Red Hat Bugzilla – Bug 111785
Rhythmbox fails to read the artist tag of ogg files under Turkish locale
Last modified: 2007-11-30 17:10:34 EST
Description of problem:
Under the Turkish locale, uppercase 'i' is not I. Hence, when
rhythmbox reads song information from an ogg file, the artist
tag is not probably detected. In other words, some code either
in gstreamer or rhythmbox assumes lowercase of I is always i
when checking the tag...
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. export LC_ALL=tr_TR.UTF-8; or choose the Turkish locale under gdm
2. Run rhythmbox and add ogg songs with ARTIST tags
The artist tags are not be detected.
The tags should be detected.
This may not be a rhythmbox problem if it relies on some other library
to read the tags. If it is the case, the bug may need to be "forwarded"...
The very same problem affects titles too because of the uppercase 'I'
in the TITLE tag. I did not notice it first because filenames were
being used by rhythmbox instead.
I think gtk+ (or perhaps a related) library must have some functions
to do locale dependent tolower, toupper etc. instead of the regular
ones. Those should fix the problem here...
We already use g_utf8_casefold. If that doesn't behave correctly in a
Turkish locale, it's a glib bug.
By the way, could you verify whether or not this bug is still present
in the latest development Rhythmbox 0.7.x? (0.7.2 should be out soon)
Please close this ticket. The problem is fixed with the newer version
of Rhythmbox, at least as far as ogg files are concerned. mp3 tags do
have a similar problem, but I guess that is due to the mp3 tag library
(?) that rhythmbox uses.
Anyway, the general inflexibility, bad design etc. of mp3 tags, _any_
version, is convincing enough to convert mp3s to oggs!
Rhythmbox uses GStreamer to read metadata (tags) now in 0.8.
mp3 tags are a whole other issue because for v1 tags there is no
character set information, and for v2 even though there is a character
set flag, it's often set incorrectly.
ogg got it right by just mandating UTF-8.