#metabrainz

/

      • ola joined the channel
      • 2016-03-07 06710, 2016

      • Gentlecat
        2 classes with 1k in each, correct
      • 2016-03-07 06710, 2016

      • alastairp
        possibly, though I'm not sure
      • 2016-03-07 06726, 2016

      • alastairp
        perhaps he had much fewer. I know his recent ones were only 200 total
      • 2016-03-07 06726, 2016

      • ola is now known as Guest93077
      • 2016-03-07 06757, 2016

      • Gentlecat
        we really need that list of jobs :)
      • 2016-03-07 06732, 2016

      • Gentlecat
        I can try and make something, but should be doing other things
      • 2016-03-07 06755, 2016

      • petepy
        ruaok: cool, has anyone suggested using youtube history in way to assemble music profiles?
      • 2016-03-07 06713, 2016

      • ruaok
        not directly, no.
      • 2016-03-07 06716, 2016

      • alastairp
        Gentlecat: yeah, the PR is there, but it needs to be updated
      • 2016-03-07 06720, 2016

      • ruaok
        interesting thought though.
      • 2016-03-07 06736, 2016

      • Gentlecat
        isn't it a list of datasets?
      • 2016-03-07 06739, 2016

      • black-perl has quit
      • 2016-03-07 06710, 2016

      • zas
        ruaok: did you get in touch with Free Music Archive at some point ? https://github.com/murdos/musicbrainz-userscripts… (it is about importing FMA releases in MB)
      • 2016-03-07 06724, 2016

      • petepy
        yeah not sure where that falls in the priorities of the project. I'm hoping to apply for the gsoc, or at the very least do cool stuff with some of the metabrainz resources.
      • 2016-03-07 06701, 2016

      • alastairp
        Gentlecat: yes, it's "just a list", but I want filtering, and paging too
      • 2016-03-07 06711, 2016

      • alastairp
        in fact, the "just a list" is done
      • 2016-03-07 06746, 2016

      • Guest93077 has quit
      • 2016-03-07 06720, 2016

      • ariscop has quit
      • 2016-03-07 06722, 2016

      • WcktKlwn has left the channel
      • 2016-03-07 06711, 2016

      • ruaok
        zas: wasn't me.
      • 2016-03-07 06738, 2016

      • petepy_ joined the channel
      • 2016-03-07 06736, 2016

      • petepy has quit
      • 2016-03-07 06702, 2016

      • petepy joined the channel
      • 2016-03-07 06754, 2016

      • petepy_ has quit
      • 2016-03-07 06715, 2016

      • pk_ joined the channel
      • 2016-03-07 06722, 2016

      • pk_
        join
      • 2016-03-07 06724, 2016

      • reosarevok
        pk_: already here ;)
      • 2016-03-07 06708, 2016

      • shubham1337 has left the channel
      • 2016-03-07 06711, 2016

      • alastairp
        Gentlecat: hmm. I see your thing still hasn't finished
      • 2016-03-07 06720, 2016

      • alastairp
        I think this is the same bug that I remember seeing.
      • 2016-03-07 06732, 2016

      • alastairp
        there's sometimes an issue with the multithreading, and it just stops
      • 2016-03-07 06747, 2016

      • alastairp
        Gentlecat: I killed a few processes and it seems to have finished the first job, and now it's looking at another
      • 2016-03-07 06741, 2016

      • Gentlecat
        issue in gaia?
      • 2016-03-07 06745, 2016

      • alastairp
        yeah
      • 2016-03-07 06749, 2016

      • Gentlecat
        :(
      • 2016-03-07 06701, 2016

      • alastairp
        the bit which does multiprocessing
      • 2016-03-07 06754, 2016

      • alastairp
        hey. your mood test is pretty good!
      • 2016-03-07 06744, 2016

      • alastairp
        I would love to see results of running the classifier on your 30k ground truth
      • 2016-03-07 06708, 2016

