#metabrainz

/

      • dseomn has left the channel
      • 2021-07-21 20208, 2021

      • dseomn joined the channel
      • 2021-07-21 20257, 2021

      • yyoung
        Sorry that my PR has caused an argument, I did feel the difficulties in improving an exisitng component considering people's custom, and I agree to postpone it a little bit for more feedbacks
      • 2021-07-21 20243, 2021

      • wargreen joined the channel
      • 2021-07-21 20202, 2021

      • yyoung
        Last night I took a long time trying to rebase another PR of mine (#2114) on master, but haven't had it done, maybe that PR would be more accepted by users, so we may also merge that for next release, continuing working on top of it, and introduce the popover when it became essential
      • 2021-07-21 20238, 2021

      • yyoung
        CatQuest: The 'x' icon should be on the same line with other icons, but it doesn't seem like that in your screenshot, could you provide more details so I can fix it?(e.g. your browser and version)
      • 2021-07-21 20244, 2021

      • yyoung
        Also the extra click issue is kind of trivial if you thinks about it carefully, as yvanzo said, you had to click on the input box before editing, while now you have to click on the edit icon, and then the input box is focused so you can start editing; If you're pasting a link, you can still edit until it blurred, which then requires a click to get it focused both before and now.
      • 2021-07-21 20253, 2021

      • d4rk joined the channel
      • 2021-07-21 20209, 2021

      • d4rkie has quit
      • 2021-07-21 20242, 2021

      • [1997kB] has quit
      • 2021-07-21 20249, 2021

      • [1997kB] joined the channel
      • 2021-07-21 20252, 2021

      • agatzk has quit
      • 2021-07-21 20224, 2021

      • agatzk joined the channel
      • 2021-07-21 20238, 2021

      • lucifer
        akshaaatt[m]: thanks. i'll take a look soon.
      • 2021-07-21 20248, 2021

      • ritiek joined the channel
      • 2021-07-21 20203, 2021

      • yvanzo
        mo’’in’
      • 2021-07-21 20232, 2021

      • BrainzGit
        [listenbrainz-server] 14jdaok opened pull request #1559 (03master…pin-timeline-backend): Pin timeline backend https://github.com/metabrainz/listenbrainz-server…
      • 2021-07-21 20209, 2021

      • yvanzo
        yyoung: Don’t be sorry, this fuss is not your fault. There is no need to revert your work for now. Please continue rebasing the other PR but keep a copy of your branch before the rebase.
      • 2021-07-21 20236, 2021

      • yyoung
        Sure
      • 2021-07-21 20200, 2021

      • yvanzo
        yyoung: I have to focus on getting search indexes back on, so I won’t be much available for anything else (but emergencies) today. Let’s have our weekly meeting tomorrow? I will be available starting from 7am UTC.
      • 2021-07-21 20208, 2021

      • yyoung
        yvanzo: Sure, that's July 22nd 7:00 UTC, right?
      • 2021-07-21 20245, 2021

      • yvanzo
        Right!
      • 2021-07-21 20243, 2021

      • yyoung
        OK, see you then!
      • 2021-07-21 20206, 2021

      • yvanzo
        Thank you.
      • 2021-07-21 20215, 2021

      • BrainzGit
        [musicbrainz-server] 14reosarevok opened pull request #2179 (03master…MBS-11798): MBS-11798: Disallow Instagram internal links https://github.com/metabrainz/musicbrainz-server/…
      • 2021-07-21 20237, 2021

      • ruaok
        lol @grouchy lB bot
      • 2021-07-21 20249, 2021

      • ruaok
        mooin!
      • 2021-07-21 20253, 2021

      • lucifer
        morning!
      • 2021-07-21 20247, 2021

      • lucifer
        ruaok: the dump ages check failed yesterday. from the error message it seems because it found the mapping dump instead.
      • 2021-07-21 20209, 2021

