ansh: hi, could you please review CB#448? And if it works for you, please rebase your existing PRs with it (I'm having trouble doing a docker build with node 12, so have upgraded it to v16)
let's add both lookups, and have a flag in the code to switch between them. At the moment I think that the MBC lookup will give better results
in fact, for a handful of the nan values that I checked, the mbc lookup does give a result
Pratha-Fish
Interesting. I'll add the option to switch then
alastairp
sweet
Pratha-Fish: when that's done, I have a few potential PRs for you to do. let me know if you're interested in them
Pratha-Fish
alastairp: definitely, getting used to the OSS workflow is a must for me
alastairp
perfect - it'll require you to set up listenbrainz on wolf (not the whole infrastructure, but the code is there). I can help you step through it, but I have a lunch meeting in ~1 hour. so up to you - we can do it, now, or we can do it when I'm back in 3-4h
ansh: 👍 thanks, will do. I think the lock file might be because I generated the updated lock file locally with node 17 instead of the 16
ansh
I noticed these things when I setup CB following the docs
alastairp
and I can see why the static builder one happened too (see that it's using `docker run` without the --user flag, this means that it'll probably be running as user 1001, which means that it can't modify the existing files in your home directory)
Pratha-Fish
alastairp: Let's do it after lunch. Till then I'll be able to finalize the mapper script as well
alastairp
this is similar to the pg_custom errors you were seeing previously, and related to a fix that lucifer made on LB node tests too
actually, I see that we don't have the --user flag in CB's develop.sh like we do in LB. interesting. looking more
I'm just rebuilding the docs now. I guess they're on manual. I still don't understand how building docs works :( same issues with LB
ansh
I think we can have a script that runs every time a new release is made
alastairp
yes, we tried that with LB, but it was creating new versions each time, which made things very messy in the readthedocs interface, because sometimes we would make many releases a week
doc builds are failing. I suspect a similar reason to what lucifer found in LB last week. when I'm back after lunch I'll look at them
ansh: `search_editon_group` just noticed this 😅
lucifer
alastairp: probably jinja2 incompat version. can open a PR to fix.
lucifer: alastairp : I cleaned up old "Top 100 top artist..." playlists that were old and stale.
riksucks
hi lucifer, I wanted to ask you something
lucifer
hi!
riksucks
so the api rn is sending back the userids of the followers. I was wondering the following approach.
the frontend has an access to all the people following me
so in the frontend, I can do the mapping, and show the usernames of the people I sent personal recommendation to
if a name is not present in the frontend, but the userid is there, frontend would assume, that the person has unfollowed him, hence not show anything
lucifer
riksucks: but external api users do not have access user ids so they can't do such a mapping
riksucks
oh right, yes
lucifer
thus, the api should only deal in user names, it should convert usernames to user ids when writing to db then convert user ids back to user names when sending events to users.
riksucks
also another question. We should be saving data with the ids, but when serving them, we would be populating it with usernames. right?
because again, usernames can change
right
lucifer
yup db should store user ids for this reason
riksucks
right thanks. also, there exists `get_many_users_by_mb_id` to get info about users from mb_id. making a `get_many_users_by_id` sounds like an okay approach to you for this right?
lucifer
riksucks: yes. that sounds good.
but i'd suggest to first try adding a join to "user" table directly in the events query.
you can user something json_array_elements_text to expand the array to 1 row per id, join with username then collect back by grouping on event id.
i can help with that if you get stuck.
akshaaatt
Thank you mayhem! Received.
chinmay
lucifer: how often does lb get a new prod build?
lucifer
chinmay: do you mean how often we do a new release?
chinmay
yes
lucifer
there's no fixed timeline, whenever we have enough stuff to release or there's an urgent fix so on. sometimes its multiple times a week sometimes a couple of weeks without any release/
chinmay
oh, alright
I wanted to try out fresh releases api and I was avoiding setting up musicbrainz server locally
mayhem
we could deplay to test.mb.org and let you do that, chinmay
ansh: I think I worked out the reason for the permission denied errors. newer versions of npm have this funny behaviour where they will see if they are running as root, and if the owner of the directory that they are running in is different to root, they will try and chown the output