#metabrainz

Gentlecat: do you think we need schema changes for CB in order to improve the DB access?

alastairp: I know. I might, even, depending on whether I end up having an internship over the summer or not

ruaok: I'm getting dmitry to review the content search proposal. plus that would be 7?

no, I think schema is just fine

that's what I thought too.

it's the way the database accessed. it's a mess

do you need me to chime in on that proposal at all?

partially my fault that I didn't notice it soon enough

but that's 3 AB proposals. We'd need another mentor for AB

we'll worry about the mentor/slot distribution later.

sure, if you have something to comment on

yeah, of course

alastairp: I can try to help with something there

👍

will probably be working on it during the summer anyway

would be nice to coordinate

as ruaok says, let's see what happens on the 25th

Yo have 5 people working on ab over the summer would be amazing!!!!!!

To have

do we have some sort of penalty for cliches in proposals? :)

can you call pep8 a software requirement? :)

sweet that fixed it, cron appears to be working correctly now, just getting messages saying can't download the packets but assuming that's because i'm already up to date

[21:02:06] <reosarevok> bitmap: while I'm on the "probably repeated questions" topic, does the PG 9.5 thing help us at all wrt area containment queries and the ws?

ruaok: are you just salty because you got rickrolled?

No, but it’s only an issue as long as someone insists on using the view.

I'm not salty. I took a shower yesterday. :-p

Hmm. So it would not be a huge problem to display the containment for the relevant areas in WS results?

Probably not (although everything making WS results larger could be problematic by itself).

Are we talking "serious chances of being problematic by itself" or "well, I guess if we pile a lot of these things it might be bad"? :p

(if the latter, I'd still want to go for this one because there's no great way to get them otherwise, but it could be a separate include so that it's only added when people actually require it)

where does Musicbrainz store data about events? in which relation? is it in "place"?

adesh: events have their own table

Where they took place is stored with a relationship, that'd be l_event_place (or l_area_event if only the city is known)

Same for the performers, l_artist_event

i can't find the table "event" in the schema at https://wiki.musicbrainz.org/MusicBrainz_Databa...

It might be not up to date for that one

oh that's why :/

alright

adesh: https://github.com/metabrainz/musicbrainz-serve...

it might be easier to just look at https://github.com/metabrainz/musicbrainz-serve...

thanks

n one more doubt

what is gid for? or what does it stand for?

That's the musicbrainz ID

global ID, AKA MusicBrainz ID aka AMBID

s/AMBID/MBID

oh that's the musicbrainz ID? okayy thanks a lot

ruaok: emailed you the note I've written so that we can reclaim @BookBrainz

responded.

twitter sucks: "Invalid Case Status for upload"

...

I'll ask them to create a new upload link/reopen the case

ruaok: apparently they've closed the case without telling us and want us to make a new ticket

\o/

does anyone else have the same level of discourse admin as freso?

Omnifone also launches an innovative Music Information Retrieval platform, the Acoustic Lab, which promotes the analysis of the DNA of a digital music track, creating a library of audio and musical features that provide unique metadata sets. Omnifone offers customers the world leading audio analysis services, which scan across its 55 million track catalogue.

The Acoustic Lab can be used to enrich a customer’s end user engagement by providing unique service functionality. The Acoustic Lab is partnered with Queen Mary University London’s Centre for Digital Music, and provides a unique source of music-based research information for academia.

alastairp: it seems I might

What do you need?

can you tag dmitry as AcousticBrainz team member: https://community.metabrainz.org/users/dbogdano...

He has a cat, so I'll make the effort :p

done!

Does that news thing mean someone else (one of our data users, too!) is doing exactly the same as AB instead of joining efforts? :p

so it seems

https://twitter.com/cweichen/status/71223484479... looks ominous

(the reply)

looks like the data may be closed too

:D

Well I assume it is closed, it's a B2B and blah blah blah industry babble company :p

You could answer at that guy from @AcousticBrainz like "you could come talk to us instead ;) " :p

if the project is with c4dm, they'll be using sonic annotator

their framework for music analysis

"that guy" is chingwei, friend of musicbrainz and soundcloud guy

Heh

We have too many friends, I can't know them all! :)

It's a shame Twitter didn't add a "Haha" reaction like FB, or I'd Haha the "sales" reply from the MB account :p

I suspect they've "just" run c4dm analysis tools over the audio

but it will be difficult to know if they've changed any of the algorithms

and so we're in the same position as MSD

Guys, I've just had a very interesting conversation with the powers that be, which ended with: Could you find out if they could fix it? So, here is me finding out… :)

You know how musicbrainz IDs sometimes change/get removed?

No/yes :p

:D

(I guess I know what you mean with "change" though :p)

Go on :)

I mean change as in: they get removed :)

well, it appears that for an external client there is no way to differentiate between an ID that was never used and an ID that used but removed.

UmkaDK_: Not only for an external client :(

this leads to quite a few dead-end urls on google (for example) which end in 404

Right now, we don't store removed MBIDs anywhere

I mean, I guess they're in the edit table somewhere maybe? Not even sure

But yeah, when we split an artist (like I assume happened to "Glen Hansard & Markéta Irglová"), there's nothing left behind

AFAIK for us it's also as if the MBID had never existed

So I guess it wouldn't be a trivial thing to the response code from 404 to say 410 (intentionally removed) for ids that where removed?

LOL

I guess you can say that since you're out of her reach. :)

UmkaDK_: not currently, but that's something we've wanted to change for a while

ruaok: tell UmkaDK_ to tell someone to give us money to fix this :p

Hahahah!! :D

yeah, good luck.

(more seriously, the change might actually be kinda trivial, so maybe we should just get it done?)

(trying to find all the old, removed IDs that should be added to the "410" table though, that'd be tricky)

(but having a table to store removed IDs for the future would probably be useful anyway)

bitmap: opinion? :p

UmkaDK_: http://tickets.musicbrainz.org/browse/MBS-8380

Just an idea but how about I add an "improvement" issue and you leave a few comments on that … and I'll cross-link it into our tracking system...

Hahaha! Awesome!! :D

Knowing someone has a direct, requested use of this should make it more interesting to work on... I wonder what we'd want to store in that table

Just the fact the MBID existed? The entity type too?

Ok, I've cross-referenced that ticket on our system, so it will get seen. But if at any point you want me to highlight something, ping me and I'll make it bold, red and flashing.

Gentlecat: that stuff's broken :( and we've made tons of manual changes to webservice.sql for years so it's hardly automatically generated

that's why I don't like these huge monolithic test sql files

hmm

Do they *have* to be huge and monolithic?

Or is this a sign that we should be splitting them?

reosarevok: re mbs-8380, it'd be easy to just add the tables, less so to populate it with old stuff

Table*s*? You mean you'd have one per entity type or?

(I had thought of just one table because then we can fish for old MBIDs and add them all there without having to worry *what* they were, but dunno)

the problem is that I'm trying to run these tests and it fails on one of those queries, but I'm not sure which one exactly

well,, you could probably just do with one table

error message is not that helpful either :(

paste?

we'd need some triggers to insert deleted stuff there too

maybe next schema change

the test sql doesn't have to be monolithic but it's very tedious to split at this point

since a lot of the rows reference eachother so you have to pick it apart by hand

let me finish that lunch and I'll paste 🍴

good plan