      • ruaok
        yeah, I'll have to adjust the checks.
      • 2021-07-21 20201, 2021

      • lucifer
        👍
      • 2021-07-21 20227, 2021

      • Etua joined the channel
      • 2021-07-21 20224, 2021

      • ruaok
        lucifer: I've removed the MBID dumps from the FTP/ftp/backup volumes for now. the dump check should quiet down until we merge the PR that will soon have an improved dump age check.
      • 2021-07-21 20238, 2021

      • lucifer
        +1 thanks
      • 2021-07-21 20213, 2021

      • Etua has quit
      • 2021-07-21 20254, 2021

      • Lotheric_ has quit
      • 2021-07-21 20216, 2021

      • Lotheric joined the channel
      • 2021-07-21 20237, 2021

      • akshaaatt[m]
        lucifer: I want to finalize things for the production release by tomorrow from my end.
      • 2021-07-21 20216, 2021

      • akshaaatt[m]
        2 changes will be incorporated before that, 1) Tagger UI will be taken care of mildly as of now to suit a finished look for release. 2) Assets attribution will be finalized.
      • 2021-07-21 20230, 2021

      • akshaaatt[m]
        If you have any more things in mind, do let me know.
      • 2021-07-21 20248, 2021

      • alastairp
        morning
      • 2021-07-21 20259, 2021

      • ruaok
        moin
      • 2021-07-21 20253, 2021

      • alastairp
        have we used the "--cpus" flag anywhere in the MeB stack to limit the number of CPUs available to a container?
      • 2021-07-21 20220, 2021

      • alastairp
        I had a quick search in docker-server-configs and see that I used it earlier this year to limit the CPUs for jenkins
      • 2021-07-21 20212, 2021

      • alastairp
        as I understand, the effect is to expose only x cpus to the container. but I don't know what the practical effects of that are. can the container only "see" x cpus, or does docker/kernel prevent it from using x00% usage?
      • 2021-07-21 20229, 2021

      • akashgp09 joined the channel
      • 2021-07-21 20254, 2021

      • ruaok
        alastairp: I haven't but I tend to think we should have the programs in the container do this themselves.
      • 2021-07-21 20259, 2021

      • ruaok
        what needs limiting?
      • 2021-07-21 20230, 2021

      • alastairp
        gaia
      • 2021-07-21 20243, 2021

      • alastairp
        it just looks for how many cores are availalbe and uses _all_ of them :)
      • 2021-07-21 20202, 2021

      • alastairp
        no config option (although I just opened a PR for it there)
      • 2021-07-21 20209, 2021

      • monkey
        Just like in real life, using all of Gaia's resources. How fitting
      • 2021-07-21 20219, 2021

      • alastairp
        yeah
      • 2021-07-21 20252, 2021

      • alastairp
        OK, I think for now then we have a few options. 1) while we wait for gaia to learn how to share, I'll use --cpus on the dataset training container. 2) when this flag is added, we'll switch that that, 3) we'll add this functionality to the sklearn training too, so it'll work the same when we switch it over
      • 2021-07-21 20202, 2021

      • atj
        alastairp: I'd be surprised if it's possible to restrict the number of CPUs the container actually sees
      • 2021-07-21 20218, 2021

      • atj
        Especially given you can specify 0.5 CPUs for a container
      • 2021-07-21 20227, 2021

      • alastairp
        atj: I found setaffinity, but that's not really what I want either
      • 2021-07-21 20235, 2021

      • alastairp
        atj: yeah, that was my exact thoughts
      • 2021-07-21 20258, 2021

      • atj
        I expect Gaia is using one of the system calls to detect the number of calls, which is probably using the same info as /proc/cpuinfo
      • 2021-07-21 20205, 2021

      • atj
        s/calls/CPUs/g
      • 2021-07-21 20217, 2021

      • alastairp
        I don't mind if the cgroup prevents dockre from using more than 2.5 CPUs, but if multiprocessing.num_cpus() (python) still says that there are 16 cores available then it's still going to start 16 threads
      • 2021-07-21 20232, 2021

