#metabrainz

yes, shared translation memory is possible

(but not with the dedicated cloud thing apparently)

looks like we can't download the existing TM from transifex, though, unless we have a payed plan there :(

ew

so if we want tshared translation we need the 17eur version?

if I understood the prices page correctly

outsidecontext: ok, that was the main point (using both weblate and transifex for Picard to start with), thanks.

If we decide to switch to Weblate, could we do a one-time payment to Tfx to fetch the TM from there and import that to WL?

I worked on MB OAuth integration too, so I’m curious if it could be possible to have it with paid hosting.

That's what I was thinking too

I'm confused that shared memory is only available on the 17 tier, but SSO only on 25? WTF?

"You can add screenshots to ease translation of UI"

wow

Either way, seems like a decent trade-off to get out of exactly that vendor lock-in. The longer we stay in the lock-in, the more the sunken cost will be.

ruaok: I think this note about shared TM is meant different: On 17€ you only get a TM you have to share with others

ruaok: yea, that seems weird. so i think we might ask the guy

oooohhhhhh

I think we should go with the 25€ version for full control of settings

ah, ok, yes.

should I just go sign us up?

I mean we could test picard *first* but

...

"You can integrate your own oauth provider, so login with MB credentials would be possible"

oh, loads more options. let me know what we need, I have no clue what our current needs are.

man I say go!

Maybe outsidecontext and ruaok et al can figure out the specifics outside of the meeting. :)

yeah.

nothing i've seen on weblate or testing it some time earlier made me hate it :d

ruaok: I'll probably not have time to set things up this week, not sure yet. wait a bit, I'll check a few things what is possible and needed and then we can talk about signing up

yvanzo: Got more comments, or good to go? :)

CatQuest: not really out of the box, it requires some glue code.

looking further, I think 99eur is actually what we'll need. :(

sure of course

aw

I think we’re all done, so…

Thank you everyone for your time! Stay safe out there, and enjoy the sun (while wearing a mask and using sunscreen and otherwise staying safe!) if you have some where you are!

</BANG>

Thank you!

alastairp: https://github.com/metabrainz/brainzutils-pytho...

ruaok: however, they also offer that same plan free to open source projects, so maybe there's some wiggle room to ask for that but offer to pay something

Thanks !

Have a good week everyone!

the cache works but this time pulling the image took more time.

yeah, interesting

if there is a free plan, then I'll just donate 500eur per year and be done.

yeah, right.

this looks nice to support weblate one way or another

for sure.

the actions' docs say that pulling from docker hub is almost always faster.

I think the open-source plan is the same (for free) as the 99€/m plan, if I read correctly

though did outsidecontext say that all open source projects share a server and it's sometimes slow? might be something to consider

_lucifer: I think it's saying there "pull complete images from docker hub every time instead of caching the pulled image"

once we understand our needs I'll chat them up and see if we can reach some middle ground consensus that doesn't lump is into the OSS much

muck

alastairp: yes, that's what i intended to say 😅.

_lucifer: ah, right

maybe this will have a bit more impact on larger projects

yeah.

with the original basic version: 1m to build, 1m to pull, 30s to run tests

[musicbrainz-server] reosarevok opened pull request #2057 (master…edit-preview-id-warning): Fix Catalyst warning: id is missing in edit previews https://github.com/metabrainz/musicbrainz-serve...

I just re-ran and the pull was faster: https://github.com/metabrainz/brainzutils-pytho...

yup, i saw.

but it's 1m pull, 20s uncache, 30s run tests, 20s re-cache

jasondk: added comments to your proposal.

see how build was only 7s though!

yes, this should be useful in LB.

so in this case 20+20+7 is the same as the 1m to build

but if in LB it means we can do 20+20+7 instead of 5m, that'd be great

i'll add login to this PR and work on LB next.

maybe if we use github's docker registry, it gets a bit faster?

yeah, I was wondering about that

do you know why the cache has so many different items?

i think its caching all the layers

if you expand the log, it's uncoompressing hundreds of files

ah, because it tells docker to write layers to /home/runner/work/_temp/, and then it just compresses everything?

yes, see the post run log

/usr/bin/sh -c docker save '93323ae7d41a' '93323ae7d41a' 'd4b80a89fb19' '539e6638127c' 'fd3a0bfdf6e5' '73f376d0faa9' 'd2b96ca0ec01' '1258b0a636fa' 'd674c0ed597b' '3b94a3228b2f' '31fb65592bb8' '12224c6bf952' 'cf3f2a822d1d' 'e8c5cba13786' '87c80a349ede' '87c80a349ede' 'f5b92d01403e' 'deb6b0db88c8' 'fd3a0bfdf6e5' '73f376d0faa9' 'd2b96ca0ec01' '1258b0a636fa' 'd674c0ed597b' '3b94a3228b2f' '31fb65592bb8' '12224c6bf952' 'cf3f2a822d1d'

