#metabrainz

(all except the file created timestamp so on)

ok so its like someone uses the same audio file to generate fingerprint and that audio file has no meta data but our database has meta data linked with that fingerprint then we can return the meta-data to the user?

yes

ok that's a nice feature!

it is forcing me to use ndk version 21.2.6472646 while i have 21.3.6528147

you can set the ndkVersion in the build.gradle file

`ndkVersion "21.3.6528147"`

worked thanks

why can't we give login/ signup in the app itself?

Its a basic feature in any application

MusicBrainz has its own OAuth

we are using OAuth instead of Http Digest

we could try using a webview inside the app instead of opening a browser but i don't think that would make much difference. also not all people like chrome

never heard of it before

_lucifer yeah we can do that

can we use this

?

white_shadow: already on my list to integrate the app auth with account manager :)

but you can work on it if you want. i do not have a timeline to implement this

so we can do login/signup from app itself by using this?

no it is not that simple

the user will still have to login using the current workflow

it is just that the auth token will be managed by the account manager api instead of our app directly

yeah it seems difficult, but i think API can we written in Auth0 so why can't we have the API for application?

because we already have our own authentication apis

which are used by numerous other projects and applications

ok but they are not compatible for android use?

but we are using them so they are compatible

the workflow is different from other applications that's the only differene

[bookbrainz-site] MonkeyDo merged pull request #489 (UserCollection…fix-checkEntityTypeBeforeAdding): fix: throw error when trying to add collection, area, editor to a collection https://github.com/bookbrainz/bookbrainz-site/p...

Mr_Monkey: I saw your review. I wanted to talk about whether the entities in collection should be in ascending order or descending order

ok

I changed it to DESC because if it's ASC then user might not be able to see the added entity

because of pagination

Yeah, I figured that was the reason.

should i change it back to ASC?

I'd be more comfortable with showing a success message and refreshing the collection (if it was a success)

Project bookbrainz-site build #3314: passed in 3 min 29 sec: https://travis-ci.org/bookbrainz/bookbrainz-sit...

Hmm okayy

Honestly I'm not sure about our sorting. We'll definitely want to replace it so that users can sort with the column and order of their choosing, but it does require a sizeable refactor.

How about showing the success message and changing the page number to the last

so user can see the added entity

I don't think that's a huge issue if there's a clear success message

and as a user, I prefer not to be taken somewhere I didn't request to go

Let's shelf that one as "issues that will be solved with proper sortable tables"

yeah makes sense

Showing a success message however is good either way, and kind of solves our current predicament

white_shadow: if you want a nicer ux with login, the only option i see is http digest. it could be added as a secondary option for users to enable if they want. but then it is essentially a UX v/s Security tradeoff. we would have to ensure that the password is safeguarded and that is difficult. if you can ensure that, it can be added but only as a secondary feature

Umm showing a message is tricky

because we are refreshing the page

white_shadow: also there has not been any request by the users in this regard so i do think it is not only unsafe but unnecessary

What I suggest is: instead of refreshing the page, call a method passed in the modal props (something like `succesfullyAddedItems`) that will do three things in the collection component: 1. close the modal 2. show success message and 3. call the pagination endpoint to refresh the page

It complicates the component a bit, but I think the seamless user experience is worth it

how many active users we have in our app? _lucifer

1000-5000

but i think it will rise once we get the tagger and listening working

Customers don't like to get redirected to any other app, and i think the website is not compatible with the mobile view itself

Yeah alright. and let's keep the entities in ascending order right now. With the most recently added in the last

i had to scroll and zoom in, zoom out

reosarevok: thanks, made it public!

but yes that all depends how much concerned we are regarding our customers count

not customers but users

yeah

if the users request this feature in decent numbers we can reconsider

on how to approach the issue

but for now there haven't such requests

its fine man!

2020-08-11 06:54:04,296: Failed to import recording with id in bounds (233327, 249149)

2020-08-11 06:54:04,477: (psycopg2.OperationalError) server closed the connection unexpectedly

This probably means the server terminated abnormally

before or while processing the request.