      • alastairp
        and I'd much rather start just 4
      • 2021-07-21 20249, 2021

      • BrainzGit
        [listenbrainz-server] 14jdaok opened pull request #1560 (03master…pin-timeline-frontend): Add pin event to TimelineEvent component https://github.com/metabrainz/listenbrainz-server…
      • 2021-07-21 20200, 2021

      • alastairp
        so yes, our software is definitely buggy to not allow this to be manually configured (fixed in https://github.com/MTG/gaia/pull/116)
      • 2021-07-21 20230, 2021

      • atj
      • 2021-07-21 20213, 2021

      • alastairp
        lucifer: btw, turns out that about 6 months ago we added py3 support to gaia anyway :)
      • 2021-07-21 20233, 2021

      • ruaok
        alastairp: ok that approach makes sense.
      • 2021-07-21 20218, 2021

      • ruaok
        param: can you please have a look at more of the open PRs from jasondk? And I think there are some PRs that you approved that are safe to merge -- it would be nice to work down the list of open PRs a bit.
      • 2021-07-21 20232, 2021

      • ruaok
        sorry that so much of this project ends up being react work. :(
      • 2021-07-21 20203, 2021

      • alastairp
        atj: I guess I could always LD_PRELOAD and override sysconf(_SC_NPROCESSORS_ONLN) ;)
      • 2021-07-21 20256, 2021

      • ruaok
      • 2021-07-21 20225, 2021

      • alastairp
        groan
      • 2021-07-21 20231, 2021

      • atj
        alastairp: indeed you could, but god would kill a kitten everytime you started gaia
      • 2021-07-21 20255, 2021

      • param
        ruaok: on it today!
      • 2021-07-21 20207, 2021

      • param getting pfizered today
      • 2021-07-21 20215, 2021

      • alastairp
        🎉
      • 2021-07-21 20241, 2021

      • alastairp
        ruaok: I switched to my linux desktop for development yesterday. it's so much faster than the laptop
      • 2021-07-21 20207, 2021

      • alastairp
        the only thing that's annoying me is all of my now-ingrained keyboard shortcuts
      • 2021-07-21 20225, 2021

      • ruaok
        param: great on both accounts! 💉
      • 2021-07-21 20228, 2021

      • ruaok
      • 2021-07-21 20235, 2021

      • atj
        could you not add a max_threads option to the configuration file loaded by ClassificationTaskManager?
      • 2021-07-21 20247, 2021

      • alastairp
        ruaok: 😍
      • 2021-07-21 20253, 2021

      • alastairp
        let me try that
      • 2021-07-21 20254, 2021

      • ruaok
        kinto remaps your keys to mimick a mac. I use it on both my linux boxes, works great.
      • 2021-07-21 20209, 2021

      • atj
        then just default to cpu_count() if the option isn't set
      • 2021-07-21 20222, 2021

      • ruaok
        which desktop are you using? Pop OS! has been fantastic -- feels much like a mac, but with other improvements.
      • 2021-07-21 20224, 2021

      • alastairp
        atj: yeah, I just added the functionality to do that. Just waiting for my coworkers to merge it
      • 2021-07-21 20236, 2021

      • ruaok
        I'm super happy and kinda grumpy about using my macbookpro for the next 6 weeks.
      • 2021-07-21 20242, 2021

      • ruaok goes to forage for food
      • 2021-07-21 20244, 2021

      • alastairp
        unfortunately we have some other problems, because AB currently uses an older version of gaia with py 2.7, and since then we've updated gaia to support py3 only (and released a 0.0.1 update on it. we should have at least updated the version by 0.1 when we made that backwards incompatible change :(
      • 2021-07-21 20258, 2021

      • atj
        what's slower on the Mac?
      • 2021-07-21 20200, 2021

      • alastairp
        however, there are some ways that we should be able to backport this setting
      • 2021-07-21 20207, 2021

