#metabrainz

now to fix it, i guess the easiest way would be to set a date before which it should never look and just error out

another way would be to just look at most a year or few months and error out otherwise?

i don't think we want to further back than that to find the latest listen

iliekcomputers: thoughts?

sorry what's the bug?

the function you mention is used to find the latest listen in the spark cluster.

the cluster was empty and it went back till the medivial times to find a listen

ah, interesting.

yeah, we should have a lower limit there i guess.

hesitant to set it to the current year as it could lead to errors on jan 1.

the year before the current year maybe.

6 months or 1 year back from today sounds good to me

yeah, sounds reasonable to me.

👍

the absolute lower limit is jan 1 2005.

yay dump is nearly imported.

then mapping

I'll test my code to write user similarity locally now.

_lucifer: import succeeded. now importing mapping.

yay!

Done! Time taken: 0.04

lol

do we have any code yet to display the user similarities?

shivam-kapila shared a mockup for that yesterday. he might know

The semi circular dial is almost done.

got link?

To the mockup or code?

mockup

A sec please

I was also wondering about the similar users that would be shown on your own friends page.

[listenbrainz-server] paramsingh opened pull request #1316 (master…param-recommend-recordings-from-listens-page): [wip] Allow users to recommend recordings from the listens page https://github.com/metabrainz/listenbrainz-serv...

ruaok: 'You mean people similar to people I follow?

I think so, yes.

I added something to the figmas board. something like that works?

*figma

hmm. tightly packed sidebar, empty main space.

what else goes into the main space here?

we were not yet sure. I think we were gonna ask users to pin something (like their fav. recordings) to their profile?

how about putting the "similar users" into the main area for now?

I have a few questions

mapping import done

does the similarity need to be a dial?

how does it fit on the listens page?

we can use linear scale if thats preferable

last.fm does a simple progress bar line which fills up and changes color with different levels of compatability.

(I mean thats an easier react component :D)

_lucifer: dataframes requested, with 365 days to start with.

> I have a few questions

Do we want to make the profile page for those whom I follow and whom I dont exactly to be same

👍

2. Do we show similarity score on profile of people whom I already follow?

ruaok: did you send job_type for similarity?

python3 manage.py spark request_dataframes --days=365 --job-type=user_similarity

3. Do we show people similar to John Doe(whom I dont follow yet) when I visit their profile

we should rename the job type to "similar_users" which is what we call this feature in other places.

i have renamed it and pushed the change

but its not updated in the container yet

thanks.

iliekcomputers: How about these: https://getbootstrap.com/docs/3.3/components/#p...

shivam-kapila: yep, something exactly like that

okay

1. Do we want to make the profile page for those whom I follow and whom I dont exactly to be same

2. Do we show similarity score on profile of people whom I already follow?

3. Do we show people similar to John Doe(whom I dont follow yet) when I visit their profile

ruaok: Mr_Monkey: iliekcomputers: _lucifer: alastairp Can you share your preferences for these

my take: 1. not sure. 2. sure. 3. yes, I think so.

are we going to followers while fetching similarity? if not, might as well display it for all users

1. probably? What would be the differences?

2. yes

3. I think that's a good way for users to explore more music tastes without being too invasive. Definite yes for me.

Mr_Monkey: can you review https://github.com/metabrainz/listenbrainz-serv... Once this is done, I can get started on the changes we need to make to feed API endpoint.

I am also not quite sure about the differences yet

iliekcomputers: OK!

iliekcomputers: Where should we place the similarity score component on listens page

Food first, but top of my list.

Above BP?

[mbsssss] reosarevok opened pull request #60 (master…SEARCH-646): SEARCH-646: Prioritize exact matches for tag search https://github.com/metabrainz/mbsssss/pull/60

Under the user name?

shivam-kapila: i'd say we can put it somewhere just above the listens or something.

We should probably refactor in the future and have a UserName component we can drop in anywhere that has the user name (duh!), compatibility score and follow button.

yvanzo: ^ can you see if that PR seems legit? If so, how do we test it? :)

_lucifer: cron snook an incremental import into the mix. lets see how that goes. :)

Mr_Monkey: yep, and the supporter badge idea we had.

That too !

And anything else we end up with that we want to associate with a short presentation of the user

iliekcomputers: Reg. the failing build numpy has dropped support for python 3.6.x and we are using 3.6.9

:/

did it happen like yesterday?

i think numpy 1.20.1 and a pr was recently merged to update numpy

we need to revert that for spark, i'll open a PR

shivam-kapila: oops, good catch. that was my fault

let's downgrade again

what version were we previously using?

could you revert that and also add a comment that explains why we use this version?

sure thing

I think that _lucifer was looking at upgrading the java version of the spark images, so once we do that we can also upgrade python (see https://github.com/metabrainz/listenbrainz-serv...)

ruaok: this incremental dump import is taking a long time ;)

` __init__: Copying old listens into '/temp'`

if this is moving a whole dump out of the way... yeah.

not the best timing, cron.

yeah, that's a thing we need to fix.

sheesh, cron. timing is supposed to be your one job.

it wouldn't be a problem if our data in spark was partitioned by day instead of month.

_lucifer: gives me a moment to debug my code. :)

Mr_Monkey: once you're back from lunch, let's chat about the things i should return from the feed endpoint once before i get started.

[listenbrainz-server] shivam-kapila opened pull request #1317 (master…revert-numpy-version): Revert numpy version to 1.19.4 https://github.com/metabrainz/listenbrainz-serv...

alastairp: done

shivam-kapila: there is a 1.19.5 released too

oh yes it also supports 3.6. Updating

shivam-kapila: can you add a comment explaining why we can't upgrade in the requirements.txt

done

shivam-kapila: sorry, one more thing. can you put the comment one line above, instead of on the same line?

[listenbrainz-server] amCap1712 merged pull request #1295 (master…spotify): Check if user has revoked authorization if refreshing token fails https://github.com/metabrainz/listenbrainz-serv...

ruaok: alastairp: if its fine i'd like to make a release to beta to test the above PR

alastairp: done

_lucifer: what service does this affect? spotify reader?

_lucifer: we don't run spotify reader in beta, so it'll go in prod.

it's only present on prod

oh ok!

maybe we should take a bit more time and set up a full stack on test?

oh come on another one :(

get some popcorn _lucifer

ruaok, can you please check the queue and cancel in case there are more?

only our request.

😄

I just finished debugging my code. let me push a new image and get that deployed.

but I am happy to terminate the request consumer to allows us to move on.

Mr_Monkey: hey.

sure if it won't end up corrupting the dump

So this is what I have so far: https://docs.google.com/document/d/1_rVY55eHksV...

unsure. let's hope for the best.

🤞

yo