#metabrainz

/

      • Shubh joined the channel
      • BrainzGit
        [musicbrainz-server] 14mwiencek opened pull request #2540 (03master…selenium-search-setup-docs): Document running Selenium tests that use search https://github.com/metabrainz/musicbrainz-serve...
      • chinmay joined the channel
      • chinmay
        monkey: akshaaatt: just read the conversation about linting styles using prettier workflow. Was going to suggest linting styles too.
      • .less files in LB are messy. I was thinking of some solution that'll run the linter with an npm script similar to how eslint is set up.
      • hackerman has quit
      • odnes joined the channel
      • KevlarNoir joined the channel
      • Freso
        Man, kids these days really be out there making Backlinks and getting juice :\
      • Etua joined the channel
      • Shubh has quit
      • akshaaatt
        Hi chinmay! I'm working towards a solution for the .less files currently. But your idea does sound good to me.
      • chinmay
        Alright :)
      • Tell me if you need any help.. I came across stylelint
      • monkey
        I agree chinmay, the LB less files could use a bit of cleanup and organizing. That's basically all my doing 😅
      • MRiddickW joined the channel
      • Don't know stylelint, will have a look
      • Etua has quit
      • chinmay
        Hahaha 😁
      • Yes please :)
      • I'll try something out locally and update you
      • monkey
        Looks like there's lightly better support for SCSS than for LESS, but…
      • Also thanks for the BrainzPlayer PR
      • chinmay
        My pleasure :)
      • I'll check out that package
      • monkey
        And also also, happy 2000th pull request ListenBrainz !
      • 🎉
      • lucifer
        ooo 🎉🎉
      • chinmay
        YESS💥
      • BrainzGit
        [listenbrainz-server] 14amCap1712 opened pull request #2001 (03master…recs-improvement): Some misc improvements to recommendation processing and display https://github.com/metabrainz/listenbrainz-serv...
      • [listenbrainz-server] 14amCap1712 merged pull request #1997 (03master…master): Add `listenbrainz-mpd` submission client https://github.com/metabrainz/listenbrainz-serv...
      • mayhem
        atj: are you able to unblock IPs from the gateways?
      • lucifer: regarding #1998 (which was an awesome year, all around, but that is off topic)... I see what you did and it makes me...
      • want to complicate things. 🤦
      • I have this, perhaps incorrect, impression that there will be a flurry of activity around new tracks around friday.
      • lucifer
        i see. so rechecking old stuff may overwhelm the mapper?
      • mayhem
        it would be nice if we could do some sort of exponential back-off here -- check a track frequently when it is new, but less frequently as time goes on.
      • overloading the mapper is possible, but really hard to predict. I'm more concerned about people having to wait 2 weeks for a track to be remapped.
      • lucifer
        ah i see
      • mayhem
        e.g. I listen to a track. I realize its not in MB, I add it. then wait 2 weeks.
      • but in order to do a back-off we would need another piece of information to build the back-off.
      • tricky.
      • lucifer
        yeah we'd need listened_at in the table i think.
      • which is not as simple because the table is on msid and multiple people could have listened to it. or same person multiple times.
      • mayhem
        slightly OT question: Will the last_listened timestamp be added to statistics too? I think it should -- because stats will be a source of info for troi and I would want to know when a track was last listened and having it in the stats API would be ideal.
      • indeed, it is tricky.
      • lucifer
        sure thats doable
      • mayhem
        that would be most useful!
      • lucifer
        for recordings currently, for releases and artists will need to implement it but not hard.
      • mayhem
        I think recordings are the ones I care about the most. but the others could be useful, but not imprtant now.
      • lucifer
        makes sense
      • mayhem
        oh and I think I will try my hand at making a daily-jams-energy list and a daily-jams-chill list.
      • I think I will try to classify top level genres into energy and chill and then pick those tracks from recommended tracks.
      • lucifer
        how about building the backoff based on last_updated times. store all last_updated times in database. check after `number of times already checked days` or a multiple of it. so check on day 1, then day 2, day 4, day 8?
      • the daily jams energy and chill idea sounds good! :D
      • mayhem
        would we need to store all last_updated times or just last and previous-to-last?
      • if we have previous to last we can just double the difference between the two last timestamps...
      • actually more in reverse.
      • how about a check-again, timestamp?
      • lucifer
        thinking more maybe add another column of number of times checked? but previous to last also makes sense.
      • mayhem
        last_updated and check-again.
      • lucifer
        check again makes sense. so first time set it +1 day, after that keep doubling check again - last udpated time?
      • mayhem
        so, if we get a listen for a track that has no match, check if we are past check-again. If so, check again. if still not found, push check-again out by twice the difference before and move on.
      • yes!
      • lucifer
        yes makes sense
      • mayhem
        yeah, that feels right to me.
      • lucifer
        cool, i can work on adding this.
      • mayhem
        how should we proceed with #1998? close for now or accept for now?
      • lucifer
        lets accept for now and release, i am not sure the change will be done today so maybe let the mapper run for 2-3 days on it.
      • (but no strong opinions either way)
      • mayhem
        done.
      • lucifer
        👍
      • mayhem heads out to the office
      • BrainzGit
        [listenbrainz-server] 14amCap1712 merged pull request #1998 (03master…recheck-listens): Recheck unmatched incoming listens periodically https://github.com/metabrainz/listenbrainz-serv...