      • alastairp
        atj: mostly docker inside virtualisation
      • 2021-07-21 20223, 2021

      • alastairp
        and spanish summer, causing it to overheat as soon as it sees some hard work
      • 2021-07-21 20236, 2021

      • alastairp
        therefore limiting 50% of available cpu cores
      • 2021-07-21 20246, 2021

      • alastairp
        (2019 i9 16" MBP)
      • 2021-07-21 20254, 2021

      • alastairp
        ruaok: currently still using ubuntu here
      • 2021-07-21 20210, 2021

      • alastairp
        ruaok: one thing that I was experimenting with was tailscale + docker remote-context-over-ssh
      • 2021-07-21 20216, 2021

      • atj
        hah, I've got the same model. Was thinking last night about how how it always was
      • 2021-07-21 20222, 2021

      • atj
        *hot
      • 2021-07-21 20233, 2021

      • alastairp
        so you run "docker-compose x" on the mac, and it executes on the remote machine
      • 2021-07-21 20235, 2021

      • atj
        We're currently experiencing some Spanish weather at the moment
      • 2021-07-21 20256, 2021

      • alastairp
        unfortunately auto-mounting code volumes doesn't work very well, so I gave up on that solution
      • 2021-07-21 20200, 2021

      • ruaok
        atj: tapas with some chance to sangria?
      • 2021-07-21 20236, 2021

      • alastairp
        atj: but yeah, for a long time I just "assumed" that a build took this long. Then I remembered that docker on a mac is virtualised anyway, with limited CPU and memory on top of that
      • 2021-07-21 20246, 2021

      • alastairp
        every time I start a process, the fans spin up
      • 2021-07-21 20208, 2021

      • atj
        ruaok: I prefer Mahau to Sangria personally
      • 2021-07-21 20213, 2021

      • atj
        *Mahou
      • 2021-07-21 20205, 2021

      • alastairp
        a coworker of mine just got the new M1 ipad pro + keyboard + LTE, and configured https://github.com/cdr/code-server, which is remotely-hosted vs code
      • 2021-07-21 20206, 2021

      • atj
        alastairp: I think Apple are hamstrung by the Intel CPUs
      • 2021-07-21 20215, 2021

      • alastairp
        he went on holiday last week and said it was perfect
      • 2021-07-21 20245, 2021

      • atj
        not surprised they went to ARM. The thermal management on the higher end Intel laptop CPUs is terrible
      • 2021-07-21 20209, 2021

      • ruaok
        I've arrived at a point where I don;t want to buy any more intel products.
      • 2021-07-21 20210, 2021

      • alastairp
        yeah, I got the i9 because I thought it was a good idea to future-proof it, but in hindsight I think I should have gone with the i7 and a lower spec discrete gpu
      • 2021-07-21 20212, 2021

      • atj
        My 2015 MBP hardly ever used to spin up the fans.
      • 2021-07-21 20222, 2021

      • ruaok
        well, actually I arrived there in 1991, but hey
      • 2021-07-21 20215, 2021

      • alastairp
        meanwhile, at my desk I have a thunderbolt dock with DP+ethernet+usb-a+power delivery, and 1 in 3 times that I plug it in, I kernel panic the mac
      • 2021-07-21 20238, 2021

      • lucifer
        akshaaatt[m]: currently none. will let you know if there are any in future.
      • 2021-07-21 20242, 2021

      • ruaok runs out to find noms
      • 2021-07-21 20254, 2021

      • lucifer
        alastairp: hah! nice. so let's move AB to newer gaia?
      • 2021-07-21 20214, 2021

      • atj
        yeah, it pissed me off no end they ditched the HDMI port and magsafe. Apparently they're bringing them back for the M1 MBPs
      • 2021-07-21 20236, 2021