      • Gentlecat
        you got a script for that?
      • 2016-03-07 06733, 2016

      • alastairp
        not really :(
      • 2016-03-07 06701, 2016

      • Gentlecat
        I can try doing it locally
      • 2016-03-07 06709, 2016

      • alastairp
      • 2016-03-07 06713, 2016

      • alastairp
        this is supposed to do it
      • 2016-03-07 06728, 2016

      • alastairp
        the idea is to create a `models` row for it
      • 2016-03-07 06700, 2016

      • alastairp
        but there is no easy way to take all mbids in a dataset, and only run the hl on those ones
      • 2016-03-07 06705, 2016

      • alastairp
        (instead of all items in the database)
      • 2016-03-07 06732, 2016

      • alastairp
        docs = db.data.get_unprocessed_highlevel_documents_for_model(model_id, includes)
      • 2016-03-07 06739, 2016

      • alastairp
        this was really slow, so what I ended up doing was creating a temporary table containing the mbids I wanted to compare against, and doing a join on to that table
      • 2016-03-07 06748, 2016

      • kosmasK has quit
      • 2016-03-07 06753, 2016

      • alastairp
        but that's a little hacky. so I think we need a better way of doing it
      • 2016-03-07 06757, 2016

      • Gentlecat
        btw, after you merge and deploy schema changes and all this stuff, don't forget about generated models
      • 2016-03-07 06708, 2016

      • Gentlecat
        the actual files, I mean
      • 2016-03-07 06722, 2016

      • alastairp
        what do you mean?
      • 2016-03-07 06725, 2016

      • alastairp
        the .history files?
      • 2016-03-07 06729, 2016

      • Gentlecat
        yep
      • 2016-03-07 06739, 2016

      • alastairp
        are they stored in the checked out directory?
      • 2016-03-07 06747, 2016

      • alastairp
        so I'll have to copy them from my ~ to ~acousticbrainz?
      • 2016-03-07 06749, 2016

      • alastairp
        ok
      • 2016-03-07 06706, 2016

      • Gentlecat
        to one that is specified in the config file
      • 2016-03-07 06710, 2016

      • alastairp
        ah, ok
      • 2016-03-07 06711, 2016

      • alastairp
        cool
      • 2016-03-07 06723, 2016

      • Gentlecat
        you can actually point them to the same place
      • 2016-03-07 06733, 2016

      • alastairp
        there's also no way of comparing computed hl files to the ground truth of a dataset
      • 2016-03-07 06738, 2016

      • alastairp
        that's kind of the "next" thing to do
      • 2016-03-07 06744, 2016

      • alastairp
        though everything is the next thing to do
      • 2016-03-07 06715, 2016

      • demonimin has quit
      • 2016-03-07 06740, 2016

      • alastairp
        https://github.com/metabrainz/acousticbrainz-serv… ahh this also expects to have the train/test splits
      • 2016-03-07 06752, 2016

      • alastairp
        which your dataset won't have because you didn't choose the filtering option
      • 2016-03-07 06700, 2016

      • alastairp
        this is a bit of a mess
      • 2016-03-07 06709, 2016

      • demonimin joined the channel
      • 2016-03-07 06729, 2016

      • demonimin has quit
      • 2016-03-07 06756, 2016

      • demonimin joined the channel
      • 2016-03-07 06709, 2016

      • github joined the channel
      • 2016-03-07 06709, 2016

      • github
        [listenbrainz-server] pinkeshbadjatiya opened pull request #70: Fix [LB-13] | Export Listens (master...export-listens) https://github.com/metabrainz/listenbrainz-server…
      • 2016-03-07 06709, 2016

      • github has left the channel
      • 2016-03-07 06750, 2016

      • armalcolite
        ^work was almost done. could not resist. :)
      • 2016-03-07 06744, 2016

