#metabrainz

/

      • lucifer
        not familiar much with sentry's setup but in general looks good!
      • 2021-09-07 25038, 2021

      • reosarevok
        lucifer: the gids are there too :)
      • 2021-09-07 25040, 2021

      • lucifer
        let's deploy it!
      • 2021-09-07 25009, 2021

      • lucifer
        reosarevok: nice, we can use them to extract the relevant data.
      • 2021-09-07 25051, 2021

      • lucifer
        the works' indexing has been running for a few hours on bono now. in progress no errors yet.
      • 2021-09-07 25022, 2021

      • reosarevok
        Ok, let's see
      • 2021-09-07 25059, 2021

      • reosarevok
        I'm guessing we should have only a few entities of each type, plus all relevant artist types, work types, etc :)
      • 2021-09-07 25015, 2021

      • lucifer
        +1
      • 2021-09-07 25021, 2021

      • lucifer
        alastairp: testing out flask-cors, it does not seem to obey views's method parameters. it will just return the methods set in flask config. alternative is to use its @crossorigin decorator but then we are back to where we are now.
      • 2021-09-07 25023, 2021

      • lucifer
      • 2021-09-07 25054, 2021

      • lucifer
        the current OPTIONS request returns POST only but flask-cors will return GET, POST and whatever else is in its default
      • 2021-09-07 25036, 2021

      • lucifer
      • 2021-09-07 25018, 2021

      • alastairp
        lucifer: what is the query that you're making to get that request? `OPTIONS /endpoint HTTP/1.1`?
      • 2021-09-07 25008, 2021

      • alastairp
        and you're saying that it always returns "Access-Control-Allow-Methods: OPTIONS, GET, HEAD" regardless of what the methods in the blueprint route are?
      • 2021-09-07 25018, 2021

      • lucifer
        yes that.
      • 2021-09-07 25032, 2021

      • alastairp
        that's definitely wrong!
      • 2021-09-07 25045, 2021

      • lucifer
        the gist i posted is wrong will fix it. (its GET result instead of options)
      • 2021-09-07 25053, 2021

      • alastairp
        oh wait - I was reading the wrong output
      • 2021-09-07 25010, 2021

      • alastairp
        flask-cors doesn't send an Access-Control-Allow-Methods header?
      • 2021-09-07 25052, 2021

      • lucifer
        huh, its not sending it currently. i think i borked my setup. let me recheck it.
      • 2021-09-07 25010, 2021

      • lucifer
      • 2021-09-07 25011, 2021

      • alastairp
        it looks like Access-Control-Allow-Methods is optional, anyway
      • 2021-09-07 25041, 2021

      • lucifer
        ok fixed it. it only sends it when the Access-Control-Request-Method is present in the preflight request
      • 2021-09-07 25053, 2021

      • alastairp
        right. keep in mind that ACAM isn't the same as `Allow`: https://developer.mozilla.org/en-US/docs/Web/HTTP…
      • 2021-09-07 25006, 2021

      • lucifer
      • 2021-09-07 25032, 2021

      • alastairp
        so it's not saying "you can only make this type of request to this endpoint"
      • 2021-09-07 25013, 2021

      • lucifer
        yes. i saw that difference too. but Allow is also just a reference right?
      • 2021-09-07 25031, 2021

      • alastairp
        you can definitely try and make a request using a method that isn't in Allow
      • 2021-09-07 25011, 2021

      • lucifer
        right that's my understanding as well
      • 2021-09-07 25057, 2021

      • lucifer
        so ACAM is returning an extended list, including things which actually aren't supported. i think it'll work anyways in this case but just adds more confusion.
      • 2021-09-07 25034, 2021

      • lucifer
        i.e. not use Flask-CORS if we can't stop this behaviour
      • 2021-09-07 25034, 2021

      • Sophist-UK joined the channel
      • 2021-09-07 25024, 2021

      • lucifer
      • 2021-09-07 25035, 2021

      • lucifer
        i see its an open issue
      • 2021-09-07 25029, 2021

      • alastairp
        interesting
      • 2021-09-07 25046, 2021

      • alastairp
        so, any alternatives? Keep our own decorator? Make an extension of this model to support it?
      • 2021-09-07 25042, 2021

      • lucifer
        keep using our own decorator for now. i'll try to get the contribution in the Flask-CORS extension. it seems they do want it. if that doesn't work out, create our own extension.
      • 2021-09-07 25034, 2021

      • alastairp
        tried to do sentry upgrade, but it supposedly requires newer docker to what we have. Looking into it more to see if we can omit this requirement
      • 2021-09-07 25003, 2021