It always seems to be happening on "recordings"

yvanzo This is the entire output https://pastebin.com/wnm1ktrE

"recordings" is the largest index, you can add "--entity-type recording" to rebuild only that index

most probably some recently added data makes the indexer to fail, I will check that

double-check the log output of postgres too - I had a database import failure a few weeks ago and it was because there wasn't enough memory in docker for mac allocated, postgres was corrupting data and restarting

thge "server closed the connection unexpectedly" reminds me of that error

nelgin: for the above, run docker-compose logs -t db

db_1 | 2020-08-10T14:42:03.341858653Z 2020-08-10 14:42:03.336 UTC [24] LOG: database system was shut down at 2020-08-10 14:41:06 UTC

db_1 | 2020-08-10T14:42:03.341871974Z 2020-08-10 14:42:03.341 UTC [1] LOG: database system is ready to accept connections

db_1 | 2020-08-10T14:53:10.989199804Z 2020-08-10 14:53:10.989 UTC [42] FATAL: database "musicbrainz_db" does not exist

db_1 | 2020-08-10T18:46:42.630713171Z 2020-08-10 18:46:31.822 UTC [1] LOG: server process (PID 5048) was terminated by signal 9: Killed

I need to run to a doctors appointment. I'll try allocating more memory. How do I fix the DB in the meantime?

bbiab

This one is a false positive: FATAL: database "musicbrainz_db" does not exist

(actual an innocuous error triggered by an init script)

[musicbrainz-server] reosarevok opened pull request #1646 (master…MBS-11031): MBS-11031: Convert historic Edit Track edit to React https://github.com/metabrainz/musicbrainz-serve...

MBS-11031: Convert historic Edit Track edit to React https://tickets.metabrainz.org/browse/MBS-11031

[acousticbrainz-server] alastair merged pull request #257 (master…preferences): AB-98: Add a feature to select SVM parameters https://github.com/metabrainz/acousticbrainz-se...

AB-98: Generate custom project file during model evaluation stage https://tickets.metabrainz.org/browse/AB-98

nelgin: welcome back

if you run `dmesg`, does it show any text about Out of memory? (or OOM killer?)

how much memory does the VM have?

I had 8gb, now increased to 10. I'm just trying to start up the docker but it looks like it is wanting to install and compile a whole bunch of stuff for some reason s I'll let it finish.

and you said that you made some changes to memory settings as well? it sounds like this might be the cause of the crash

I increased pgsql from 2gb to 4gb per the suggestion

the other suggestion, which might not be listed in the documentation, is that the postgres memory value should be no more than 25% of your total ram

so if you set it to 4gb, ideally you should have 16gb memory available

OK, that I did not see.

There is nothing here about any such 25% limit.

it's a best-practice recommendation of postgres - https://wiki.postgresql.org/wiki/Tuning_Your_Po...

Something is terribly wrong here. I'm just going to reinstall the vm. Also, any way to have it use a us rather than eu mirror for downloading the files?

under the documentation for shared_buffers: "If you have a system with 1GB or more of RAM, a reasonable starting value for shared_buffers is 1/4 of the memory in your system."

I have 10 gb that's all I can give it.

It sounds like a good idea that we should add a note there about this, or a link to the postgres documentation

in that case, you might be able to get away with leaving it at 2gb, although this will make things quite a bit slower

alastairp: I have read that before, ideally yes but otherwise it isn't supposed to affect stability iirc

yvanzo: as I said previously, my experience with shared_buffers set to 4gb and only 8gb in docker-for-mac was that postgres would reliably corrupt when building db indexes during data import

increasing docker ram made the problem go away

here is the doc https://www.postgresql.org/docs/12/runtime-conf...

alastairp: ok thanks, I never tested with this platform

OK, reinstalling ubuntu. I'll take a snapshot this time once It has done the updates :)

Building the docker image.

yvanzo, bitmap: can you find anything that uses root/edit/details/add_link.tt ?

I'm at the "Download latest full data dumps and create the database with..." but. How can I get this to use a US rather than EU mirror?

Nobody?

reosarevok: looks unused to me

nelgin: you should be able to just add the URL of the US mirror to the createdb.sh command

iliekcomputers: Hi

heyo

how goes it?

going well

I started writing the code for the backend today

awesome!

I spent the evening thinking about a new way to store the entity stats

i don't think we need to plan much for this week, i think the first sitewide graph will take this week entirely.

Yeah about that

I was wondering, instead of storing the entity stats in a json what if we store it in a new table in a relational way

it will solves a lot of problems for us as well as allow us to implement new features easily

what I was thinking was we create a table with the following schema for top artists suppose

i think that is a good idea, and eventually we will probably do that. my instinct is that it isn't in the scope of your current project.

we initially went with json because we didn't know what data we would calculate and in what format.