#metabrainz

/

      • bitmap
        yeah
      • reosarevok
        And it would still need an extra confirmed boolean, then, I guess
      • bitmap
        you can use confirm.submit for that, actually
      • reosarevok
        What to use for the "next" button to submit the first form?
      • Well, "first form" :)
      • bitmap
        0/1 could work or you can use strings like review/submit
      • reosarevok
        You mean as in, give the submit a value and then read that?
      • bitmap
        yup
      • is that what you were asking, or were you asking about what string to display for the button?
      • reosarevok
        No, I was asking about how to make the same button do two things, I haven't done that before
      • So, yeah, basically that was what I was asking
      • I can try :)
      • bitmap
        cool
      • the merge form does something similar iirc except it passes the 'submit' variable it in the url (and the merge entities are stored in the session)
      • reosarevok
        Hmm. I can keep the post parameters, but I guess that means I have to load the editors twice then?
      • (since the post parameters are just usernames)
      • Maybe I'm being extra slow, not feeling super bright today :D
      • bitmap
        yeah, you will
      • Etua has quit
      • Etua joined the channel
      • reosarevok
        Well, I used == instead of eq and accidentally removed all editors while skipping the confirmation page
      • Good that I'm testing this properl :D
      • *properly
      • MRiddickW has quit
      • bitmap
        definitely not pro-perl
      • reosarevok
        bitmap: also, while I look at that. Can you see my questions on https://github.com/metabrainz/musicbrainz-serve... ? :)
      • I'd like to get that one fixed this week if I can
      • bitmap
        sure!
      • reosarevok
        And from https://github.com/metabrainz/musicbrainz-serve... I understand the next step for me is to make that return paginated?
      • bitmap
        yes I think I suggested making /ws/2/genre a browse endpoint
      • yvanzo
        bitmap: about running plackup in development mode, it doesn’t seem to need -r to actually load latest files under lib/MusicBrainz/Server/Entity/URL/
      • but using -r makes plackup to kill the server and restart it when changing these files.
      • I did test using these files because it was easier to see changes to the sidebar of NiN artist’s page.
      • bitmap
        you mean adding new files right? our perl probably use something like load_class on those
      • but restarting the server when they change needs -r, yes
      • yvanzo
        I mean changing the sidebar_name value in IMVDb.pm for example.
      • bitmap
        oh, ok. I'm guessing it's passed to load_class somewhere
      • yvanzo
        Found it: it's loaded from Data::URL using URL_SPECIALIZATIONS map.
      • Will add -r for classes loaded more conventionally, thanks.
      • alastairp
      • bitmap: maybe you too - I see you made some changes recently to the repo, but it looks like it was mostly just updating the postgres host
      • lucifer
        oh! finally
      • !m alastairp
      • BrainzBot
        You're doing good work, alastairp!
      • reosarevok
        bitmap: heh, it's complaining with
      • I guess that's related to postParameters somehow
      • alastairp
        not sure if I should interpret "oh finally" as "you took way too long to do that" or "great, new toys!"
      • reosarevok
        Por qué no los dos
      • lucifer
        the latter :)
      • bitmap
        alastairp: cool. I looked at doing that semi-recently but ran into some problems with the upgrade scripts, iirc, and thought it might be safer to do the upgrades release-by-release
      • alastairp
        bitmap: recent versions (and the version that I'm upgrading to) come with `install.sh` which appears to do everything. The only trick was using a custom docker-compose.prod.yml and sentry config file that points to the other db
      • bitmap
        looks like you already tested the upgrade process though, so I was possible doing something wrong or it got fixed
      • alastairp
        given that it's late in the day, I'll do the upgrade tomorrow. perhaps we can get through all of these "stopper releases" - should require 3 separate updates and then we'll be up to date
      • and then we can check out the new github bot after that. I've set up this version of sentry + github at MTG, I believe that we need to create our own github app for self-hosted sentry. I have instructions on how to do that
      • reosarevok
        bitmap: I expected it to be solvable with delete(%$post_params{'delete-users.submit'}); but seems not
      • lucifer
        alastairp: how about update the master branch of the repo to that tag. so that PR contains only meb specific commits to make it easier to review?