      • lucifer
        :/
      • 2021-09-07 25013, 2021

      • alastairp
        zas: is it possible to do docker upgrade on serge? Otherwise we might have to move sentry to a more recent server
      • 2021-09-07 25004, 2021

      • lucifer
        alastairp: regarding reviews/feedback, i also forgot to mention https://github.com/metabrainz/listenbrainz-server…
      • 2021-09-07 25027, 2021

      • alastairp
      • 2021-09-07 25048, 2021

      • alastairp
        this is the commit which did it. it's interesting, they're only reporting that it's "possibly" the solution to their problems
      • 2021-09-07 25057, 2021

      • alastairp
        let me see if I can skip the version check
      • 2021-09-07 25013, 2021

      • alastairp
        ok, thanks lucifer
      • 2021-09-07 25045, 2021

      • opal joined the channel
      • 2021-09-07 25053, 2021

      • monkey
        LB team, another question regarding listens submitted from brainzplayer: what value should additional_info.listening_from have? `listenbrainz`? `brainzplayer`? `https://listenbrainz.org/` ?
      • 2021-09-07 25046, 2021

      • lucifer
        depends. if in future, we allow BP to be embeddable then do we want it to BP for all sites?
      • 2021-09-07 25012, 2021

      • monkey
        There's also origin_url that could help differentiate
      • 2021-09-07 25005, 2021

      • lucifer
        right. but ideally i think there should be one field for source.
      • 2021-09-07 25052, 2021

      • monkey
        What is your suggestion then?
      • 2021-09-07 25021, 2021

      • lucifer
        listenbrainz
      • 2021-09-07 25055, 2021

      • monkey
        That would be my favorite option too
      • 2021-09-07 25056, 2021

      • alastairp
        is this the time to address the ticket that talks about the split between service and tool?
      • 2021-09-07 25013, 2021

      • lucifer
        that would make sense
      • 2021-09-07 25023, 2021

      • lucifer
        plus also standardizing a "source" field
      • 2021-09-07 25021, 2021

      • prabaaaaal joined the channel
      • 2021-09-07 25038, 2021

      • monkey
        That's LB-321 and LB-442
      • 2021-09-07 25039, 2021

      • BrainzBot
        LB-321: Report the tool that a user is sending listens from https://tickets.metabrainz.org/browse/LB-321
      • 2021-09-07 25039, 2021

      • BrainzBot
        LB-442: Support "source" as additional_info https://tickets.metabrainz.org/browse/LB-442
      • 2021-09-07 25046, 2021

      • monkey
        `source` definitely makes sense in my use case: `listening_from: "listenbrainz", source: "spotify"` for example
      • 2021-09-07 25052, 2021

      • reosarevok
        Heh, I see the WS does not include language names anyway, just zxx
      • 2021-09-07 25003, 2021

      • reosarevok
        So I didn't have to worry
      • 2021-09-07 25002, 2021

      • BrainzGit
        [musicbrainz-server] 14reosarevok opened pull request #2255 (03master…MBS-10902): MBS-10902: Stop changing zxx language name at the Perl level https://github.com/metabrainz/musicbrainz-server/…
      • 2021-09-07 25029, 2021

      • bitmap
        reosarevok: re MBS-960 I do agree using triggers to manage these is the correct/cleaner solution (and feel free to do that if you prefer) but I figured modifying the existing methods might help the PR get merged quicker, so
      • 2021-09-07 25030, 2021

      • BrainzBot
        MBS-960: Manage tags (rename/remove) https://tickets.metabrainz.org/browse/MBS-960
      • 2021-09-07 25040, 2021

      • reosarevok
        It probably would, but it feels a bit odd to have the query first find all the entity IDs, then pass it to withdraw or whatever so that it will deal with them one by one (actually, would that even need modifying? In theory it could just be called in a loop, I guess)
      • 2021-09-07 25048, 2021

      • reosarevok
        And the whole thing sucks so much anyway
      • 2021-09-07 25031, 2021

      • reosarevok
        Also, does the PR above seem sensible? IIRC you warned me back in the day that doing it like that would eventually become messy :p
      • 2021-09-07 25037, 2021

      • reosarevok
        So I'd like to avoid a second messy thing
      • 2021-09-07 25020, 2021

      • bitmap
        oh I see, the bug is from the language objects being reused I guess?
      • 2021-09-07 25056, 2021

      • reosarevok
        Yes
      • 2021-09-07 25001, 2021

      • bitmap
        not sure I remember what I warned about specifically :) I just remember saying we have too many of these [No lyrics] fallbacks spread out across the code
      • 2021-09-07 25037, 2021

