[mb-users] bad puid

[Frederic Da Vitoria]

2006/10/6, Wendell Hicken <whicken at gmail.com>:
>
> On 10/6/06, Frederic Da Vitoria <davitofrg at gmail.com> wrote:
> >
> >
> > But I continued my investigations and I checked the second track's
> > PUIDs. Now I am quite confused (and/or worried!) There seem to be a lot of
> > collisions in the PUIDs (or did I accidentally stumble on a coincidence?) Is
> > this expected? If so, shouldn't Picard use PUID+duration instead of PUID
> > alone?
>
>
> I'm not sure where the data in the MusicBrainz db is coming from in this
> case, but  PUIDs are already
> accounting for duration - no tracks that are farther apart than 10 seconds
> should ever have the same
> PUID.  (In other words, probably the original connection was borked).
>
> I know there were some bogus PUID connections from the original pass, and
> AFAIK they are still pending
> scrubbing - this may be one of those.
>

I would have understood mixing two releases of the same album, but not of
two different releases of the same artist (ovo / up) or of two artists PG /
U2!). I did not test Picard on this particular point, but I believe Picard
currently lets a distracted or careless user send about anything. So this is
bound to happen again. I think Picard should refuse to send PUIDs if the
timings differ too widely from the release's timings (the famous 5 seconds
limit), or at least, the user should get a dialog telling him there is a
problem and asking for his confirmation.

What did you mean by "from the original pass"?

-- 
Frederic Da Vitoria
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20061006/21a4bf3c/attachment.htm