#metabrainz

/

      • pbnoxious has quit
      • 2017-06-20 17118, 2017

      • ruaok
        k
      • 2017-06-20 17130, 2017

      • ruaok
        same outright replamcement.
      • 2017-06-20 17105, 2017

      • modwizcode has quit
      • 2017-06-20 17122, 2017

      • ruaok
        sent
      • 2017-06-20 17159, 2017

      • modwizcode joined the channel
      • 2017-06-20 17152, 2017

      • D4RK-PH0ENiX joined the channel
      • 2017-06-20 17113, 2017

      • suhas2go joined the channel
      • 2017-06-20 17139, 2017

      • support joined the channel
      • 2017-06-20 17147, 2017

      • support has quit
      • 2017-06-20 17108, 2017

      • Slurpee joined the channel
      • 2017-06-20 17108, 2017

      • Slurpee has quit
      • 2017-06-20 17108, 2017

      • Slurpee joined the channel
      • 2017-06-20 17147, 2017

      • ruaok
        cage replaced, zas.
      • 2017-06-20 17102, 2017

      • ruaok
        root passwd in hetzner group
      • 2017-06-20 17124, 2017

      • alastairp
        Quesito: https://community.metabrainz.org/t/are-there-diff… you should probably send a friendly note here saying that if they need the data for commerical or non-profit/academic use that they should contact you
      • 2017-06-20 17116, 2017

      • alastairp
        oh wait, this is the deezer guy :)
      • 2017-06-20 17115, 2017

      • ruaok
      • 2017-06-20 17121, 2017

      • ruaok
        I know you have some thoughts on this one.
      • 2017-06-20 17125, 2017

      • Quesito
        hmm. well his projects are for the moment his own --from my understnding --however yes. what ruaok said...
      • 2017-06-20 17137, 2017

      • alastairp
        ruaok: interesting
      • 2017-06-20 17150, 2017

      • alastairp
        isn't there a minimum song length in last.fm?
      • 2017-06-20 17156, 2017

      • alastairp
        might help us define the threshold
      • 2017-06-20 17120, 2017

      • ruaok
        I don't know.
      • 2017-06-20 17121, 2017

      • Clint
        audioscrobbler protocol forbids submitting a track <30s
      • 2017-06-20 17142, 2017

      • ruaok
        Clint: ah, that is very helpful.
      • 2017-06-20 17112, 2017

      • alastairp
        yeah, that's what I was thinking of
      • 2017-06-20 17123, 2017

      • Clint
        actually <=30s
      • 2017-06-20 17137, 2017

      • Clint
      • 2017-06-20 17151, 2017

      • alastairp
        ruaok: what about approaching this from a different direction
      • 2017-06-20 17109, 2017

      • alastairp
        because there's a huge amount of effort going into supporting these people on alpha
      • 2017-06-20 17131, 2017

      • alastairp
        - if you import from last.fm, you're not allowed to import from alpha. ever
      • 2017-06-20 17154, 2017

      • ruaok
        what is your motivation/reservation for suggesting this?
      • 2017-06-20 17112, 2017

      • alastairp
        I guess trying to make things easier for us :)
      • 2017-06-20 17115, 2017

      • ruaok
        I thought was to put this in place long enoughfor people to import from alpha, then get rid of it.
      • 2017-06-20 17128, 2017

      • ruaok
        well, yes.
      • 2017-06-20 17138, 2017

      • ruaok
        great idea, but now we have working code, thanks to iliekcomputers
      • 2017-06-20 17100, 2017

      • ruaok
        if this code works for the time being, in 3 months we can nuke alpha and then this code.
      • 2017-06-20 17102, 2017

      • alastairp
        do we have code in the lastfm importer to only import stuff scrobbled since the last time data was added to our account?
      • 2017-06-20 17110, 2017

      • ruaok
        not yet, no.
      • 2017-06-20 17122, 2017

      • ruaok
        ideally, I would like to hold off on that since that is a new feature.
      • 2017-06-20 17127, 2017

      • alastairp
        (that's another solution to work around the bad data in alpha. OK, let's go with the patch since we have code)
      • 2017-06-20 17139, 2017

      • ruaok
        we're still trying to get to a stable/scalable feature parity with alpha.
      • 2017-06-20 17101, 2017

      • ruaok
        got a moment to review the PR?
      • 2017-06-20 17108, 2017

      • ruaok
        easier than my counting one.
      • 2017-06-20 17109, 2017

      • alastairp
        yes, looking at it now
      • 2017-06-20 17116, 2017

      • ruaok
        woo
      • 2017-06-20 17127, 2017

      • alastairp
        min_time = users[user_name]['min_time']
      • 2017-06-20 17134, 2017