      • reosarevok
        Me neither tbh, just remember being told to be wary :D
      • 2021-09-07 25032, 2021

      • alastairp
        monkey: right, I like that we can show both of them, then. Do we need to go through existing submission tools and see what they use?
      • 2021-09-07 25007, 2021

      • monkey
        There are basically three fields: "listening_from", "origin_url" and the suggested "source"
      • 2021-09-07 25028, 2021

      • monkey
        I believe it's limited to that
      • 2021-09-07 25035, 2021

      • yvanzo
        reosarevok, bitmap: do you want to decide the next milestone now?
      • 2021-09-07 25038, 2021

      • alastairp
        origin_url is the url of the mp3 file/youtube link/spotify recording?
      • 2021-09-07 25007, 2021

      • reosarevok
        yvanzo: I was going to say let's meet tomorrow but that shouldn't take super long, so we can do it now too if bitmap is up for it :)
      • 2021-09-07 25013, 2021

      • lucifer
      • 2021-09-07 25025, 2021

      • reosarevok
        !m lucifer
      • 2021-09-07 25025, 2021

      • BrainzBot
        You're doing good work, lucifer!
      • 2021-09-07 25043, 2021

      • yvanzo
        Congrats!
      • 2021-09-07 25030, 2021

      • reosarevok creates the milestone for now
      • 2021-09-07 25034, 2021

      • monkey
        alastairp: Correct. For example for this browser extension : https://github.com/web-scrobbler/web-scrobbler/bl…
      • 2021-09-07 25044, 2021

      • reosarevok
        We're hoping for releases every two weeks again, right, yvanzo? :)
      • 2021-09-07 25050, 2021

      • yvanzo
        right
      • 2021-09-07 25056, 2021

      • bitmap
        sure, let's do that
      • 2021-09-07 25008, 2021

      • yvanzo
        so next beta on Monday
      • 2021-09-07 25016, 2021

      • yvanzo
        next beta +freeze
      • 2021-09-07 25019, 2021

      • lucifer
        :D
      • 2021-09-07 25023, 2021

      • yvanzo
        (we can update beta earlier)
      • 2021-09-07 25032, 2021

      • reosarevok
        So the milestone should be due on Monday I guess?
      • 2021-09-07 25038, 2021

      • lucifer
        let's move this on to sir-test then?
      • 2021-09-07 25039, 2021

      • reosarevok
        Rather than on release day?
      • 2021-09-07 25045, 2021

      • yvanzo
        yes
      • 2021-09-07 25047, 2021

      • alastairp
        monkey: cool, that sounds good to me. the only other thing that came up on the forums about that was a version for `listening_from`. do you have any thoughts on that?
      • 2021-09-07 25048, 2021

      • reosarevok
        Ok
      • 2021-09-07 25006, 2021

      • reosarevok
        lucifer: should we run a full index and see if anything else broke now?
      • 2021-09-07 25028, 2021

      • monkey
        We could accept either a string or an object of shape {name, version} and any other field considered useful?
      • 2021-09-07 25035, 2021

      • lucifer
        sure, will take almost another day though
      • 2021-09-07 25054, 2021

      • monkey
        as you pointed out in that post though, `version` is a fluid concept
      • 2021-09-07 25059, 2021

      • reosarevok
        I'm pretty sure you have enough stuff to do in the meantime :) (and I'm planning to be off after this talk here anyway)
      • 2021-09-07 25007, 2021

      • lucifer
        works alone took 6 hrs.
      • 2021-09-07 25020, 2021

      • lucifer
        yup, triggered full reindex on bono.
      • 2021-09-07 25035, 2021

      • reosarevok
        Plus, nothing stopping us from trying to put some test data together at the same time, right? :)
      • 2021-09-07 25040, 2021

      • reosarevok
      • 2021-09-07 25008, 2021

      • yvanzo
        0 Open, that was quick!
      • 2021-09-07 25017, 2021

      • alastairp
        my thoughts about this structure is that we said "we're open to what you can submit, just send stuff and we'll see what sticks!" and then everyone just does exactly what we put in the docs. so I'd rather we just explicitly tell people what to put
      • 2021-09-07 25058, 2021

      • alastairp
        and given that we have years of listening_from in a speicfic format, I think I'd rather we just recommended a "listening_from_version" key
      • 2021-09-07 25010, 2021

      • reosarevok
      • 2021-09-07 25011, 2021

      • monkey
        Yeah, that's sensible
      • 2021-09-07 25022, 2021

      • reosarevok
        If you're ok with that, I'd put the edits still open there :)
      • 2021-09-07 25030, 2021

