Fingerprinting ... was Re: Thanks for MusicBrainz patch

[Daniel Dewald]

On Monday, 4. October 2010 16:51:59 Mark Kretschmann wrote:
> On Mon, Oct 4, 2010 at 4:06 PM, Stefan Derkits <stefan at derkits.at> wrote:
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
> > 
> > On 10/04/2010 11:29 AM, Mark Kretschmann wrote:
> >> Hey Sergey,
> >> 
> >> I just wanted to say a big "thank you!" for you MusicBrainz/MusicDNS
> >> patch. Our whole team has been impressed by the speed and accuracy of
> >> your work, and also about the constructive communication with you.
> >> 
> >> 
> >> Thanks, and we'd love to continue working with you, if you want to :)
> >> 
> >> 
> >> Mark.
> > 
> > Hey everyone,
> > 
> > I did not try the MusicBrainz Patch yet, but I must say I'm also
> > impressed about the Speed Sergey reacted to Feedback.
> > 
> > For me there is one important Location where Fingerprints should really
> > be, and that is in the Database, created via the Collection Scanner
> > (because then we have many possibillities of new Features).
> > Daniel Dewald, who is working on the lastfm Fingerprinting, wrote in a
> > previous Mail that this is at the Moment not possible because Phonon
> > can't output the Audio faster then normal Speed and Fingerprinting would
> > take too long.
> > 
> > So my Questions is:
> > - -) Till now (as I understand), the Collectionscanner doesn't use Phonon
> > and can only scan everything that Taglib can read. Wouldn't be Phonon
> > just for Fingerprinting (in the Collectionscanner) be a not really
> > needed additional Dependency?
> > So wouldn't it be better to use liblastfm Fingerprinting [or maybe also
> > MusicBrainz/libofa Fingerprinting] (at least in the Collectionscanner)
> > without Phonon?
> > 
> > Just seems a little unfair that MusicBrainz/libofa is in Trunk, but
> > liblastfm Fingerprinting not because Daniel wanted to base his work on
> > Phonon whereas Sergey just used xine or ffmpeg.
> 
> I'd say Daniel Dewald should use FFmpeg too.
> 
> The advantage is, we depend on it anyway (soon), due to Teo's
> Transcoding patch. Plus, it's cross-platform, and well maintained.

As stated in IRC and here before there is a reason why I chose phonon as audio 
source. Using ffmpeg (or any other library for that matter) would mean that 
the users can play files x,y,z.. and fingerprint either just a subset of those 
files or more files then he can play with his current backend. For transcoding 
this ain't a problem but in my opinion playback and fingerprinting should be 
possible with the same files. Inconsistency is just bad here. I'd rather throw 
my code into the trash then cripple it that way. As Sergey's code already is 
in trunk now I don't see my code going there anyway. In fact since my code was 
intended to produce fingerprints not only for tagging purposes but for the 
database etc. having 2 fingerprint solutions would just create even more 
confusion for the user (and the devs for that matter). I guess we are far 
beyond discussing intersections between those 2 codes now. The decision has 
already been made.

Daniel

P.S. For those who wonder why I didn't intervene earlier: I was on vacation 
till 25.09.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2921 bytes
Desc: not available
Url : http://mail.kde.org/pipermail/amarok-devel/attachments/20101004/1cd59b34/attachment.p7s