With either 1.4.3 or 1.4.4, on Ubuntu Edgy.

The collection scan stops at about 30% percents and displays :

"Désolé mais l'analyseur de collection a échoué suite à de trop nombreuses erreurs.
Advice: A common source for this problem is a broken 'TagLib' package on your computer. Replacing this package may help fixing the issue.
The following files caused problems:
/media/docs/mp3/Films/X-Files/9x08_trust_no_1_teaser.rm
/media/docs/mp3/Films/In the Mood for Love/09 - Quizás, Quizás, Quizás (Perhaps, Perhaps, Perhap
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m
/media/docs/mp3/Club/Bob Sinclar - III/10 - Métro Blanche.m"

So as you can see, some unicode problems with mp3s, OGGs (second file). And also an unrecognized .rm file.
Any idea ? It didn't happen with Amarok 1.4.3 on Dapper. All files are local.

Thanks a lot for reading... And thanks ever so much to the Amarok team for their amazing work,

mahen

BTW, I tried to downgrade taglib and put the one provided with dapper instead : it makes no difference...

Hm, I saw about a dozen persons report similar issues on French forums. Maybe it does not affect persons who don't use not iso 8859 characters ?

But it is not only due to Amarok itself, as it worked under Dapper. Any idea ?

PS : another thread about this on ubuntuforums : http://www.ubuntuforums.org/showthread. ... ght=amarok

Hello,

I had the same problem, but with 1.4.3.
I didn't find the explanation but at least I found a way to have a collection built: See http://amarok.kde.org/forum/index.php/t ... 213.0.html

It seems it was due to a few files with exotic caracters in the file name. These files had names which were not even changeable with konqueror.

The main problem was that it was really difficult to identify these 2 or 3 files.
Because, after the crash, Amarok reported errors on files with accents and other exotic french caracters even they were not responsible for the crash.

Have you tried to index only the bob sinclar files for example ? And only the In the Mood for Love file ?

I really think there is a bug, even with bad encoding of the files, amarok collection builder should not crash.

Thanks for the reply !

However... Being French I also have many files with special characters (and also some Japanese ones), and they all appear perfectly under Nautilus (ext3 / utf8 I assume). And they all worked perfectly under the same versions of Amarok, but under Dapper. I don't really feel like removing the accents from my filenames

Nonethless, it's a good beginning

mahen wrote:However... Being French I also have many files with special characters (and also some Japanese ones), and they all appear perfectly under Nautilus (ext3 / utf8 I assume). And they all worked perfectly under the same versions of Amarok, but under Dapper. I don't really feel like removing the accents from my filenames

In my case the problems occurs in konqueror not in nautilus
French accent shouldn't be a problem, in my case, I had to rename these in order to find the other files which led to the problem.

In your case, maybe your japanese files are the cause of the trouble....

I do think it's something linked to filenames encoding, maybe not all are encoded in UTF8?
I don't know if in KDE there is somethinkg to the  G_BROKEN_FILENAMES=1 env variable used by glib to handle filenames wich are not in ut8 encoding.

I would like also to solve this problem, I am not eager to change filenames each time they have accents...

Hi again !

So I compiled the SVN version, with debug on., to see if it made a difference. (Ubuntu Edgy -- it used to work with Ubuntu Dapper with the same collection : local ext3 directory)

I can see hundreds of error messages like the following ones :

amarok:    [CollectionDB] [ERROR!] [virtual int SqliteConnection::insert(const QString&, const QString&)] sqlite_step error.
amarok:    [CollectionDB] [ERROR!] columns url, deviceid are not unique
amarok:    [CollectionDB] [ERROR!] on insert: INSERT INTO tags_temp ( url, dir, deviceid, createdate, modifydate, album, artist, composer, genre, year, title, comment, track, discnumber, bpm, sampler, length, bitrate, samplerate, filesize, filetype ) VALUES ('./mp3/Club/Bob Sinclar - III/13 - So High.mp3','./mp3/Club/Bob Sinclar - III',1,1162335651,1045582470,1,1,1,1,'1','So High','Encodé par CDex 1.40 Fr', 13 , 0 , 0 , NULL,311,191,44100,7459711,1)
amarok:    [CollectionDB] [ERROR!] [virtual int SqliteConnection::insert(const QString&, const QString&)] sqlite_step error.
amarok:    [CollectionDB] [ERROR!] columns url, deviceid are not unique
amarok:    [CollectionDB] [ERROR!] on insert: INSERT INTO tags_temp ( url, dir, deviceid, createdate, modifydate, album, artist, composer, genre, year, title, comment, track, discnumber, bpm, sampler, length, bitrate, samplerate, filesize, filetype ) VALUES ('./mp3/Club/Bob Sinclar - III/14 - Fantasy.mp3','./mp3/Club/Bob Sinclar - III',1,1162335652,1045582644,1,1,1,1,'1','Fantasy','Encodé par CDex 1.40 Fr', 14 , 0 , 0 , NULL,728,165,44100,15017319,1)
amarok:    [ScanController] RestartEvent received.

and when it displays the error message in Amarok itself :

amarok: END__: virtual bool ScanController::doJob() - Took 1.7e+02s
amarok: END__: virtual void ThreadWeaver::Thread::run() - Took 1.7e+02s
amarok: [ThreadWeaver] Job aborted: CollectionScanner. Jobs pending: 0
amarok: BEGIN: virtual ScanController::~ScanController()
amarok:  [ThreadWeaver] Threads in pool: 2
amarok: END__: virtual ScanController::~ScanController() - Took 4.1e+02s
amarok: [virtual KDE::ProgressBar::~ProgressBar()]
amarok: [ThreadWeaver] Threads in pool: 2

At the beginning, I can see a few strange messages as well :

*** stack smashing detected ***: amarokcollectionscanner terminated
amarok:    [CollectionDB] [ERROR!] Already-scanned file at /media/docs/mp3/Jeux/Ikaruga/Ikaruga - [09] Final Chapter (The Stone-Like).mp3 has same UID as new file at /media/docs/mp3/Jeux/Ikaruga/Ikaruga - [09] Final Chapter (The Stone-Like)-1.mp3


I have absolutely no idea whether it can help or not ! But I think my filenames are correct, and that the problem rather lies in some id3 tags parsing.

Problem solved, for me.

It turned out I had to remove *.rm files. It actually had nothing to do with UTF8 or samba shares.

This is worth reporting as a bug on http://bugs.kde.org if you (or someone else) hasn't already done so. Particularly mention .rm files.