#musicbrainz

Immortal Vinyl Records; Virgin Records America

with 2.3 picard thinks it succeeds, but windows explorer and winamp show the old labelname

with 2.4, windows explorer shows no labelname at all and winamp shows "Virgin Records America"

weird

do you have delete existing tags on?

maybe winamp still reads the old 2.3 tag?

no

no @ tags

hm

no i let winamp read the metadata again

deleting tags does not help

oh wait

anyway... the point is really with the files, not the way I read them

well,many audio players don't support multi-valued tags?

because basically, when I exit picard

and restart it and let it read in all the files again

then it should show the value it last showed, but the recordlabels are back to old instaed

murk: They are mp3s?

yes

does it happen to the other tags too?

or ONLY recordlabel?

only duration/date/recordlabel it seems

murk: If you don’t have “delete existing tags” on, it won’t delete the existing tags.

So that would fit with reading back the old values

you should also check if winamp supports multi-valued tags

from my experience the least audio players do

Besides which, it sounds like it does read the label info right

You said it was “virgin records america” and “immortal vinyl records”, right?

in reverse order, but yes

Right, but those are the two labels

both before and after

well, it seems like my annoyance is resolved now through enabling 2.4 and 'delete existing tags'

bute before seperated by "/" and after ";" ?

on restarting picard i no longer see *'s everywhere i saved

noobie: that’s just a difference in how Picard can read them back.

most audio players will detect / as a single value though

If I recall, the “;” is used for multi-value tags, while "/" is used in the raw tag when the file format only supports single-value tags.

oh ok :)

yes if I click to edit the recordlabel in picard

it shows two separate entries

the semicolon is what picard uses as a delimiter

murk: But only for display.

Unless you have a specifically defined delimiter in the options

Modern formats actually allow multiple values for the same tag.

so it doesnt support it :)

they argument that 2.4 is hardly used

But, before I forget it: Thanks a lot for your help :-)

murk: id3 is a terrible, terrible format. :-)

metaflac \o/ :D

there's not a way to maintain the current view picard has of all your files, is there?

so that when you close and open it that it will show the same thing as before

what do you count a s current view?

the files loaded in picard?

yes

guess not

the file explorer in picard saves it's last location i think

It would be a cronjob on my server

no command line support as far as i know

yep, I know :(

etu: We all would, I think. Or at least most of us

hawke_: that could be a point for picard 2.0? Make the gui optional?

etu: Maybe. Ask the Picard devs. :-)

[quote name='weston123' timestamp='1342808274' post='16643223'][list]

[*]In-game name: weston121

I know someone was talking about decoupling the gui from the backend.

oop.s Sorry. >_>

*Oops

that'll be awesome

hawke_: THere's an open ticket for that.

noobie: ^

Sec, I'll find it.

oh, then I sign up for updates on it

it's a pain in the ass to do, but yeah, several people want it

I don't remember if I actually opened the ticket but I know I considered it :P

(the issue is that at present Picard mostly uses internal Qt objects for data, HTTP requests, etc.)

might be easier to write a cmd tool from scratch

:o

noobie: maybe yes...

"all" it needs would be to fetch the data from the db and pass them to mutagen

noobquestion: what's the purpose of 'updating' already tagged tags? :$

heh, yup, it was me that reported that

files*

murk: the data in the database isn't strictly constant

murk: people add things, correct things, etc.

so occasionally it's good to update tags to bring in those changes :)

well,it should be possible to tag new files too

i see :)

ianmcorvidae: I think we would need something like NES to be able to do that reliably, no?

hm?

hawke_: I just mean manual checking occasionally, not automatic looking at if changes have happened or whatever

and obviously changing picard to be a daemon-client model is basically untouched by any database stuff

and you don't need NES for automatic checking either, but I doubt anyone will look at fixing it before then anyway

O_O daemon-client model? That seems excessive.

I just meant the ability to track merged/split entities.

yeah, which I didn't talk about at all :P

Where an MBID becomes no longer valid.

which is why I was confused

In order to update the tags, you’d need to do that. :-)

tracking merging/splitting is all server-side

we already do that :P

well, merging anyway

Not reliably.

reliably unless people delete instead of merge

NES doesn't really make a difference there either, except *possibly* for splits, which we currently don't really track

That I can’t answer, but I know from time to time I try to load a release in picard and it just says that it can’t find the release. Yet I look it up and it’s there.

and no, to update the tags you do the same thing you do when you match them to start :P

I have never checked whether it was deleted though.

otherwise it's the automatic stuff I mentioned

it's probably deleting rather than merging

when there are merges entries get put in the gid_redirect tables and it'll redirect you to the new version

Like I said, I never investigated further when that happened: maybe it was deleted, maybe it was a transient error, maybe the gid_redirect doesn’t always work properly, I don’t know. I just looked up the release on the web page, loaded it in the tagger, and went about my business.

Maybe I’ll look into it more next time it happens

yeah

NES won't fix the delete-rather-than-merge problem :)

You sure about that?

It should allow us to go back into the edits and at least find the damn thing.

Instead of just “sorry, it’s gone forever, no evidence that it ever existed”

depends how you implement deletes in NES

which is how a lot of stuff in MB is now: it’s really hard to find the history of any given entity.

it's easily "delete the master pointer for this" and then you have some sort of garbage-collector that'll delete unused trees/revisions

which is basically equivalent to how we do our current trimming of the DB automatically when things are unused

True.

But I wouldn’t think it would also delete the edit history associated with those trees/revisions.

there's only trees/revisions :)

there's no separate edit table or whatever, it's all tracked in terms of stuff done in the entity tables themselves

There is edit history; there is a place that comments and votes and everything are stored.

there's merge requests, but those probably also get deleted when they aren't referenced (however deeply) by anything that's a master revision

a merge request is comments, votes, and a pointer to a revision, basically

ocharles: "EXPLAIN and log analysis is about as far as you get with a code profiler"

ocharles: There's the pg_stat_statements contrib module for profiling procedures: http://www.postgresql.org/docs/current/static/p...

(FYI)

any eta on NES?

Next year some time, I think.

Anyone have any advice on how to title http://beta.musicbrainz.org/release/734443ab-44... ?

I wouldn't hold your breath, it could probably take unbounded amounts of time

ianmcorvidae: It seems like NES has a more concrete plan than NGS at least. :-)

hawke_: except for the last 10%, which always takes 90% of the time :)

“5 great tone poems: Including foo, bar, baz”? “5 great tone poems, including foo, bar, baz”? “5 great tone poems: foo, bar, baz”? “five great tone poems” (as it is now)?

hawke_: I'd keep it as it is, I think, I don't know that that's supposed to be a subtitle