      • Nyanko-sensei joined the channel
      • 2016-03-07 06749, 2016

      • Nyanko-sensei has quit
      • 2016-03-07 06751, 2016

      • D4RK-PH0ENiX has quit
      • 2016-03-07 06715, 2016

      • D4RK-PH0ENiX joined the channel
      • 2016-03-07 06745, 2016

      • rahulr joined the channel
      • 2016-03-07 06723, 2016

      • github joined the channel
      • 2016-03-07 06723, 2016

      • github
        [listenbrainz-server] pinkeshbadjatiya opened pull request #71: page-count-bug-scraper (master...page-count-bug) https://github.com/metabrainz/listenbrainz-server…
      • 2016-03-07 06723, 2016

      • github has left the channel
      • 2016-03-07 06731, 2016

      • Gentlecat
        Freso: looks like we need more categories (https://community.metabrainz.org/t/listenbrainz-a…)
      • 2016-03-07 06706, 2016

      • github joined the channel
      • 2016-03-07 06706, 2016

      • github
        [listenbrainz-server] pinkeshbadjatiya opened pull request #72: Fix [LB-68] Memory Leak | Scraper (master...memory-leak-scraper) https://github.com/metabrainz/listenbrainz-server…
      • 2016-03-07 06706, 2016

      • github has left the channel
      • 2016-03-07 06736, 2016

      • pk_ has quit
      • 2016-03-07 06751, 2016

      • CallerNo6 joined the channel
      • 2016-03-07 06721, 2016

      • loujine joined the channel
      • 2016-03-07 06707, 2016

      • alastairp
        armalcolite: cool, thanks for the patches
      • 2016-03-07 06750, 2016

      • kanha has quit
      • 2016-03-07 06700, 2016

      • armalcolite
        alastairp: np.
      • 2016-03-07 06725, 2016

      • armalcolite
        can u suggest any other issue?
      • 2016-03-07 06758, 2016

      • armalcolite
        planning to go for 'case sensitive username in LB'
      • 2016-03-07 06701, 2016

      • alastairp
        not right now, I'm busy with other stuff
      • 2016-03-07 06723, 2016

      • alastairp
        oh, cool. I think we had a patch for that in AcousticBrainz too - https://github.com/metabrainz/acousticbrainz-serv…
      • 2016-03-07 06723, 2016

      • alastairp
        I think ruaok mentioned it a bit earlier today - listenbrainz has been in the background for us for a few months. with more people interested in it we will try and put more time towards working on the next plans
      • 2016-03-07 06733, 2016

      • alastairp
        you can see that there are already a number of open pull requests from last year
      • 2016-03-07 06741, 2016

      • alastairp
        and the tickets remaining on Jira
      • 2016-03-07 06714, 2016

      • alastairp
        I see your lastfm account is very new :)
      • 2016-03-07 06731, 2016

      • alastairp
        did you know about last.fm before you looked at listenbrainz? it has had a very long history
      • 2016-03-07 06753, 2016

      • petepy
        i'm a little confused how this chat works
      • 2016-03-07 06707, 2016

      • alastairp
        one of our goals is to replicate the core song recording functionality (called scrobbling in last.fm, 'listening' in LB)
      • 2016-03-07 06755, 2016

      • armalcolite
        yeah. knew it. and dont use it muh
      • 2016-03-07 06707, 2016

      • armalcolite
        *much.
      • 2016-03-07 06718, 2016

      • kanha joined the channel
      • 2016-03-07 06726, 2016

      • alastairp
        so this is what most of the tickets on jira are about, but also the projects that we have proposed for SoC are part of the next things we want to work on to make it feature compatible with the main last.fm functionality
      • 2016-03-07 06732, 2016

      • petepy
        last.fm no, not irc either. this chat is really messing with my head
      • 2016-03-07 06709, 2016

      • alastairp
        so one thing that you could do is look at last.fm some more, and see how our proposed listenbrainz features fit in with replicating this functionality
      • 2016-03-07 06731, 2016

      • armalcolite
        can u elaborate "core recording functionality" ?
      • 2016-03-07 06700, 2016

      • alastairp
        I mean scrobbling
      • 2016-03-07 06721, 2016

      • armalcolite
        submitting the current playing song?
      • 2016-03-07 06723, 2016

