#musicbrainz-devel

Yes. Dunno how we can improve that yet though, i think we could get rid of UI_LANGUAGES somehow.

I think you can set a minimum level of translation percentage in the .transifex file for it to pull translations.

Using that, it could build all the translations available.

And it would also make it easier for people to test new languages, as they could literally just drop in a new .po to compile.

Though I guess Picard would have to discover it somehow and know the name of the language for it...

Yes, i thought about it, but i'm reluctant to limit by percentage, because i think if people can see it is badly translated they will not contribute to translation.

Well, if you want that granular control, I don't think there's a way around having a custom list. Whether you call it UI_LANGUAGES or something else.

Hmmm, http://en.wikipedia.org/wiki/List_of_ISO_639-1_... has native names for languages as well as iso--639-1 code

even better http://en.wikipedia.org/wiki/ISO_639-1_language...

zas__: wouldn't that make people *more* willing to contribute to it?

I mean, if it's well translated there's nothing to contribute to :p

reosarevok: it is what i meant ... ;)

s/can/can't/

how come we have sco.po ? https://www.transifex.com/projects/p/musicbrain...

sco isn't a ISO 639 code and all translations are at 0%

Freso: https://www.transifex.com/projects/p/musicbrain... --> you should know ;)

zas__: Uhm. What? When did "sco" stop to be an ISO 639 code?

well, my bad, sco is a valid ISO 639-2 code, but translation is totally empty

Sure.

One of the reasons I've neer pushed for it to be added to UI_LANGUAGES.

:)

Or *nay

Man, I miss messing with Scos.

*Scots

Also, my typo level is high. I need to go to bed.

Same here.

Have a good night.

is that something we did?

alastairp: http://beta.musicbrainz.org/edit/17806039 editor Mohana

suspect it was a mistake though

right, thought so. I should have checked that :)

also. Other is qaa?

that explains why on my TV I can change language between Spanish and qaa. I always wondered what the qaa was

qaa-qtz are local use, qaa is our code for artificial languages I guess

"Original audio", it seems: https://mailman.videolan.org/pipermail/vlc-deve...

since we don't include the 639-5 codes

(choosing qaa makes it go back to english for dubbed shows [or french, german, etc if it's dubbed from that language])

nikki: finally getting people to look at attributes

there's mumbling that we should have attribute aliases. I'm trying to convince them that we don't need them

yay

we don't have a way to do aliases right now anyway

right. and if it's a fixed list, the problem of dealing with different spellings goes away somewhat

I mean, maybe we *should* have them

But for now, they'll have to live with that :p

alastairp: the timing on your elasticsearch comments is uncanny.

I inserted 25k documents and elastic search never stopped consuming CPU.

now trying solr as a logstash backend. hopefully it uses less magic

ruaok: hah. I feel vindicated!

the downside is that the fancy kibana interface isn't going to be available.

I distinctly recall giving you a funny look when you said geordi used ES, ianmcorvidae ;)

I'm currently moving bits of geordi into postgres :P

there we go

:)

yeah, that's funny. the freesound guys use ES for logs, and don't seem to have a problem

the scale is pretty different for geordi anyway though. hard to grow it huge amounts at once

ianmcorvidae: we may want to consider using another store.

alastairp: that magic may be with them then.

ultimately, what I want geordi to be using ES for is search. which seems rather appropriate

right, comes back to the doc storage/doc search ting

Currently containing 2.749.329 messages. Oldest message is from 10.11.2013 - 14:14:50. Stored 50 messages in the last 10 minutes.

the current plan does keep it in a reduced doc-storage role though, which we may need to revisit

but once it has a real database to move things to that's easier anyway, so :)

I wonder how big the messages are

reosarevok: right, but adding aliases to attribute values, that makes it almost an entity in its own right

I thought aliases were supposed to be light-weight :)

er, s/aliases/attributes/

wow.

apparently ES never finished processing everything that was stuffed into redis.

fired it up with solr and its going to town. :)

nice

that's what I never got, because they both use lucene

this is the sort of performance I was expecting.

one apparently uses less magic. :)

right

please review https://github.com/musicbrainz/picard/pull/209

zas__: this happens after filename shortening: https://github.com/musicbrainz/picard/blob/mast... :)

xlotlu: what happens ?

zas__: it creates files like "filename (3).mp3" if one with the same name exists

it adds extra-characters, which is bad. :)

i'll leave a comment in the code that it creates a bug. i intend to consolidate all that code under filenaming, but if i don't get to it this week it won't happen too soon

xlotlu: it will be great if you can fix it, i'm not knowing this part well enough

ruaok: if you can pull this off it'll be kind of hilarious because the community has been trying to gather this kind of data for years

I'm taking a very Brewster Kahle sort of approach to this. :)

Yeah. I guess academics are naturally cautious

its been on my radar for quite some time. but the time is ripe now.

The things that are happening now are going to be pretty awesome I think

I agree with that.

Got it a year or so for it to sink into the community and we might get some great stuff

things might be really fun a year or so on. ;)

snap

:)

Though, I would be surprised if it takes more than 'a' year ;-)

its like what we spoke about yesterday. there are some great tool about that make all of this much much simpler than before.

Science!

doesn't look like it :P

and with Xavier at the helm, open source is going to win.

Not just at the mtg

Qmul, McGill, ccrma, .....

But I guess in academia it's cool?

ruaok: we need to get you to an ismir

Taiwan this year ;-)

alastairp: heh. not sure about that.

reosarevok: not just academia.

Yeah, you'll get stabby

ruaok: well, it seems useless for the average user and you can't sell it to companies, can you

but we get 10 good ones.

win!

Hmm

Ok, I can see that one

Of course, without an evolving dataset that'll stagnate quick

So you'd need to convince people to scrobble to you :p

We've started citing mbids in our papers

alastairp: cool!

Need to convince others to too