      • reosarevok
        Then maybe we can stop the special-casing too
      • 2021-09-07 25034, 2021

      • bitmap
        I was gonna suggest putting the edit ones yea
      • 2021-09-07 25020, 2021

      • yvanzo
        +1
      • 2021-09-07 25050, 2021

      • reosarevok
        Ok, I'll start with that then
      • 2021-09-07 25055, 2021

      • monkey
        I suppose we could somehow get the git hash of the current commit, send it to the front-end and use that for the listening_from_version of BrainzPlayer
      • 2021-09-07 25008, 2021

      • alastairp
        listening_from, listening_from_version, source, and origin_url, then? source and listening_from are strings, not URLs. That's backwards compatible with what we have now, too
      • 2021-09-07 25034, 2021

      • alastairp
        yeah. that value is available as an env variable in the webserver container
      • 2021-09-07 25054, 2021

      • reosarevok
        I also added one trivial change that I considered even sending straight to master (https://github.com/metabrainz/musicbrainz-server/…)
      • 2021-09-07 25005, 2021

      • reosarevok
        And one small thing that might help Freso (https://github.com/metabrainz/musicbrainz-server/…)
      • 2021-09-07 25024, 2021

      • reosarevok
        And I guess an aggregator is fine too, heh (https://github.com/metabrainz/musicbrainz-server/…)
      • 2021-09-07 25019, 2021

      • reosarevok
        And there's a bunch of other Bug ones that all could be nice to merge but I see nothing that is super urgent
      • 2021-09-07 25027, 2021

      • monkey
        +1 to the combination of `listening_from` `listening_from_version` `source` and `origin_url`.
      • 2021-09-07 25027, 2021

      • monkey
        I think that should cover all cases I've seen mentioned
      • 2021-09-07 25038, 2021

      • reosarevok
        So I'd rather you two picked stuff you feel you'll have time to review, or just leave them for next week :)
      • 2021-09-07 25009, 2021

      • alastairp
        thanks monkey! I can update the docs, and then we can close those 2 issues!
      • 2021-09-07 25021, 2021

      • reosarevok
        (like, this is a minimum hope, I guess? so if we end up reviewing more, the reviewer can always just add them to the milestone or whatever)
      • 2021-09-07 25022, 2021

      • bitmap
        I'll try to prioritize the bug ones but the rest you've linked + the edit ones seem fine in addition to that
      • 2021-09-07 25022, 2021

      • monkey
        *chef's kiss*
      • 2021-09-07 25037, 2021

      • reosarevok
        Neat
      • 2021-09-07 25041, 2021

      • bitmap
        yea of course I'll try to review more
      • 2021-09-07 25008, 2021

      • reosarevok
        I want to keep working on genre list and tag editing, but I doubt they'll be ready this week anyway :)
      • 2021-09-07 25009, 2021

      • BrainzGit
        [musicbrainz-docker] 14yvanzo merged pull request #203 (03master…fix_validator_image): Update the validator image https://github.com/metabrainz/musicbrainz-docker/…
      • 2021-09-07 25033, 2021

      • monkey
        alastairp: So I guess when submitting listens from BrainzPlayer I should fill `origin_url` with the current browser URL?
      • 2021-09-07 25046, 2021

      • alastairp
        what do you mean?
      • 2021-09-07 25059, 2021

      • alastairp
      • 2021-09-07 25016, 2021

      • monkey
        For example yes
      • 2021-09-07 25039, 2021

      • alastairp
      • 2021-09-07 25045, 2021

      • monkey
        I'f you're listening to my listens it would be /monkey. If listening to recent listens, lb.org/recent
      • 2021-09-07 25052, 2021

      • alastairp
        (or spotify:track:4mvOKMtq3o1q5U2N5SEcm0)
      • 2021-09-07 25002, 2021

      • alastairp
        though I guess that's a uri, not a url...
      • 2021-09-07 25016, 2021

      • monkey
        There's different fields for that *sometimes* (`spotify_id` for example)
      • 2021-09-07 25043, 2021

      • alastairp
        I don't think that the page that someone is on when they listen to a song is useful
      • 2021-09-07 25059, 2021

      • alastairp
        consider your example - that song isn't going to be on the page in a few hours time
      • 2021-09-07 25028, 2021

      • monkey
        I guess that's my point. That's currently how some submitters use it.
      • 2021-09-07 25030, 2021

      • alastairp
        or imagine that I'm on soundcloud, and the player has moved on to another track but I'm still on the page of the first song that I listened to
      • 2021-09-07 25048, 2021

      • alastairp
        right - which is why I asked if we had to go and look at existing tools and see what they use