#musicbrainz-devel

incremental stuff is unfinished yet

yes. just little comments to make setup easier.

Computers are hard

I mean the pull request

ah, no. I was just using your import-export branch working through the setup bits

and I was goingt to test the dumping/import stuff.

and when you have the incementeal stuff, I can test that too.

yeah incremental stuff is in the same branch

I need a place to test the live update branch too.

but usual dumping is supposed to work

alastairp: what room are you in? still in bldg 55?

Gentlecat: good.

importing with database initialization too

python ./manage.py init_db

psycopg2.OperationalError: FATAL: role "acousticbrainz" does not exist

54.005

Tallers building

database is supposed to be created

Go in the door and take first left

user too

Gentlecat: can we please get a manage option to do that for us?

like in CB?

not sure if CB does that.

sure, I think we can

it does

that would be grand. thanks.

I wouldn't mind :)

I'm kinda over it to be honest. I didn't realize it would be such a pain.

for a fluff feature.

yeah, turns out new tech doesn't work that well yet

not in python at least

lesson learned. :)

wow. 1.7M files in ABZ. cool.

ruaok: so how is that dumping thing? working as expected?

still working to submit a track or two to my server.

before I can dump.

how can I run the server on a different port/ip through manager?

through manager? I don't think you can run it through manager

any way to do it now?

I just use server.py

k

there are a couple of arguments, including the port

oh, it moved.

I thought it had gone

yeah, sorry for confusion

no worries.

I should document how to run this thing

+1

and actually try to use other stuff apart from web server

odd.

pip install -r requirements.txt?

yep.

and I just re-did that.

still can't find it?

weird...

Requirement already satisfied (use --upgrade to upgrade): flask in /home/robert/ve/lib/python2.7/site-packages

are you sure the python called by sudo is the virtualenv one?

Mineo: OH!

you SMRTer than me

album submission now running.

album submission?

abz

we're professional open source nuts. :)

I thought you could only submit tracks one by one

our shit is much more repeatable than researchers.

heh, that will never happen again

the export ran instantaneously. :)

its so fast since there were only 11 tacos. :)

Yeah, we know you can deal with 11 tacos pretty fast

no, I mean "album submission" part

first point of feedback. I'd like the tar file to contain the subdir acousticbrainzdump-20150113

rather than extracting things into the cur dir.

so I had this idea, not sure how feasible it would be

ok

assigning cover art to mediums

and also, lets call it acousticbrainz-dump-20150113, no acousticbrainzdump-20150113

reosarevok: these are not nommable tacos.

Awwww

I was lamenting that they had no name as of yet.

it's like a wad of data, I said.

wad? wad. oh, taco!

thats what wads are in spanish. :)

and we had tacos. :)

Gentlecat: ignore my comments about album submission. just being silly.

I thought for a minute that you had some magic proprietary script that can submit albums :(

sorry, no.

how do you pass the data dump argument to init_db?

python manage.py init_db dump.tar.xz

tried. that. :(

hm, was working before

try "python manage.py init_db -a dump.tar.xz"

\ø/

looks like it works.

magic

so, aside these small bits of feedback, you seem quite on track.

and now I'm ready with a sandbox to test your upcoming patches.

lol

btw, that subdir in tar file

should I rename abdump into that?

or just put everything there?

tar xJf acousticbrainz-dump-20150113.tar.xz

- acousticbrainz-dump-20150113

- acousticbrainz-dump-20150113/COPYING

- acousticbrainz-dump-20150113/TIMESTAMP

- acousticbrainz-dump-20150113/abdump

- acousticbrainz-dump-20150113/abdump/<tables>

put everything there, got it

and please add a - between acousticbrainz and dump

yep

ok, you're fully on the right track. keep on!

any updates on long term plan?

I need to run off to a class in any minute.

but I can give you a general overview before we write up a spec

but I might get interrupted. just fyi.

ok, so the idea is to crowd source building the training data sets, right?

ruaok: we're on break

alastairp: ok, I'll head over in a couple

10m or so

k.

Gentlecat: feh.

don't worry

Sorry Gentlecat!

I stole him

I've got that googledoc I can look at

if you're about in an hour or two, I can continue.

k. sorry.

and it's not like I can start anything before dumps are implemented

Leo_Verto: There's a ticket for that already.

ah nice

chirlu linked it for you a day or two ago.

thanks, I might have gone offline by the time it was linked or missed it because I didn't get pinged

ruaok: https://gist.github.com/alastair/c81114d2e9b489...

fitst question: is the low level data the same?

if not, is the source data mistagged?

yeah

about to start looking into that

though, same length...

possibly.