      • alastairp
        we don't like to use the word scrobble when talking about listenbrainz, because this is a term specifically for last.fm
      • 2016-03-07 06726, 2016

      • alastairp
        yes, that
      • 2016-03-07 06741, 2016

      • armalcolite
        yeah i read wiki for scrobbling
      • 2016-03-07 06742, 2016

      • alastairp
        unfortuately there is no simple word to mean that (except for scrobble)...
      • 2016-03-07 06734, 2016

      • armalcolite
        will try to get a clear idea of how last.fm works
      • 2016-03-07 06707, 2016

      • alastairp
        last.fm also has a lot of other functionality (friends, following, likes, tagging, etc)
      • 2016-03-07 06714, 2016

      • alastairp
        which we are not interested in replicating
      • 2016-03-07 06752, 2016

      • petepy
        yeah i sort of see that as an important part of the whole scrobbing thing, just like in spotify/facebook.
      • 2016-03-07 06701, 2016

      • Gentlecat
        the focus is probably on presenting listens, analyzing them
      • 2016-03-07 06724, 2016

      • alastairp
        one of the main focuses of metabrainz is to provide high quality data to anyone who wants to use it
      • 2016-03-07 06729, 2016

      • armalcolite
        ok. we are concerned about data.
      • 2016-03-07 06749, 2016

      • alastairp
        so there is definitely value in using this type of data in the same way as last.fm does
      • 2016-03-07 06705, 2016

      • alastairp
        but that's not a core requirement for metabrainz. we will let someone else do it
      • 2016-03-07 06704, 2016

      • petepy
        hypothetically would a media player with built in listening to metabrainz be useful?
      • 2016-03-07 06722, 2016

      • petepy
        or just a plugin for existing media players or web players'
      • 2016-03-07 06734, 2016

      • alastairp
        there are already a small handful of plugins
      • 2016-03-07 06742, 2016

      • alastairp
        not for many players though
      • 2016-03-07 06718, 2016

      • alastairp
        you can see that one of our proposed projects is to make an api which is compatible with last.fm's api so that existing players can submit data directly
      • 2016-03-07 06725, 2016

      • alastairp
        in fact, this is what libre.fm also does
      • 2016-03-07 06752, 2016

      • armalcolite
        i had the same project in mind. :)
      • 2016-03-07 06700, 2016

      • armalcolite
        like a proxy.
      • 2016-03-07 06704, 2016

      • alastairp
        exactly
      • 2016-03-07 06714, 2016

      • armalcolite
        users submit to use, we submit to lastfm.
      • 2016-03-07 06738, 2016

      • alastairp
        perhaps. we're not sure if we should submit to last.fm as well
      • 2016-03-07 06749, 2016

      • armalcolite
        :p
      • 2016-03-07 06704, 2016

      • armalcolite
        just replicating the api would be easy.
      • 2016-03-07 06715, 2016

      • armalcolite
        but we should work on improving it
      • 2016-03-07 06740, 2016

      • armalcolite
        or just replicating is enough ?
      • 2016-03-07 06744, 2016

      • alastairp
        in fact we also have a starting: https://github.com/Jonty/scrobbleproxy
      • 2016-03-07 06715, 2016

      • armalcolite
        good work. can surely use it.
      • 2016-03-07 06755, 2016

      • kepstin
        armalcolite: no need to improve the last.fm api; it's only there for compatibility. listenbrainz has its own native api where improvements can be made
      • 2016-03-07 06709, 2016

      • darwin
        armalcolite: I've imagined a pass through proxy that sniffs for submissions *to last.fm* and proxies them
      • 2016-03-07 06727, 2016

      • darwin
        armalcolite: because otherwise, spotify scrobbles, hypem scrobbles, etc. are lost?
      • 2016-03-07 06738, 2016

      • armalcolite
        kepstin: ok.
      • 2016-03-07 06741, 2016

      • kepstin
        darwin: the traditional way to redirect scrobbles to an alternate service for apps that don't support it natively is to edit the hosts file to redirect the audioscrobbler submission server to a different ip