      • alastairp
        what is min_time ?
      • 2017-06-20 17159, 2017

      • ruaok
        unfold the code immediately above that segment.
      • 2017-06-20 17100, 2017

      • alastairp
        OK. users is a dict, because a bulk add can be many users?
      • 2017-06-20 17107, 2017

      • ruaok
        yes.
      • 2017-06-20 17122, 2017

      • alastairp
        and each username is a list of listens, plus the boundary timestamps of the oldest and newest listen
      • 2017-06-20 17127, 2017

      • ruaok
        we never make a guarantee about what is in a batch from rabbitmq.
      • 2017-06-20 17133, 2017

      • ruaok
        ding
      • 2017-06-20 17101, 2017

      • alastairp
        then for each user, you get all listens that fall within that range, and... only import the ones which didn't get returned?
      • 2017-06-20 17109, 2017

      • alastairp
        after doing some filtering for fuzzy timestamps
      • 2017-06-20 17126, 2017

      • alastairp
        does this importer know if these items are from alpha or from last.fm?
      • 2017-06-20 17138, 2017

      • alastairp
        because it'd be nice to replace bad->good timestamps
      • 2017-06-20 17141, 2017

      • ruaok
        it does not attempt to know.
      • 2017-06-20 17142, 2017

      • alastairp
        (yes yes, new feature)
      • 2017-06-20 17155, 2017

      • ruaok
        oh, that is a great observation.
      • 2017-06-20 17100, 2017

      • ruaok
        but....
      • 2017-06-20 17102, 2017

      • ruaok
        :(
      • 2017-06-20 17107, 2017

      • ruaok
        updates to influx/bigquery
      • 2017-06-20 17115, 2017

      • alastairp
        feh
      • 2017-06-20 17118, 2017

      • ruaok
        yeah.
      • 2017-06-20 17139, 2017

      • alastairp
        I still vote we get a really long stick and smack around the people who put real data in alpha
      • 2017-06-20 17101, 2017

      • ruaok
        one of the things I want to do is to make it clear that people should only use the alpha importer if they submitted new listens not already in last.fm
      • 2017-06-20 17109, 2017

      • alastairp
        yeah
      • 2017-06-20 17120, 2017

      • ruaok
        ocharles is in the UK. I got sticks long enough for htat.
      • 2017-06-20 17148, 2017

      • ruaok
        but I think iliekcomputers' PR is "good enough"
      • 2017-06-20 17157, 2017

      • ruaok
        to cover for the numpties in the world.
      • 2017-06-20 17106, 2017

      • alastairp
        the comments on the code seem reasonable, but it's too late in the day to verify the logic of the actual timestamp diff sweeper
      • 2017-06-20 17137, 2017

      • ruaok
        I'm happy enough with it to deploy to beta and then to engage ocharles for a real world test.
      • 2017-06-20 17130, 2017

      • alastairp
      • 2017-06-20 17135, 2017

      • alastairp
        what are all those 0000s in the timestamp?
      • 2017-06-20 17142, 2017

      • ruaok
        influx nonsense.
      • 2017-06-20 17148, 2017

      • alastairp
        😞(
      • 2017-06-20 17154, 2017

      • ruaok
        influx timestamps have to be in NANOseconds.
      • 2017-06-20 17121, 2017

      • ruaok
        unless you can specify the time precision, but that isn't possible in a lot of places.
      • 2017-06-20 17136, 2017

      • ruaok
        and this appending 8 "0" is found in a lot of example code.
      • 2017-06-20 17101, 2017

      • ruaok
        > do we do stuff with dates in this format from influx often? If so, it'd be nice to have this abstracted away somewhere
      • 2017-06-20 17112, 2017

      • ruaok
        that is a fair comment, but I'll make a bug for it for later.
      • 2017-06-20 17118, 2017

      • alastairp
        yeah, of course
      • 2017-06-20 17124, 2017

      • alastairp
        I just added another "later" comment
      • 2017-06-20 17129, 2017

      • ruaok
        saw it.
      • 2017-06-20 17146, 2017

      • ruaok
        I'm just aggressively trying to get to beta now.
      • 2017-06-20 17141, 2017

      • ruaok
        LB-178
      • 2017-06-20 17141, 2017

      • BrainzBot
        LB-178: Improve/abstract out influx date nonsense functions https://tickets.metabrainz.org/browse/LB-178
      • 2017-06-20 17138, 2017

      • ruaok
        !m Quesito
      • 2017-06-20 17138, 2017

      • BrainzBot
        You're doing good work, Quesito!
      • 2017-06-20 17123, 2017

      • ruaok
        > creating deltas is left as an exercise to the reader
      • 2017-06-20 17135, 2017

      • ruaok
        alastairp: do we need to do an "academic intervention"?
      • 2017-06-20 17129, 2017

