MusicBrainz Summit/5

From MusicBrainz Wiki
< MusicBrainz Summit
Revision as of 03:34, 19 September 2012 by Ianmcorvidae (talk | contribs) (Ianmcorvidae moved page musicBrainz Summit/5 to MusicBrainz Summit/5 over redirect)
Jump to navigationJump to search
The printable version is no longer supported and may have rendering errors. Please update your browser bookmarks and please use the default browser print function instead.


Summit Notes

Thursday, November 20th 2003 (Dave and Rob in Bedford)

    • Switch data dumps to every other day to save bandwidth
    • Put replication in place and then switch full dumps to once a week
    • Possibly add another replication off site machine to do RDF dumps
    • Synchronize the data dumps/RDF dumps to align with replication packet serial numbers
  • Review of progress on the Advanced Relationships work
    • AdvancedRelationships coming along -- perhaps a week's work (system time) work left to beta.
    • Do a complete event model later
    • Release Events: (complete)
      • Create a new table that covers: release event (optional month/date), territory ISO two letter country code
      • * May optionally add release notes later if need arises
      • * Add a default two letter country code to preferences
    • Consider a general notes concept that allows us to attach notes to any entity
  • Linking to MB
    • Make the artist and release pages use MBIDs to allow weblogs and other pages to easy link to MB.
      • Perhaps create an internal caching scheme to not impact the database looking up slower ids
    • Create a JavaScript method for copying links to MB with icons
    • Consider creating a supercession table to allows forwarding of MBIDs in case releases get merged, deleted, etc.
      • Keep track of last used date and lookup counts to allow pruning this data
  • Discuss necessary changes for the mb_server for replication
    • Use separate user for replication and let this user write to the DB. Normal musicbrainz_user would have no write access on the slaves
    • Add a mirror sever setting to
    • Have the login page redirect to the main server.
    • Have the possibility of replicating sanitized moderator information.
    • Deny RDF submits for mirror servers
    • Check out what other pages need to redirect to the main server (reports)
  • Further replication tasks:
    • Schema version # (monotonically increasing, starting at 1)
    • Store current schema version # in
    • Require users to hand update from CVS initially if schema version # changes. Later we will add a more automated method for updating.

Friday November 21st 2003

In attendance:

  • Jamie Munro (host)
  • Dave Evans
  • Robert Kaye

Karma system

  • Not all contributions can be redeemed for all benefits. (eg. a cash contribution should not give moderation trust)
  • Contributions:
    • Code, testing, documentation -> leads to developer trust
    • Moderating, voting -> leads to moderator reputation/trust
    • Cash -> leads to cash reputation
    • Referrals via the HTML links to MB -> leads to promotion reputation/trust
  • Benefits
    • automod status can only be archived via mod/vote contributions
    • Advanced tagger features can be accessed by achieving a certain karma level
    • schwag/amazon credit (give out a percent of of $$$ taken in)
    • credit/visibility (visual cues, email@mb alias)
    • Core supporters (identify and credit the core people who make MB happen)


  • Use points not cash
  • Core supporters can give points for thinker or other non-tangible contributions
  • Use a weighting system to determine if people qualify for a certain benefit
  • Automod contributions use a sliding window (must maintain votes/mod to maintain status)
  • Cash donations unlock advanced tagger features for a fixed period of time.

Improve voting issues:

  • Finish the one moderation release edit feature
  • Number of votes required to pass a moderation is inversely proportional to quality of the moderator (good moderators will require few or no votes, bad/unknown moderators require more votes)
  • Automoderators can cast more votes than one (moderators can chose how many votes to cast)
  • When logged in, add an iframe to the artist/release pages that shows one moderation to vote on:
    • Show votes for this artist or if none, then
    • Show votes for artists in the subscribed list, or if none, then
    • Show votes for related artists, or if none, then
    • Show random votes

Other suggestions:

  • Jamie suggest that we need to create a mozilla search plugin a la MusicMoz (got to and search for music)
  • Jamie suggest that we should have a type of moderator that is specific to a band -- hopefully a band member. This moderator could have veto power over the changes made to the artist in question.

Saturday November 22nd 2003

In attendance:

Audioscrobbler/ notes:

  • Audioscrobbler to host own MB database mirror to handle plugin lookups
  • AS plans to integrate tagging features in AS plugins
  • MB tagger should do some statistical gathering to create a graph of tagger accuracy
  • Relationships: Artist to artist and release to release, no track to track because of resource issues. May become possible once MB is fully integrated
  • People to people matching profiles
  • Some moderation voting atop Audioscrobbler pages if required
  • for RJ and/or add ripping support to Tagger to aid the ripping process of
  • Special MB moderator should be created that submits moderations from AS/
  • How do we share logins?
    • At first AS will require regular MB logins
    • Later we may create a Web service for creating login
    • or share logins with as_username
  • Jamie will contact the Liberty Alliance about single login -- perhaps we can be an example case


  • Contains accurate info that doesn't disappear
  • Uses the same structure as dmoz
  • Contains reviews, releases, general info about artist
  • Whole database is a hierarchy.
  • Runs on donated hardware/software
    • requires a new more powerful server for mirror
  • CC license is a question to put to the community, uses DMOZ license currently

MusicMoz integration plan

  • Goal: Add MusicMoz links to the MB website and encourage MB users to participate in MusicMoz
  • initial match of artists and loading: provide artist name, sortname, mbid, releases and release ids to RF
  • for creating a new artist musicmoz sends artist, sortname, uuid, and musicmoz url to MB via the webservice
  • artists that are linked with musicmoz cannot be changed
    • if mb users want to edit the name then a note is shown and http POST is done to MM site to request the change
    • MM editors can edit artist names via the webs service
  • MB cannot delete artist if linked to MusicMoz
  • MusicMoz linked artists cannot be merged into another artist. Other way around is ok.

MusicMoz new webservice features:

  • AddArtist
    • Send: name, sortname, moz URL
    • Ret: MBID
  • EditArtist
    • Send: MBID, newname, newsortname
    • Ret: OK
  • PendingActionNotice
    • Send: text, mod id, mod link
    • ret: ok

Misc Notes:

  • Can we use FOAF to express musical interest (mail Dan Brickley and cc: RJ) (done)
  • Mail BPM database guy. Our official response is: Hmmm
  • Once the triple alliance (AS, MB, MM) gets going we should do a press release
  • May consider adding details of the alliance to the metabrainz pages.
  • ISMIR: We should submit paper about MB to let the ISMIR crowd know about MB.