do not work in the Fetch Cover tool. The Fetch Cover tool does not seem to work with broad queries, e.g. sending a partial album name and displaying a list of possible matches. EDIT: well it does work, sometimes; maybe the query just takes awhile. Busy cursor?

Cover fetching problems are mostly solved by now in the git version, expect this for Amarok 2.3.2

semper_non_sequitur wrote:* Moodbar doesn't "just work", even with the script installed. Again, a script problem and not an Amarok problem, but this kind of thing (along with the previous two items) was simple and automatic in 1.4. I read the instructions for generating Moodbars and decided eh, too much work. I'd rather spend a fraction of that time writing a complaint!

Yes, it's so much easier to rant than to actually do something...*sigh* Moodbar is an external element we only display, expecting people to add the moods to their track with the external software. Amarok is not a tagging software, as I already said above. Maybe talking to the Moodbar developer(s) asking them to make mood generating easier could change our mind, so far this is not going to change.

semper_non_sequitur wrote:* Various Artists acts weird. Some tracks with Unknown album are listed under Various Artists, some are not. There seems to be no way to move these tracks from Various Artists, as the context menu 'Do not show under Various Artists' is not shown for the Unknown album. Perhaps albums should have a dedicated details dialog instead of repurposing the Track Details dialog.

Please the Amarok planet on http://amarok.kde.org/ for explanations about this by our developer jefferai, he has explained the problem over and over again.

semper_non_sequitur wrote:* There is a lot of wasted space in the Context Pane widgets (the Info widget is a stellar offender here). It would be nice if some of these could have a configurable a size limitation (e.g. 25% of Context Pane) so that larger widgets could be stacked on smaller widgets. Of course getting them full size would never happen... maybe a double-click on the Context Pane would open an item in 'dedicated' mode. Or maybe there should be layout/container/view widgets in the context pane. OK, let's face it, the Context Pane acts weird.

It's not the Context Pane, it is the individual widgets that don't work all well together, please file individual wishes or bugs about that in http://bugs.kde.org. Hint: most if not all of what you address is already filed.

semper_non_sequitur wrote:On the subject of the Context Pane, how are new widgets added? Are they .so's, or does is JS supported as in KDE's plasma widgets?

The widgets are plasma based and can't be added by external scripts. So far it is limited to be accepted by the Amarok core developers, as it is too big a security risk to let scripts add widgets. Please have a look at the code if you want to know more: http://git.kde.org/amarok

semper_non_sequitur wrote:Finally, I don't want to give the impression that I had a uniformly bad experience with Amarok2. In fact, Amarok2 has largely been awesome, and that always makes the annoyances stand out more (raised expectations and so forth). Fixing the tags in my rather motley collection should make things go more smoothly.

Thank you Our developers do the best they can in their limited time, so maybe turn down too high expectations as this is Free Software developed by volunteers, not money based development. Of course, everybody who can is welcome to give a hand

This is a deliberate choice of the Amarok team, as Amarok is a music player, not tagging software. You can tag your tracks with a lot of specialized software like Easytag, kid3 or Picard, Amarok does expect the tracks to be correctly tagged to have a reliable collection. So please use an external tagger, this is not something that should be solved with scripts either.

I understand the rationale, though I don't fully agree with it. Amarok 1.4 was impressive for how well it handled chaotic collections of mp3s. After importing 10K or so mp3s, there were about a hundred or so albums that had faulty tags. I was able to fix those as I encountered them during random play, as it was only a right-click away. With Amarok2, there are at least a thousand albums whose tags had problems, and changing each one is quite tedious. If I was coming to Amarok without having seen 1.4, I would not be as blown away.

This has made me finally get around to normalizing of the tags in my collection. The last time I did this (four years ago?) I used a shell script and id3v2 to build the tags out of the genre/artist/album/#_track directory structure. This time, I left Picard running overnight -- 1400 unmatched files, so it's about the same success rate as 1.4 (~ 100-150 albums).

I guess I would classify what I liked about 1.4, and what I am finding lacking in 2.x, as Collection Management. Everything about it -- tagging songs, generating moodbars (heh, snuck it in again!), fetching covers -- was easy and automatic in 1.4. In contrast, 2.0 is more about Collection Enjoyment (have you guys trademarked that yet?): building playlists (which in 2.0 are quite awesome, btw), enjoying song/artist 'context' (another excellent idea), searching, etc.

Many of the complaints about "1.4 worked so much better" (and there are tons) boil down to this difference between the intent of 1.4 and 2.0. Maybe 2.0 should have a first-run wizard that automates the import of a collection(including tagging + cover fetching), prompting the user for what tools to use (e.g. picard) and launching those programs. Of course that would mean that first-time import is different from the scan-directory import, so the wizard would have to be (optionally) used for every import. Needs more thought, but it could work.

Yes, it's so much easier to rant than to actually do something...

Always!

Seriously, though, Amarok is a media player. I've just built a new box, got the custom kernel compiled, got E17 configured, got mediatomb and netatalk up and running so all these mp3s can be used on other devices, and I still have to set up a couple of databases, my dev environment, and test instances of my company's servers in Virtual Box. The last thing I am going to spend time on is configuring a media player!

The widgets are plasma based and can't be added by external scripts. So far it is limited to be accepted by the Amarok core developers, as it is too big a security risk to let scripts add widgets. Please have a look at the code if you want to know more: http://git.kde.org/amarok.

That is unfortunate; it means that Content Pane widgets won't be developed by third parties, so they will remain somewhat half-baked. Actually it looks like they are already available as shared libraries (amarok_context_applet_wikipedia.so and friends), so really it's only a matter of publishing an API so that they can be built outside of the Amarok source tree. I know, I know, submit a feature request

Our developers do the best they can in their limited time, so maybe turn down too high expectations as this is Free Software developed by volunteers, not money based development. Of course, everybody who can is welcome to give a hand

Yup, that's the nature of open source development: no pay and no thanks! My own development time is pretty full; I can't see myself writing anything more than the occasional glue script (or a content pane widget, if they were less of a burden to build).