#musicbrainz-devel

Anyone around here who may be able to help me? Can't get musicbrainz server to start: http://pastebin.com/bth9nEMK

ocharles: around by any chance?

Afraid he's on holiday for a few days

shucks.

kepstin: heh, thank you, thank you. I'll try and slip an 'eh' into there :)

heh

I notice Picard has '>' and '<' as translatable strings, are they supposed to be there?

kepstin: Do we spell it catalog or catalogue?

navap: I've been using catalogue

(the only thing I've really done so far was replace all the 'catalog' strings I could find with 'catalogue' ;)

You missed one ;) https://www.transifex.net/projects/p/musicbrain...

oh, I've only looked at server

picard was added since I last checked :)

Ah, I've only looked at picard

feel free to do stuff like put in fancy punctuation “”… in the translations

huh, those < > strings are weird

everything has probably changed since you last looked :P

ianmcorvidae: a bit, yeah :)

navap: http://i.imgur.com/DBrcr.png

that's where the < > are from

heh

What happens when source strings change? Do the translated strings get removed?

hmm. most places turn them into 'fuzzy' strings, which are left in place, but marked as needing review

I dunno how transifex handles it

I just came across five instances of 'other' for server

That's not right, right?

mmm, server has some redundancies in it.

lol @ http://www.btb.gc.ca/btb.php?lang=eng&cont=690

that's the official list of country names in Canadian English and French

kepstin: yeah, we have lots of redundancies

navap: there should be five in attributes

because of case differences in other languages, they're from different tables

in server, unsure

incidentally, transifex tends to break things when source strings change

Ah, you're right they're in attributes

which is why we back up everything hourly :)

yeah; if you open up the details thing you'll see they have different context strings

(which are tables, in this case)

the eo translation actually has slightly different translations for some of them, so it's even been useful :P

in any case, gettext doesn't actually even *allow* duplicates unless they have different msgctxt strings

so if you see duplicates, they're either distinguished somehow, probably being stupidly hidden by transifex, or they're in different domains

(there's probably some redundancy between statistics and server, but probably not anywhere else unless someone put a literal name from the DB into a string somewhere, which would be stupid)

(of course, with and without . or : or whatever is still redundant, but different enough gettext won't complain :/)

luks: do you think it's worth deciding what to do with release types in picard before 1.0?

I was told to find a mbcontrol script here: http://paste.pocoo.org/show/555245/ but it no longer exists?any ideas?

I was told to find a mbcontrol script here: http://paste.pocoo.org/show/555245/ but it no longer exists?any ideas?

bitmap: I'm fine with releasing 1.0 without that

shit?finally got musicbrainz installed but my mirror aint showing some updated info.

Do you have replication set up?

(it will always run 1 h behind the main server, in any case)

I think so? :\ To be honest, I'm not sure, lol?.

Just need a cron with "carton exec -- ./admin/replication/LoadReplicationChanges" right?

I love you bro.

SexyKen: what finally got it working?

ianmcorvidae: You betcha! you are the man. updated cpan and started all over. a bunch of shit still didn't get installed by carton, so did all as "carton install bologne" and the final test?launching?and it worked!

ah, cool

so I guess it was just some weird cpan dependencies problem on your particular platform

at least you got it working finally :)

!

Just gotta get updates to load auto and stuff

Any ideas for auto start and update? :)

for update, you should be able to add 'carton exec <absolute path to /admin/cron/slave.sh>' to run at 10 minutes after the hour

in crontab

LoadReplicationChanges would also work, but slave.sh has better logging

Hmm, logs rotated or no?

My wife wants some info for Flo Rida from 2012 but mu mirrors not showing anything

luks: I think I fixed all the tickets I wanted to get in, so we can release any time today or tomorrow

bitmap: how do you feel about the copy files patch?

I haven't looked at it yet, I can do that now

no, not necessary

I'd rather not include it in this release

and I don't have any other changes on my mind

okay, well I'll look at it later anyway :)

I should mention the new tag comparison in NEWS, I don't see it there

how can I tell plack to listen on 127.0.0.1 not 0.0.0.0?

alastairp: probably with the --listen directive

yeah, just found that

alastairp: but search 'plackup' on metacpan, that should give you everything you need

wondered if I could put it in the .psgi file

nah

for rika I just wrote a script to run the servers

rika is also using fastcgi and such of course, but

right

hmm, maybe you can help here

so, I have plack running the devserver

on :5000

and nginx on :80

ah

I'd recommend using plack with FCGI on a unix socket, if you're passing to nginx

ok, sure

look at the sample nginx configuration for beta

and will that still reload the site and stuff on changes?

for the nginx end

"for beta"?

if you pass -r to plackup it still will

cool

admin/nginx/002-beta

what's the args to plackup to make it fcggi?

-s FCGI

thankms

and --listen for that can take a file path (for UNIX sockets)

rika's run.sh is just: carton exec -- plackup -Ilib -s FCGI --listen $HOME/.fcgi.socket -r

(well, after a cd into the proper directory)

2012/06/02 13:03:20 [error] 1556#0: *1 FastCGI sent in stderr: "[error] Caught exception in MusicBrainz::Server::Controller::Root->begin "Can't locate object method "forget_dead_hosts" via package "Cache::Memcached::Fast" at local/lib/perl5/Cache/Memcached/Managed.pm line 765."" while reading upstream, client: 174.91.230.205,

does fcgi do memcache differently?

no

don't use Cache::Memcached::Fast for sessions

well, I wasn't!

:)

get rid of the "memcached_class => 'Cache::Memcached::Fast'" that's in SESSION_STORE_ARGS

the Cache::Memcached and Cache::Memcached::Fast interfaces aren't the same, sadly

so, what differed between the 2?

meaning, I'm pretty sure I wasn't seeing it when plack was hosting directly

ok, sorted: http://musicbrainz.music.mcgill.ca/

thanks

shouldn't really make a difference, maybe you hadn't logged in before?

but cool, anyway

nah, it was when I was just hitting /

shrug

luks: should the version be updated to 1.0 final?

yeah, who knows -- I don't remember how Cache::Memcached::Managed decides if it needs to take the branch that hits forget_dead_hosts

may be some incidental thing

ianmcorvidae: ah. that might be it