#musicbrainz-devel

ocharles: I don't have any objections to moving git to github (you'd need to add a commit hook thingy to post to mb-commits of course :P) my problem was with the code review stuff, there doesn't seem to be any way of saying I want to get emails for that

is mb not loading for anyone else? I can't connect to the main site or rika

seems to load fine for me

can't connect to jira or anything else either

guess I'll just wait

bitmap: seems to be working for me too, yeah, though it's been a bit now

it started working again a few minutes ago. very strange

who knows

networks be crazy stuff, I guess :P

ianmcorvidae: were you doing anything with multiselects still? 'cause I started writing my own jquery plugin for the instrument tree

I haven't looked at it, no

ah, okay

I'd still like to improve that interaction, but I haven't done the research into alternatives to chosen

the plugin I wrote is on my sandbox, if you try to edit an instrument or vocal AR http://bitmap.mbsandbox.org/release/ade577f6-60...

there's still a few bugs

haha

the dropdown also hides where the actual instruments end up

which is slightly confusing

not sure how to fix that, possibly put the instruments above the dropdown/search box

hm, good idea

heh, comment longer than ticket description for today: http://tickets.musicbrainz.org/browse/MBS-5054 :P

could someone help me a bit with postgres permissions?

maybe, what's the problem?

I don't have permission for the schema musicbrainz

I'm connected to the database "musicbrainz_db_static" with my user

plaintext: oh, yeah, meant to tell you about that

you want to connect as user musicbrainz

ooh

thanks!

ahh lukz

lukz: is this an appropriate place to ask about picard?

yep

lukz: nice. did freso already ping you? i was asking in #musicbrainz the other day. i'm working on the musicbrainz support for XBMC.

ianmcorvidae: one more problem: now I don't have permission for the table I have created earlier

I'm not sure, I wasn't on irc for a few days

should I just delete it, and add it again with the user musicbrainz?

xbmc recently got a musicbrainz scraper. I've just rewritten a lot of support for xbmc so it can fully utilise MBIDs

for artist, album lookups, etc etc

plaintext: possibly, if that won't take too much effort -- the other option is to just change the owner on it, presumably

lukz: but right now, we scan the tags, insert the items into the db and then go look up all the artist metadata afterwads

ianmcorvidae: thanks

that would be very difficult to change on the xbmc side.

so today we don't have anything good to insert for the artists

plaintext: but I don't know if you can change the owner without superuser permissions, so if it's not troublesome to just delete/recreate I'd do that

it's just test data, so I can simply drop the table

ty

lukz: cut a long story short, what i'm looking to know is if there would be any way picard would implement a tag to store the individual artist names

i.e. a 1:1 mapping with TXXX:MusicBrainzArtistID

night199uk: I'm not sure, but I'd definitely not want that to be enabled by default

an add-on might be okay

I actually think that looking up the MBIDs in the application is a better solution

yeah, i agree

and we do that

but the problem is the architecture is scan the MP3s and then go look up the data

there is no way I'm going to be able to change that in the short or medium term

which means right now we inject a bunch of artists named by their MBIDs

is that a problem though?

yes and no

we've not released the feature yet

it will go in shortly, maybe next week

I guess it depends how much time there is between the scan and the lookup steps where someone might accidentally see the not-actually-correct name

i'm trying to anticipate the support load. also we're debating the right way to insert these artists into our db.

some folks want to see us insert the artists with multiple artists named the same… i.e. for 'Mark Knopfler & Chet Atkins', we'd end up with 2 artists called 'Mark Knopfler & Chet Atkins' until we can scrape and download the correct data

right now we just inject the artists named by their id, so 'xxxxxxx-xxxx-xxxx' and then the users can easily see whats scraped or not

so at this point i'm just fact finding, i wanted to see if there was any appetite to do anything on the MB side

well, I'd guess that you don't have this problem for over 99% of cases

because there is only one artist

yeah, maybe

but i need special case code to detect that now

is this possible with an add-on today?

even if we add this to tags in picard, all existing files will not have the new artist tags and people using other applications to tag their files will not have them either

yeah, i know

yes, it's possible to write a plugin for picard to add the extra tags

but i guess picard would set the benchmark other taggers would follow

but it would be complicated

yer?

erk :-(

for example, if you have two MBIDs and two artist names, how do you know which one is which?

i guess i want something i can point people too if they are concerned about the behaviour

you won't need to create some hacks like prefixing the artist names with theMBIDs

well we'd order the tags in the same orer

we always retrieve the MP3 tags in order

I'm not sure you can depend on that

many tagging applications do crazy things

so MusicBrainzArtistID[0] == MusicBrainzArtistName[0], etc

not to mention, id3v2.3, which is a must for any windows user, does not support multiple values per field

well yes. I can only speak for xbmc, i can rely on that

indeed. how do you write multiple artist IDs for v2.3 though?

with the " / " separator?

I think so, yes

yeah, so then we can still rely on the order

xbmc already supports separating tags by the " / "

what if there is an artist with / in the name?

i'm just in the process of converting us to use tag lib now

yeah, thats why i'm migrating to tag lib so we have v2.4 support

i mean there is nothing 'perfect' with ID3 tagging :-)

thats what i've learnt from this process

but i'm trying to get a good general case

yeah, which is why I'd suggest to not use something else

er, *to use

?

as I said, for most cases you have only one artist

so you can safely insert that

in the cases when you have multiple artist IDs, why not use the combined name until you look up the IDs?

tags are really not reliable and picard is not the only musicbrainz-enabled tagging application

well thats the recommendation from some folks. but then we end up with 2 artists called e.g. 'Chet Atkins & Mark Knopfler'

and you need to handle the case when the user doesn't have the tags anyway

well yeah. i have to start somewhere though :-)

and yeah… but i'm trying to reduce the number of special cases

well, by adding this, you are increasing the number :)

if picard were willing to support this, then it'd be at least something i can point users to that complain about one of those two previous options

nah. if you were willing to add this i can just point users to upgrade picard and retag :-)

probably an addon is the way, then, I'd say

can't scan the file and only insert the ids into the db after you have looked them up?

I definitely do not want this in picard by default, because it solves a very specific problem that most poeple do not have

luks: i will look at that over the longer term, but thats not the way its architected today

i can understand your case

okay, maybe i will look into an addon

you think an add-on for this would be complicated though?

i'm hoping we don't get too many complaints, i don't think users will notice if our scrapes work fine. but i just want to pre-empt the users that moan about it, i'm sure there will be some.

night199uk: I think the add-on should be pretty easy

okay

thanks for looking

when looking up an mbid in the database, do I have to know what entity it is?

or is there an easier way than lookin in all possible tables (artist, release, etc)?

you have to know the entity type

oh

since we can't really tell by the splunk query, the script might have to look in all tables then

or we can pass some additional info in the YAML file, telling the script which table to look in