      • lucifer
        alastairp: hi! thoughts on releasing https://github.com/metabrainz/listenbrainz-serv... now and fixing issues if those come up later? it looks ready-ish to me also tested a couple of times on beta.
      • atj
        mayhem: possibly, can you PM me the IP and I'll check?
      • Shubh joined the channel
      • alastairp
        lucifer: sure, let's try it!
      • lucifer
        alastairp: thanks! i'll update the PR with latest master and then merge.
      • alastairp
        monkey: by the way, I've been told that there's a direct TGV from CDG to St Pierre-des-Corps, which solves most of the problems we encountered
      • monkey
        Well that's handy.
      • Yes to the CB review PR ! Let me know if you need any front-end help
      • alastairp
        although he has just suggested that I take a 6:15am flight on monday which will get me to the venue 5 minutes before they start. I think I'll politely decline and go on Sunday instead :-P
      • monkey: got a moment to give some feedback on a react question?
      • monkey
        Sure
      • Shoot 🔫
      • alastairp
      • you may have seen some similar code to this before - it's the stuff that uses Verovio to generate some svg text then we dangerouslySetInnerHTML it
      • I want to highlight some elements on the svg (the 2nd useEffect) but I found that sometimes the effect triggers just before the render happens, and so the elements don't exist. hence I hacked it together by putting it in a setInterval
      • I'm aware of course that directly modifying the dom after it's been rendered is a bit naughty, but I'm thinking of other ways to do it
      • two options I thought of: instead of modifying the element directly, render a <style> block which applies a style to each #id; or load the svg into a DomParser after it's generated and modify it before rendering the component (I think you and I may have talked about this idea before?)
      • monkey
        Option one sounds a bit cleaner. But I think it might be best to investigate why "sometimes the effect triggers just before the render happens" instead, if that would fix the issue cleanly
      • alastairp
        yeah. it was happening in the right order on my local machine, but when I deployed it to a slower server it got out of whack
      • monkey
        Hm, I don't have any better suggestion
      • alastairp
        ideally I'd actually want to draw a box around it anyway, which I need to look into. In that case I need to render it, identify the coordinates of the things that need a box around them, then render the boxes with those fixed coordinates
      • I have a bit of code for that already
      • monkey
        Question: why setInterval? Do you want to run this repeatedly?
      • alastairp
        because I don't know how js works? 🤷
      • monkey
        User setTimeout :)
      • User*
      • Use*
      • alastairp
        thank you!
      • monkey
        Damn it
      • Honestly the current solution doesn't seem sooooo bad.
      • As long as you're limiting it to presentational changes
      • alastairp
        yep, it's just a visual indication of "hey, here are the notes that you were interested in"
      • mayhem
        akshaaatt: ping me when you have a moment.
      • atj
        reosarevok: has anyone ever suggested that labels should support annotations?
      • reosarevok
        They... do?
      • Link is on the sidebar
      • mayhem
        reosarevok: aren't you supposed to be on vacation? 😠
      • reosarevok
        I am! I'm sitting by a fireplace in a small hut while I wait for the rain to stop :D
      • mayhem
        go play some video games then!
      • sheesh, doing vacations like me.
      • atj
        reosarevok: sorry :/
      • reosarevok
        haha, it's ok!
      • atj: were you expecting it to be part of the main editor like for release?
      • I can see how that's a bit confusing
      • atj
        yeah, that's why I was confused
      • I went to edit the label and looked for the annotation field
      • MRiddickW has quit
      • akshaaatt appears
      • akshaaatt
        pong mayhem
      • mayhem
        akshaaatt: we have received a couple of app ratings saying that the desktop does a much better job at matching tracks.
      • akshaaatt
        True
      • mayhem
        so, I wonder if we should try this endpoint in the app:
      • akshaaatt
        Interesting!
      • Seems convenient
      • mayhem
        given just the artist name and the recording name it tries to find the best entry on the server.
      • akshaaatt
        Let's test how quick I am to updating this
      • mayhem
        I think it has a lot of potential and if matching is an issue on the client, then this could be a useful improvement.
      • akshaaatt
        Sure!
      • mayhem
        thanks, I've been itching for someone to try this in this context.
      • akshaaatt
        Cool!
      • mayhem
        atj: thanks for that IP removal, that was the problem.
      • atj
        mayhem: no problem, I've messaged zas to let him know as I couldn't see any references to him adding the IPs manually
      • mayhem
        monkey: I now remembered why I didn't do release_tags at the time.
      • it required me to solve a HARD PROBLEM, so I left it. But then I had to solve the same hard problem when dialing in the cover art. and I didn't realize that I had solve the tags problem too.
      • fixing now.
      • akshaaatt
        Never seen such a sexy api endpoint, mayhem
      • alastairp
        lucifer: building schema-27 sample db now
      • lucifer
        👍 thanks!
      • mayhem
        akshaaatt: that is what I thought, but so far everyone has been "huh. ok".
      • thanks for picking this up. :)
      • alastairp
        mayhem: the thing is, it just works, and does it well. so it just melts into the background
      • if it broke all the time maybe we'd pay more attention to it ;)
      • mayhem
        lol, ok. ;)
      • alastairp
        mayhem: wolf is still on schema 26, should I go ahead and upgrade it?