#metabrainz

i, is everyone glorious?

because i know you are. GLORIOUS

and now I got to night. natta all

that reminds me, I should make backups

night!

yes. yes make backups! aaaa

night! :D

[critiquebrainz] mwiencek opened pull request #89: Fix a couple language issues in the MusicBrainz release group links (master...rg-string) https://git.io/vyP7x

[acousticbrainz-server] nupurbaghel opened pull request #231: AB-300 corrected absence of file - recordingtoartistmap.json (master...ab300) https://git.io/vyPxW

samj1912: sorry, I just saw your comment about removing an unmatched file, let me have a look at that issue

gcilou: :)

I did find mine too, so I can wear it for my presentation on Sat.

antlarr there are a couple more issues

See jira

I am traveling for the next few days

Can you please see if you can solve them

samj1912: ok

sure

They are all edge cases

Edge cases are best cases.

ruaok: fixed the link to the list of tickets and typos in the blog

Leo_Verto[m]: Do you have a ticket that should be in MineoBot?

...

Jira's user search is broken?

Is zas or Gentlecat in charge of https://tickets.metabrainz.org/plugins/servlet/... ?

Freso: awesome :)

Freso: define in charge

hey I want a MeB t-shirt too !

Same

nupurbaghel: hi.

we just had a talk about the sklearn system. there's quite a bit still missing that we haven't planned

making a start on removing variable length descriptors is a good idea

I'll add this to the ticket too, but how about we do this:

make a package train.transforms

inside there you can have your transform file

e.g. removevl.py

ohkay

with a class inside it - class RemoveVariableLength or something, with a function `def transform(input)`

the input data should be a list of dictionaries - the equilavent of the low level data files

also dmitry suggested to remove all variable length params as of now

the output should be the same list, but with all variable length descriptors removed

yeah i get it

yes. I recommend that you inspect all descriptors to find the ones which are different between each lowlevel file

(instead of a hard-coded list of descriptors to remove)

I thought it that way only

this week we will write some more tickets, with a list of transforms which we have to implement. This would be a good list for you to work from

did you manage to run this python code?

yes perfectly well , after some minor import errors

oh, great

also the requirements file needs to include scipy and numpy

so, you see that we can explicitly list transforms here: https://github.com/MTG/acousticbrainz-sklearn/b...

but some (such as removevl) is implicit

we have some others as well. we will put them in the ticket

are you sure? they are listed as dependencies of scikit-learn, so should automatically be installed

I was working with pycharm, maybe it was due to that

I will check again

we have some ideas for the design/workflow of this software - we haven't written it down yet though

we will try and write some of it down this week

You have 3 pull requests outstanding from me on the server project, right?

ah, https://github.com/metabrainz/acousticbrainz-se...

we have some feedback on that, we'll have to make some changes. I'll try and get you some feedback on this one this week too

yeah right 3 pull requests.

Yeah altering can be done on https://github.com/metabrainz/acousticbrainz-se... accordingly

samj1912_riot[m]: https://github.com/metabrainz/picard/pull/657

that should fix the issues with the cover art

antlarr ping

samj1912: pong

Can you take a look at PICARD-1001 too?

PICARD-1001: Cover art changes are indicated even if there's no change after loading a release https://tickets.metabrainz.org/browse/PICARD-1001

samj1912: about PICARD-1016, you're still right that the original cover art thumbnail doesn't have a tooltip, so I'll add it, but the glow is ok

PICARD-1016: Bugs if related tracks have multiple images https://tickets.metabrainz.org/browse/PICARD-1016

I'll try to have a look later, I was about to leave

antlarr okay, 1001 since it was part of the related changes

Also can you properly test your PR once

For all the CA providers?

how?

Change in options

You can load any arbitrary release in picard

And tag a simple 1 second of silence to any file and save

For loading arbitrary release use lookup in browser

there should be a testsuite for that, what happens if the release I try doesn't have any CA in that provider?

Basically just CAA, Amazon and local