#musicbrainz-devel

haha http://www.musicbusinessworldwide.com/beyonce-r...

ruaok: hi

hola

how did you get your tags from musicbrainz to build this training set?

from the db, I know that

were they recording tags? or release or artist? do you have the sql still?

I logged into some random server with a current DB and started wriring SQL.

release.

lemme see

nothing in my notebook, sorry.

interesting spam to info@: https://gist.github.com/mayhem/5c6fa9267ecbd8db...

alastairp: I do recall the query being simple but huge. :)

ok. thanks

huge only because you went "AND gid IN VALUES(all the mbids)"

in fact that reminds me that I probably wrote some code to support that.

ding.

oh!

I have *code*. want?

yeah, sure

anything would be good

there is not if __name__ == "__main__" in there. you sure?

I'll let it slide

sigh

I broke my server

net connectivy is really good again today.

doesn't like something the size of ab

I'm genuinely not sure if you're being sarcastic or not

sarcastic.

ok. I thought that was the case

I think.

better go read some code, because some wanker didn't write a lick of documentation before he got distracted by shiny

yeah, I just glanced at it. that seems about right

so is there one to go from all tags -> genre tags?

I did that part by hand/random sql statements.

can you remember if you used all tags as your starting point, or all tags from things that were in ab?

I ran a query to find the most frequent tags.

then I hand selected "clusters" to feed to the first script.

ok, fine. I can drive from here. thanks!

yeah, its pretty simple stuff, but should give you a starting point.

man. cut(1) is so fast

man, is the french air traffic control strike affecting the net too?

even on 2 million rows

way better than loading it all into python as csv

Is there any reason, off the top of anyone's head, why the MB oauth provider couldn't be modified to accommodate password auth?