Re: [Rhythmbox-devel] duplicate-format songs in library; speed of import

[Jonathan Matthew <jonathan kaolin wh9 net>]

On Mon, May 21, 2007 at 11:40:50AM -0400, shivers ccs neu edu wrote:
> I have two questions/issues with respect to Rhythmbox.
> 
> 1. I keep songs redundantly formatted in my music repository: every
>    song is stored full-resolution in flac and lossily compressed in
>    mp3. This means that when I import my repository into Rhythmbox,
>    I see every song twice. This completely hoses my ability to play albums
>    with any ease at all.

Why do you do this?  It seems like the easiest solution would be 
$ find /music -name '*.mp3' -exec rm {} \;   .. but maybe there's a
reason you don't want to do that.

> 2. I was really astounded by the slowness with which Rhythmbox imports
>    from my media repository. My repository has about 7500 tracks from 641
>    albums. Each album is a distinct unix directory, containing its tracks
>    in both flac & mp3 format; these files all have .flac and .mp3 extensions.
>    The directories also contain a bunch of small metadata text files, with
>    extensions like .db, .inf, and one or two with no extension at all, like
>    "md5sums". It took about 12 hours to load the songs into Rhythmbox's
>    library -- wow! It seems like reading about 100 bytes of information
>    from 15,000 files ought to take a few minutes, tops. I have no idea what
>    the problem is; I just report it.

Tag reading is a fair bit more complicated than you think it is.
Aside from that, it sounds like some of those small metadata
text files are triggering a bug somewhere in the rhythmbox metadata
reading code and/or GStreamer that stalls the metadata reader for a few
seconds on each such file.