      • reosarevok
        Ooooh
      • It's because I have FormCsrfToken but also the parameters in postParameters, I assume
      • alastairp
        lucifer: yeah, that's a bit messy. Note that the PR is to merge it into `metabrainz`. I believe that the idea is that "master" matches the upstream, and `metabrainz` matches what we deploy
      • bitmap
        reosarevok: yeah that would make sense
      • I was wondering if you were sending multiple 'submit' values somehow
      • reosarevok
        Should I drop them from postParams, or drop FormCsrfToken ?
      • alastairp
        so to make it avoid the large PR we'd have to push that tag into "metabrainz"
      • reosarevok
        yvanzo, bitmap: should we release prod, btw?
      • I can do that, I guess, if we are finally ready :)
      • lucifer
        ah right, my suggestion is to change the base just for the review and then later change back to meb branch when ready
      • alastairp
        oh sure, I guess we can do that
      • bitmap
        reosarevok: what's postParams? are those in the component_props?
      • 'cause that error comes from FormHandler
      • reosarevok
        That's my $post_params = $c->req->body_params;
      • And they are passed to component_props and then to PostParameters
      • (sorry, it's "postParameters" when passed :) )
      • alastairp
        zas: hi, currently sentry (automatic bug reports) has its own exim container for sending mails: https://github.com/metabrainz/sentry-onpremise/..., in your opinion is this OK to continue with, or do you think it'd be good to set up the MeB exim_relay container?
      • one slight issue with this is that sentry is deployed with docker-compose (as it's based on the upstream project). I can convert start_exim_relay() to a docker-compose service configuration
      • bitmap
        reosarevok: and you're passing $post_params to form_submitted_and_valid, right? there must be a duplicate parameter in there somewhere
      • reosarevok
        Yes, as I said
      • I expect these are duplicating the csrf params
      • I just dunno whether I should just not have <FormCsrfToken form={form} /> on that second file since PostParameters has them already or what
      • bitmap
        sorry I missed that
      • reosarevok
        np
      • bitmap
        I'd remove csrf_session_key & csrf_token from postParams
      • alastairp
        bitmap: lucifer: ah - one other thing I came across - latest latest sentry uses docker-compose "Compose specification", which requires newer docker-compose binary and possibly docker server. Maybe for now we can just upgrade to the last release that supports the tools that we currently have on serge (I'll double-check what the latest we can upgrade to is)
      • lucifer
        makes sense
      • reosarevok
        Ok, let's test
      • alastairp
        (on the other hand, if sentry is the only thing on serge that uses docker-compose I guess we can just upgrade that too)
      • bitmap
        that's the only thing I know of that uses it on serge, but
      • I guess the docker server version would be the limiting factor
      • reosarevok: if there's no further blockers with the external links editor I think we probably should
      • lucifer
        alastairp: bitmap: yvanzo: reosarevok: this is the function, https://github.com/metabrainz/sir/blob/master/s... which is currently causing trouble. it gets the value of the `attribute` at the given path https://github.com/metabrainz/sir/blob/master/s... for an entity. this feels like magic to me, thoughts of adding a `serialize` method to each entity instead of this generic one?
      • bitmap
        I saw zas posted some UI improvements on the forums and I agree a lot with those but I guess we shouldn't delay the release any further
      • alastairp
        serge has engine version 18.09.5, from https://docs.docker.com/compose/compose-file/ it looks like we can use compose version 3.7 with that. I'll check what the last supported version of sentry is, then. Looks like we won't be able to upgrade to latest-latest, but at least some 2021 release would be good
      • reosarevok
        bitmap: hmm. It seems that this is making a mess because the newline-separated usernames are somehow getting mangled into one line :D
      • lucifer
        (the current issue is that it unable to resolve the `path` https://github.com/metabrainz/sir/blob/master/s... it is retrieving artist_links.artist instead of artist_links.artist.gid)
      • bitmap
        reosarevok: ooh. maybe 'cause they are put into an input instead of a textarea
      • zas
        alastairp: if it works, np
      • bitmap
        try tweaking textAreaRegExp in PostParameters.js
      • alastairp
        zas: it does work, and is part of the upstream distribution, so we'll continue to use that. thanks
      • yvanzo
        lucifer: Might your debugger setup be worth documenting for us to follow?
