On Fri, Jul 08, 2005 at 03:00:24PM +0200, Jonas H. wrote:
> On Fri, Jul 08, 2005 at 02:45:19PM +0200, Daniel Stenberg wrote:
> > songdb.pl only writes hashes for mp3 files at this point, so I would guess
> > this makes oggs not possible to get runtimedb for. Are you by any chance
> > playing oggs?
songdb.pl should probably just hash the first 32kb of the ogg for now,
thats better than nothing.
> Not a bad guess, but I am in fact playing both mp3s and oggs. The info
> is kept while I'm in the same dir, (the song is still buffered perhaps?).
> If I play a file, rate it, forward to the next, and then rewind back,
> the playcount is 2 and the rating is whatever I set it to. As soon as I
> stop, however, the info disappears.
> > I _do_ think that the runtimedb code should be fixed to do a poor man's
> > approach in case there's no hash in the tagdb, since it'll be a pain to
> > always keep up the songdb prograams up to date with the recent codecs etc.
> > Or possibly we should make songdb.pl add a hash on the file name for the
> > cases we don't yet have hashes based on audio data.
> Yes, that's probably a good idea.
This would break runtime data for when you move the file or similar, its
better to do an actual hash.
> > The songdb.pl script also fails to include songs using other extensions
> > than .mp3 and .ogg, which it really should to be able to use the other
> > codecs with rundb.
> > Actually, the tagdb programs are all in rather sorry states:
> > 1) the perl version lacks all this as mentioned
> > 2) the java version isn't in CVS, it has no docs and no descriptions on how
> > to build it. There's no license info and it is >90 separate source files!
All code of it is gpl, i checked recently...
Should i check it in as source in cvs....? its a lot..
as for how to build it, javac -cp . -source 1.4 -target 1.4 *.java when
unpacked should do. aside from that you can run the .jar with java -jar
SongDB.jar, i recently wrote a tiny bit of docs on the wiki on how to
> > 3) the C version Nielse has started that was recently added to CVS is not
> > yet
> > working
> > There's still no GUI for any of these so our end users struggle.
> In my opinion, the C version should be the priority, since it has the
> best potential for being portable (keeping a perl runtime environment is
> sortof a hassle for some people, and don't even get me started on Java)
> and it could probably also be included in a gui app relatively easily.
> Jonas H
> -- I'm using Free software to post this. Are you?
Received on Fri Jul 8 16:09:36 2005