FS#4727 - Tag Database Compilation Fails at Random Points

Attached to Project: Rockbox
Opened by David Brunning (saveusfromcbeebies) - Saturday, 25 February 2006, 11:55 GMT
Last edited by Daniel Stenberg (bagder) - Saturday, 25 February 2006, 23:05 GMT
Task Type Bugs
Category Applications
Status Closed
Assigned To No-one
Operating System All players
Severity Low
Priority Low
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No


When trying to compile the tag database on my iriver (via USB 2.0 on a Windows machine), the process fails at seemingly random points in the directory / file scan.

On the java version the failure results in a clean exit and a database that is complete up to the point it reached in the scan.
On the perl version the failure is in the vorbiscomm component - no database is created.

Each scan process will proceed normally for a time, and then stop - but never at the same point in the directory structure! When I leave the executable files on the PC and point the command line at the iRiver music folder it seems at little more robust.
This task depends upon

Closed by  Christi Scarborough (christi-s)
Sunday, 19 March 2006, 12:37 GMT
Reason for closing:  Out of Date
Additional comments about closing:  Tag database is being withdrawn
Comment by Paul Louden (darkkone) - Saturday, 25 February 2006, 13:10 GMT
Is there some sort of message when it fails?

Also, the current tag database generation schemes are not really maintained any more. A replacement is in the works, so a fix for these is somewhat unlikely.
The current one only really works on Vorbis and Mp3s I believe (and the MP3s must have ID3 tags, not Ape tags).
Comment by David Brunning (saveusfromcbeebies) - Saturday, 25 February 2006, 14:38 GMT
From the perl version:

File does not exist or cannot be read. at line 55.
Can't call method "load" on an undefined value at line 111.

From the java version - simply the usual completion messages.

Not a major issue. Looking forward to replacement!
Comment by Daniel Stenberg (bagder) - Saturday, 25 February 2006, 23:04 GMT
The error at line 55 happens because it is a file it can't read or it doesn't exist.

The next error is a subsequent failure due to the first. This should be fairly easy to correct.

(my opinions about the java version is withheld)
Comment by David Brunning (saveusfromcbeebies) - Sunday, 26 February 2006, 08:42 GMT
"but never at the same point in the directory structure" - I should add that there is no contention for file access (i.e. no other process which is accessing these files). Looking at the code I am left wondering under what circumstances the file existance check could be caused to fail - and then under what circumstances that failure would happen in the inconsistent manner presented. I haven't experienced any difficulties with any other process that heavily use the USB2/iRiver combination (i.e. treecomp directory / file structure comparison, file copy to iriver, tag editing). Only other (vaguely) related issue I've had was that vorbisgain complained about some of the ogg comments tags, so will reset these and try again.