      • lucifer
        yvanzo: yes probably. i am currently using pycharm specific setup but will see if I can get something like vscode to work as well.
      • bitmap
        lucifer: how would serialize work - still accept a list of paths and call serialize again on each child?
      • alastairp
        lucifer: so instead of a declarative definition like https://github.com/metabrainz/sir/blob/master/s..., you think that we would make a method on an existing model? (modelext.CustomURL?)
      • bitmap
        s/list of paths/path list/ I guess
      • CoachBob joined the channel
      • lucifer
        bitmap: each entity knows its components. for eg: instead of passing artist_links.artist.gid as a path, we can directly hardcode and access the required property. like {'targetid': row.artist_links.artist.gid}
      • alastairp: yes.
      • yvanzo
        bitmap: should we move away from git2consul as we release next MBS?
      • bitmap
        yvanzo: yes I'd like to release that
      • Freso
        <BANG>
      • It’s Colour Blindness Awareness Monday!
      • I haven’t received any mailed in reviews, so…
      • People up for reviews today: zas, monkey, alastairp, Freso, yvanzo, lucifer, akshaaatt[m], bitmap, reosarevok, ruaok, CatQuest - and maybe akashgp09, ritiek, yyoung, jasondk? Please let me know ASAP if you want to give review (or if you’re listed and you don’t :))
      • zas: Go!
      • reosarevok
        Oh, if we do that, then I think I'll let you two release today ;)
      • zas
        Hey
      • I did usual stuff, maintenance, supervision, upgrades
      • but most notably I migrated influxdb + grafana to new server moroder
      • it cuts a lot of cost of having those on gcloud
      • alastairp
        one does not simply migrate influxdb
      • zas
        I also configured a bunch of new servers (we keep upgrading machines)
      • monkey
        alastairp: 👏
      • zas
        I also configured prometheus
      • the idea is to replace nagios with it (and its alert system)
      • peterhil has quit
      • for now, everything works, but I didn't deploy widely yet (so only few nodes are providing metrics to it)
      • I continue on this very soon
      • plus usual Picard PR reviews and merge
      • fin. alastairp ?
      • alastairp
        hi
      • I was on vacation for 2.5 weeks, it was very nice, although very warm in Barcelona
      • last week I slowly got back into the swing of things - checking emails, reviewing PRs. With lucifer we released some new versions of CB, including some long-time pending translations
      • I blogged about the AB similarity feature, nice to see people excited about that!
      • I worked on a translation of AB javascript code to typescript (thanks monkey for laying the groundwork for this with LB's configuration files)
      • I reported a bug in MB's oauth code to bitmap and fixed AB/LB/CB in some specific cases
      • bitmap
        (there's an MBS pr open for that too. thanks!)
      • alastairp
        I set up a PR to upgrade sentry to a newer version (this should go out tomorrow)
      • that's it. monkey next?
      • monkey
        Hello everyone !
      • I too was on vacation most of August, and so didn’t get around to doing much work. That was quite nice.
      • I did keep a close eye on akashgp09[m]’s project as he successfully finished the GSoC coding period.
      • After some thorough testing, the last fixes are being implemented now and the project should be ready for beta this week. Congrats !
      • Since I came back, I started slow with catching up with the many PR reviews that were awaiting.
      • This includes a new major feature: a notifications system for BookBrainz that prabal —last year’s GSoC student working with BB— started working on.
      • I also worked on a few small and self-contained tickets for LB and BB to warm up my coding muscles while improving some details, and started thinking about how to scrobble listens when using the ListenBrainz player.
      • That’s it for me! yyoung[m] go !
      • Freso
        Not sure if yyoung / yyoung[m] is still around for reviews. They used to always mail them in. :)
      • yvanzo
        yyoung[m] probably forgot to send a review by mail, go Freso!
      • Freso
        o/