      • atj
        You just have to wait for a couple of years until everything is M1 compatible
      • 2021-07-21 20254, 2021

      • alastairp
        atj: mmm, yeah. so. I don't actually mind that too much. I really like the concept of the dock (1 port to plug in and I get keyboard, mouse, gige, power, screen)
      • 2021-07-21 20209, 2021

      • alastairp
        and 90% of the time my machine is at a desk, so the lack of magsafe doesn't bother me
      • 2021-07-21 20223, 2021

      • alastairp
        and in fact, I much prefer being able to plug in on either side when I'm on the sofa
      • 2021-07-21 20237, 2021

      • atj
        Yes, that is the one advantage of the USB-C ports
      • 2021-07-21 20244, 2021

      • alastairp
        lucifer: yes, right! So, I'm just runnning some experiments to see if the training process gives the same results. If so, we'll be good
      • 2021-07-21 20258, 2021

      • alastairp
        and then yes, we should immediately upgrade to py3, even if we're not ready for sklearn
      • 2021-07-21 20248, 2021

      • lucifer
        +1
      • 2021-07-21 20216, 2021

      • alastairp
        usefully, the most recent version of gaia allows you to set a random seed, so I hope that we should be able to reproduce exactly the results
      • 2021-07-21 20229, 2021

      • lucifer
        makes sense
      • 2021-07-21 20245, 2021

      • lucifer
        https://github.com/MTG/gaia/issues/111 is still open, so not py3 changes not merged yet?
      • 2021-07-21 20221, 2021

      • Toasty joined the channel
      • 2021-07-21 20217, 2021

      • alastairp
        oh, in fact, I fixed the issue in the first comment in https://github.com/MTG/gaia/pull/115
      • 2021-07-21 20216, 2021

      • alastairp
        I don't want to jump on to yet another project without first finishing some other things
      • 2021-07-21 20238, 2021

      • alastairp
        so I'll add this to my list, and hopefully get to it in the next few weeks
      • 2021-07-21 20246, 2021

      • BrainzGit
        [critiquebrainz] 14dependabot-preview[bot] opened pull request #367 (03master…dependabot/npm_and_yarn/codemirror-5.62.1): [Security] Bump codemirror from 5.45.0 to 5.62.1 https://github.com/metabrainz/critiquebrainz/pull…
      • 2021-07-21 20249, 2021

      • BrainzGit
        [critiquebrainz] 14dependabot-preview[bot] closed pull request #366 (03master…dependabot/npm_and_yarn/codemirror-5.62.0): [Security] Bump codemirror from 5.45.0 to 5.62.0 https://github.com/metabrainz/critiquebrainz/pull…
      • 2021-07-21 20215, 2021

      • Toasty has quit
      • 2021-07-21 20225, 2021

      • reosarevok
        yvanzo, yyoung: one thing I did notice: currently, if I only want to add one URL to an artist, I paste it, the autoclean does its thing, I press Enter and it gets submitted,
      • 2021-07-21 20248, 2021

      • reosarevok
        but with the new code enter only locks it, and even a second enter won't submit the form
      • 2021-07-21 20213, 2021

      • ephemer0l has quit
      • 2021-07-21 20252, 2021

      • ephemer0l joined the channel
      • 2021-07-21 20254, 2021

      • lucifer
        ruaok: alastairp: monkey: i have generated weekly top releases from the contents of new dumps as a test. do you see any improvements?
      • 2021-07-21 20259, 2021

      • ruaok
      • 2021-07-21 20205, 2021

      • lucifer
      • 2021-07-21 20224, 2021

      • ruaok
        It is accurate, but i can't say it's better or worse.
      • 2021-07-21 20235, 2021

      • ruaok
        Hard to judge without seeing the before version.
      • 2021-07-21 20246, 2021

      • lucifer
        for me too its similar but there are now more links to MB.
      • 2021-07-21 20238, 2021

      • lucifer
        right, can you look at the month version now? i'll generate it next.