#musicbrainz-devel

worksforme

yay

ocharles: prod

Have we broken tracklist edit display again? http://musicbrainz.org/edit/14852318

ianmcorvidae: have upgraded to the newest chrome 14, now I see the timeline. So definitely not your bug.

:)

hello!

warp: can you please see if fixing whatever happened with tracklist edits now is a straightforward, simple thing, or we have to wait until Ollie comes back?

(also, hi)

Hello, warp.

(I wasn't raised in a barn.)

montana, barn. I would disagree

Western Montana isn't big on livestock. It was a shed.

reosarevok: any idea if it can be reproduced on test?

meh, I should've imported the database yesterday.

FiveOhTwo

Test is dead

I 2 502.

I S2 502 (no)

ugh, fun.

Yep :(

looks like it started again now.

"recording_id": null

I don't think that should be possible, http://www.frob.nl/tmp/edit-14852318.json

Hurray!

Is http://musicbrainz.org/edit/14857888 fucked up too?

Leftmost: http://www.frob.nl/tmp/edit-14857888.json

Ahem. I apologize for my language.

Hmm. Maybe two different bugs, then.

Leftmost: does the .json look correct in edit 14857888? All the names are changed to "Alright!". (if the json is correct, that is "just" a display bug)

urgh.. jira uses picture smileys in the comments :(

reosarevok: http://tickets.musicbrainz.org/browse/MBS-3043

warp: yea, that's what you get for turning wiki syntax on :S

ocharles: master keeps shouting "User attempted to edit but is not authorized" at me

Let's mob ocharles and make him do our bidding.

ocharles: I already logged out and in again, so my session should match the currently running code and database.

sorry for our a bit rugged dev chat part yesterday

ugh. smilies

what did we need wiki syntax for anyway?

hey guys, been out of the loop for a bit, is there a new vmware image with the latest everything ? can i get the link

would it happen to be ftp://ftp.musicbrainz.org/pub/musicbrainz/vm/Mu... by any chance or is there something newer ?

do you properly check when some submits a information service (e.g. MySpace) URL?

zazi: in which way?

that only information service valid URLs can be submitted

e.g. for MySpace http://myspace.com/[username]

re. my question, see also https://github.com/BarryNorton/D2R-LinkedBrainz... and https://github.com/BarryNorton/D2R-LinkedBrainz...

zazi: you can't assume anything about the URLs in MB

okay ;)

we have to deal with these issue right now ...

validate the format with a regex if you want to extract the username

it might be appropriated to use forms where one only have to fill in the username for the specific information service

I don't want to extract the username, I want to do transformations that are based on a consistent base URI

(of the specific information service)

all MB's URLs have to be "normalized" before you use them like that

even for wikipedia you can find URLs with different encoding (%20 vs _)

someone who possibly remembers: http://wiki.musicbrainz.org/Next_Generation_Sch... says Arranger should exist for both artist-recording and artist-work, but http://musicbrainz.org/admin/linktype only shows it for artist-recording

why is this

looks like the devs didn't look at the Track Relationships Conversion page after writing migration scripts

since re-arranged version doesn't exist either

the arranger stuff is a complete mess :(

yeah

we had worked through this in the track relationship conversion meeting

apparently nobody wanted to implement it though

luks: I think I've discovered something: when I apply http://paste.pocoo.org/show/442411/ I don't seem to run into file loading issues anymore. do you remember why the queues are set up like that?

bitmap: to be able to use more than one thread for loading

ah, I see

but maybe it doesn't make much sense

luks: would having only one queue per thread, but letting two threads share a queue work?

I don't know understand

oh, you mean merging the load/other queues?

to be honest, I don't remember much of the code

I think other handles directory listing, load file loading

if you merge them, it will first fully load files, instead of listing directories first

anyway, I'd try one queue per thread and then maybe adding one extra thread for the load queue

hmm, yes, that's what I was thinking

I can't really see why would that fix the problem

ah, right, it can

if qsize() returns something >0 and in between the other thread takes the task the thread will be stuck on the non-primary queue

then no thread will be waiting on one of the queues and whole picard is stuck

it's probably best to get rid of the multi-queue code

I agree with that, it's rather complex to debug too

luks: I don't actually see anything being added to the load_queue anywhere

only to the other_queue

"grep load_queue -R ."

:P

./picard/file.py: self.tagger.load_queue.put((

aha! I missed that one

picard's classes are really badly designed

I didn't know much about GUI applications when I was writing the code

well, at least the code is readble/understadable

the only part that scares me is the itemviews.py code :)

it's nearly 1000 lines long now, in fact

luks: having two threads each for the load_queue/other_queue seems most responsive I think

http://paste.pocoo.org/show/442450/ is what I've changed right now

ianmcorvidae: re arranger on work, it was never supposed to be automatically created by migration, but rather done manually after NGS release

murdos: the AR isn't available at all

nobody can do it manually, it doesn't exist to be used :)

I know, that why I've created http://tickets.musicbrainz.org/browse/STYLE-73

ah, wonderful :)

hm, why didn't you just add it yourself?

I'd rather let our beloved style leader do it, it would be less controversial

nikki: you're still paid by MB?

yes

...

"Internal Server Error"

how many hours per week? (or month?)

it's not actually defined in hours

I guess your pay is included in the "Expense - Development" category, in monthly finances report

possibly, you'd need to ask rob about that

did you see that brian commented on stuff on the wiki again?

no... was he logged with his nickname?

nope