#musicbrainz-devel

warp, you around?

or luks, you may actually be the best to ask - trying to figure if this is a schema thing (ie, per design), or a bug

warp around at 3am, you'll be lucky :P

lol

well, fatih noticed this when testing track merges re the [Disney]/HSM edits.

http://test.musicbrainz.org/release/142ecbb8-0b... ; look at track 2.

faith merged three different Fabulous recordings on different HSM releases, then changed the artist for the recording from Ashley Tisdale & Lucas Grabeel to [Disney] ( see http://test.musicbrainz.org/recording/abedc027-... )

The recording shows [Disney] correctly. However, the releases themselves still list Ashley/etc as the track artist (not an AC to [Disney], as far as I can tell), and there seems no apparent way to change the artist for those tracks (instances of that recording)

Isn't that correct?

you change them in the release editor

the recording's artist was changed, rather than change the artist 1xper each instance on a release.

This lets a release and it's tracks accurately store what's on the liner, while the recording has the normalized version.

so shouldn't that artist change on the tracks be consistent with the recording?

It does make it a pain to change the track artists though.

right - so I'd expect, at most, Ashley/etc to become an AC to [Disney], not the track and the recording to each now have an entirely different artist

and there's a ticket in jira already for being able to update artist credits when merging

I've got to run, be back in 30 min or so.

I think you should probably be asking acid2 actually. he's the one who did the merging edits as far as I know

makes sense to me --> Foo is a recording by Bar, Foo1234 is a track by Bar which is an instance of the recording Foo

although I just woke up again, so I can't really think right now

still makes sense to me --> Foo is a recording by Bar, Foo1234 is a track by Bar (playing as Zip) which is an instance of the recording Foo

doesn't make sense to me --> Foo is a recording by Bar, Foo1234 is a track by SomeoneElse which is an instance of the recording Foo

Think of that Quee & David Bowie song that is credited to different people on different releases.

So it's intentional? Doesn't that mean that, rather than simplifying artist changes by centralizing them all to a single recording, we're actually multiplying them, as now there's the change on the recording, plus a change on each and every release (which means potentially far more releases to be edited, with the new definition for a release)?

I'd think of the Queen and David Bowie case as being recording: "Queen" & "David Bowie", track1 "Queen" & "David Bowie" (as AC "Queen"), track2 "Queen" & "David Bowie" (as AC "David Bowie")

you can't link an artist in an artist credit and not include any text

as far as I can tell, there's a bug in the edit recording stuff because all http://test.musicbrainz.org/edit/12710294 did was edit the name, not the actual linked artist

so b/c of the bug, the recording is still linked to Ashley/etc, just with the AC for that artist being [Disney] on that recording?

I remember a similar bug where you couldn't change the first artist of an artist credit, only the credited text, so it should be somewhere in jira

maybe, it's hard to tell because there was no change to the actual artist that was linked

hmmm, yes, it's still the same artist MBID, just w/a new name... wierd

and that [Disney], whereever it appears for that recording, still links to Ashley/etc

*shrug*

I think it's the same thing as http://jira.musicbrainz.org/browse/MBS-320

that one mentions track artists, but back when that ticket was created, changing track artists used the same artist credit editor as the edit recording stuff does now

hmm...

it was on one of the versions, I thought

hm

oh wait release relationships

probably the track one you were working on ;)

never mind :)

yyyeah I may be a bit tired

it should probably be a release group one anyway

luks too

hi

hey, are you free for a brief chat? it's about editing tracks and separating tracklists between releases

I've written up a document that explains the problem, just finishing that now

it's at http://gist.github.com/593476 for those interested

what is the problem?

I think it's only an UI problem

there are some old notes here http://wiki.musicbrainz.org/TrackMerging

it's a problem of track ids changing in edits, see the scenarious in that document

the only realistic solution I was able to come up is to ask the user what to do

track IDs are not important

they are not shown anywhere

tracks itself are not important

tracklists are

but they are needed for the edits to know which track they are editing

Say I'm renaming a track, and I don't want it to affect other tracks. The edit needs to know somehow which track I'm refering to, and the track id seems the only safe identifier

ah, I think I can't help much with that

as in you're stumped too? :) or you wouldn't get into this situation?

as in I always wanted to avoid this because there is no good solution

ok, I'm all for avoiding it too then

I said many times that I'd prefer to not release NGS with the current structure of edits

we heard you

but we're running rapidly out of time

the only alternative I had on my mind was to have Edit Tracklist as the lowest edit type

Meaning that if I rename 2 tracks, we only get 1 edit?

yes

yea, me and Kuno discussed that, but that doesn't seem good - it's not at all granular

but maybe that's the best we can do for now

the problem there is that someone else cannot e.g. vote no one of the renames and yes on the other one.

but perhaps that's not that much of an issue.

no matter how you do it, using this kind of edit structure for NGS is a hack

so I guess it's always only temporary solution

I think the fastest way to get this out the door is with an edit tracklist edit

fewer edits to vote on \o/

nikki :D

I mean who added it originally and why? Was it requested?

It seems like an odd setting, I've never seen any other sites except banks ask that kind of question.

I'm sure I've seen it a few times

Okay fine, banks and a few other random sites. But do we really need it?

my guess would be that djce added it because he's all security-conscious or something, but I can't be sure

I can't remember *not* having it, which would suggest it was still djce mostly coding when it was added

+1

so it was you?

Sounds like me.

hehe

heh

djce: Would you object to not including it in NGS?

It's also nice and difficult for me to test it actually works :)

Fine by me.

One way to test it would be to log in, dick with the nginx config so that the REMOTE_ADDR is fake, then see if your cookie still works.

Then revert nginx conf. :-)

Yea, or use my server as a SOCKS proxy

yeah :P

aCiD2: Can you remove the single_ip code from r848 please (should I remind you on the review?)

navap: please remind me on the review

When I finish this branch I'll ship/amend those reviews you've commented on

hehehe http://bugs.musicbrainz.org/changeset/3916

Nice work :)

ha :-)

djce: Did you get a chance to look at http://jira.musicbrainz.org/browse/MBH-34 ? Did my config help at all?

Can't remember really. I didn't spend much time on it; IIRC it didn't work straight away, and I probably put it down and didn't go back to it.

The short (ish) answer is,

if we can take a few hours of wiki downtime,

it becomes a lot easier.

Trying to make this change with the wiki still up is hard.

Not having a staging wiki and all.

I'm sure people will survive if we do take it down

Yeah, might just have to do that one day.

I say go for it, would it be possible to put up a temporary index.htm saying "the wiki will be back soon" or something?

Affects the live site too, of course.

wikidocs etc.

yeah, I was about to say

Ah right, /doc won't work.

I wonder if it'll behave nicely or just go splat

Getting the api working is pretty important. We can't wait till NGS is being released to take the wiki down and figure this out because we need to make sure there are no issues with my wikidoc code.

djce: I'd appreciate it if you take a harder look and figure this out :)

Of, that I have no doubt :-)

We'll need something like it to be able to "clone" a release from within the release editor.

hmm?

How will a user add a new release event? If they do it through the release editor they should be able to specify that they want to reuse an existing tracklist.

I'm not talking about that (and we don't have any duplicate-this-release functionality to keep)

I mean the thing which says something along the lines of "this release is similar to these releases, are you sure you want to enter a new release?"

the number of times I've seen that bloody screen in the last few days >_<

hello ruaok

you guys being productive?

reasonably

djce: yes, that seemed to have been the right amount of great beer. :)

I just met with the Radiotuna guys from Brighton. more beer. yum.

ruaok: aCiD2 is going to have a look at my release editor branch now.