      • alastairp
        send me a proposal document, and I'll tell you if I agree with it
      • 2017-06-20 17148, 2017

      • ruaok
        with a guaranteed turn around of time 6 weeks.
      • 2017-06-20 17105, 2017

      • alastairp
        optimistic!
      • 2017-06-20 17123, 2017

      • alastairp
        yeah, the code looks good in general, except artist_msid is optional
      • 2017-06-20 17126, 2017

      • alastairp
        hmm
      • 2017-06-20 17135, 2017

      • alastairp
        *mbids* are optional
      • 2017-06-20 17139, 2017

      • alastairp
        will we always have an msid?
      • 2017-06-20 17150, 2017

      • ruaok
        I don't recall.
      • 2017-06-20 17102, 2017

      • alastairp
        artist_name is required
      • 2017-06-20 17105, 2017

      • ruaok
        quite frankly we've done a poor job documenting what is required and what is not.
      • 2017-06-20 17112, 2017

      • alastairp
        so we'll have an msid
      • 2017-06-20 17129, 2017

      • mat_ has quit
      • 2017-06-20 17139, 2017

      • ruaok
        hmm, ok, now I'm going to dig to see where my confusion comes from
      • 2017-06-20 17113, 2017

      • alastairp
        so, the contents of an object in influx is different to the stuff that is sent by a client
      • 2017-06-20 17123, 2017

      • alastairp
        because it contains msids
      • 2017-06-20 17151, 2017

      • mat_ joined the channel
      • 2017-06-20 17151, 2017

      • mat_ has quit
      • 2017-06-20 17151, 2017

      • mat_ joined the channel
      • 2017-06-20 17123, 2017

      • ruaok
        yes and it is flattened as well.
      • 2017-06-20 17138, 2017

      • ruaok
      • 2017-06-20 17147, 2017

      • ruaok
        artist_msid is required.
      • 2017-06-20 17105, 2017

      • hernantz has quit
      • 2017-06-20 17118, 2017

      • ruaok
        thanks alastairp
      • 2017-06-20 17123, 2017

      • alastairp
        👍
      • 2017-06-20 17103, 2017

      • Quesito
        ruaok: technically I lied. I have a heart of cheese.
      • 2017-06-20 17108, 2017

      • Quesito
        :D
      • 2017-06-20 17119, 2017

      • ruaok
        far better.
      • 2017-06-20 17123, 2017

      • ruaok
        can't eat gold.
      • 2017-06-20 17113, 2017

      • hernantz joined the channel
      • 2017-06-20 17120, 2017

      • Quesito
        ruaok: is last bizdev email legit?
      • 2017-06-20 17112, 2017

      • Sophist_UK has quit
      • 2017-06-20 17102, 2017

      • Sophist-UK joined the channel
      • 2017-06-20 17126, 2017

      • Quesito
        si
      • 2017-06-20 17147, 2017

      • Sophist-UK has quit
      • 2017-06-20 17100, 2017

      • ruaok
        yeah, seems quite on the up and up.
      • 2017-06-20 17112, 2017

      • Sophist-UK joined the channel
      • 2017-06-20 17116, 2017

      • ruaok
        being open source and for/non-profit are two different things.
      • 2017-06-20 17124, 2017

      • ruaok
        you can be open source and for profit.
      • 2017-06-20 17140, 2017

      • ruaok
        and that is what they are trying to do, which is fine.
      • 2017-06-20 17106, 2017

      • Quesito
        right on
      • 2017-06-20 17153, 2017

      • saifulbkhan joined the channel
      • 2017-06-20 17148, 2017

      • Sophist-UK has quit
      • 2017-06-20 17115, 2017

      • Sophist-UK joined the channel
      • 2017-06-20 17134, 2017

      • Protab joined the channel
      • 2017-06-20 17135, 2017

      • Rotab has quit
      • 2017-06-20 17158, 2017

      • saifulbkhan has quit
      • 2017-06-20 17148, 2017

      • Mycah has quit
      • 2017-06-20 17112, 2017

      • saifulbkhan joined the channel
      • 2017-06-20 17134, 2017

      • lazka joined the channel
      • 2017-06-20 17126, 2017

      • dpmittal joined the channel
      • 2017-06-20 17134, 2017

      • lazka has quit
      • 2017-06-20 17104, 2017

      • Guest61528
        Gentlecat: ping
      • 2017-06-20 17146, 2017

      • ladyfrid1y is now known as ladyfriday
      • 2017-06-20 17154, 2017

      • ladyfriday has quit
      • 2017-06-20 17100, 2017

      • ladyfriday joined the channel
      • 2017-06-20 17143, 2017

      • saifulbkhan has quit