'e8c5cba13786' '1191a3b81e8b' '1191a3b81e8b' '634c18ad9871' '4b29d21373f6' 'fd3a0bfdf6e5' '73f376d0faa9' 'd2b96ca0ec01' '1258b0a636fa' 'd674c0ed597b' '3b94a3228b2f' '31fb65592bb8' '12224c6bf952' 'cf3f2a822d1d' 'e8c5cba13786' 'e8c5cba13786' 'e8c5cba13786' 'brainzutils_test_test:latest' '93323ae7d41a' 'd4b80a89fb19' '539e6638127c' 'fd3a0bfdf6e5' '73f376d0faa9' 'd2b96ca0ec01' '1258b0a636fa' 'd674c0ed597b' '3b94a3228b2f' '31fb65592bb8'

'12224c6bf952' 'cf3f2a822d1d' 'e8c5cba13786' 'metabrainz/python:3.7-20210115' 'e8c5cba13786' | tar xf - -C .

maybe we could tell it to not do any compressing for example, to make it faster?

we'll see how it goes in LB, maybe it needs a bit of tweaking

less tweaking than jenkins though ;)

yeah, let's try that in LB so that we can have better results. in BU, the variance in results will be more i think

i was going to try this for login `echo ${{ secrets.DOCKERHUB_PASSWORD }} | docker login -u ${{ secrets.DOCKERHUB_USERNAME }} --password-stdin`

this shouldn't print the password to logs i think, but i would like to confirm.

yeah, I think that might work

rememeer that you can set the password to a fake one first and run it... see what happens ;)

if it works, the login should fail

ah yes! will do that.

seems to work as expected!

regarding LB spammers: what's the point there? I don't really get the concept, how and why does someone spam LB!?

the spammers send listens of the same song or artist a large number of times.

i am not sure what the point is but it can definitely act as a bias in recommendations or other features.

outsidecontext: beats me

utterly pointless

but at least the sockpuppets are easy to identify

maybe agents of some label or popularity thing

ok. I guess as soon as one can submit data somewhere someone will spam it

yep

googlejuice maybe?

spam is so stupid! i didn't sign up for ebay for yeeeaaars becasue i got spam from there and it put me off

alastairp: ruaok: how about adding docker password and username as organization level secrets so that we can use them in actions across all MeB projects. otherwise, we'd have to add it on to each repo individually.

only after it was explaiend to me that they thmselves didn't sned spam did i start to understand

sounds like a decent idea to me, but I don't know much about it.

i think there's probably a control for the org admin to define it there and then limit visibility to specific repos.

(in case we want to do that)

_lucifer: yeah, good idea to make it org-level

awesome! ruaok, can you do the honours please :)

_lucifer: find some docs for him to follow :)

+ names of the secrets, and where to find the values in syswiki

we need to add `DOCKER_HUB_USERNAME` and `DOCKER_HUB_PASSWORD`. their values are here https://github.com/metabrainz/syswiki/blob/mast...

Mr_Monkey: what component did you use for the date time picker

shivam-kapila: https://github.com/wojtekmaj/react-date-picker

Thanks :)

I dropped the time picker part, which comes with another package (https://github.com/wojtekmaj/react-datetime-picker)

Oh the latter is what I need. Thanks again

[brainzutils-python] amCap1712 opened pull request #64 (master…cache-misc): Miscellaneous pending cache changes https://github.com/metabrainz/brainzutils-pytho...

alastairp: what details should we add here? https://github.com/metabrainz/brainzutils-pytho...

[listenbrainz-server] amCap1712 opened pull request #1385 (master…gh): Ciao Jenkins! https://github.com/metabrainz/listenbrainz-serv...

_lucifer: just something like "calls _prep_key on the keys of dictionaries with the namespace and encode_val on the values if encode is True"

Thank you alastairp :)

alastairp: 👍, and regarding tests for `_prep_key`, do you have any specific test case in mind that should definitely be added?

remind me - are there any tests for it?

what does it do? add namespace to a key?

no

and I appreciate the feedback ruaok, I will update the proposal on the forum and also submit it on the GSoC site tonight.

it used to encode to bytes, but doesn't do that any more?

Hello everyone, so the idea I am planning on working is "Integrate more music services for recording listens and playing music" and I need some guidance how should write my proposal. If you guys have some referral implementations that would be great. I want to integrate deezer

(sorry, I don't have it open atm)

alastairp: yes add to namespace and ensure ascii

_lucifer: right, those sound like 2 great tests :)

👍

vardan: we currently have a spotify integration which may be helpful https://github.com/metabrainz/listenbrainz-serv...

Do you guys have any reference code or any reference template that I can see. That would be a great help

thx @_luc

deezer or any other music service should probably be similar

Ohkay thx