#musicbrainz-devel

or id = any(array[1, 2, 3]) perhaps, I forget array syntax

ah, right. '{ works.

cool

so how is that different from id IN (1, 2, 3) ?

warp: to do that you need to generate n placeholders, and then provide n parameters

I find it much more convenient to provide 1 query parameter

ok, so that's the only difference?

it's especially simpler in this example though, because we repeat that set of ids twice

sure

so that code would get quite unweildly having to call ' . placeholders(@ids) . ' twice

ocharles: I'm still not understanding something, that delete always seems a no-op.

it will be a no-op if there are no release group types to merge

if you have rg1=(live, remix) and rg2=(live), one of those live attributes will be deleted

What we're doing here is filtering changing the bag of release group types over all release groups being merged into a set of release group types, and then we can safely move each release group type to be on the 'new' release gorup

delete from release_group_secondary_type_join all those rows for which the secondary type does not occur within itself.

I'm parsing something wrong, or the query is wrong.

oh, the query is indeed wrong (:

the where clause should be on release_group_secondary_type_join (secondary_type, release_group)

but distinct on just secondary_type

So...

WHERE ... AND (release_group, secondary_type) NOT IN (SELECT DISTINCT ON (secondary_type) relaese_group, secondary_type ... )

ok, now I get a result when I select the query.

thanks :)

cool

we got there in the end :)

thanks for reviewing my review :P

haha, np.

in the edit table, what do the values of status mean ?

im seeing 1 and 2

VectorX: read MusicBrainz::Server::Constants

whats the location of that

VectorX: lib/MusicBrainz/Server/Constants.pm

oh, hehe

no wonder google didnt work

warp: heard of ghostery? https://www.ghostery.com/ might interest you

ocharles: yes, I've been using that for a while now.

cool, I only just found out about it, it's impressed me

ocharles: I used to run it with the blocking functions disabled, because I mostly wanted to know the crap they put on pages.

ocharles: but after some outages / dodgy internet connections I now have it turned on so pages don't stop loading if a thirdparty is unreachable.

oh, hadn't thought about that

I was curious about the tracking too, but I just went straight with a block

i already have flashblock so i used to pages breaking at first

I wasn't using flashblock until ubuntu 12.04

before I was using ubuntu I just didn't have flash. which is fine. in ubuntu 11 I just left it on, I never had problems with it.

heh

but in 12.04 there are weird rendering issues with flash, like the plugin rendering even when a different tab or application has focus.

i did try NoScript for a while, but it just breaks too much

some fun things where flash is written in gtk-2, and my browser is gtk-3, so if flash /did/ load the browser would crash instantly :)

I like javascripty things, so I'm definitely not going to run with NoScript.

I'd like to run LibreJS the way I used ghostery, but it doesn't have any option yet to only notify instead of black.

s/black/block/

ocharles: do we have a function to format those arrays for the any() call?

warp: you don't have to do any formatting, just pass an array reference

oh wait, I'm passing in the array instead of a ref to it.

DBD::Pg will do the work for you

yeah, my bad. nm

I did it correct on the DELETE, but messed up the UPDATE :)

carton exec -Ilib -- prove -v t/edits.t :: --tests ReleaseGroup::Merge

now passes, yay.

heh

DBD::Pg::st execute failed: ERROR: duplicate key value violates unique constraint "release_group_secondary_type_pkey"

DETAIL: Key (id)=(1) already exists. at lib/Sql.pm line 103, <FILE> line 1.

I always get this kind of stuff, and I never understand it.

warp: if you run all tests you have to create_test_db first

if you only run edits.t you should only have to run it once

ocharles: I often get it when changing tests even after a create_test_db run. in this case the sequence wasn't a problem, I just mucked up the sql, hence "jumps to conclusions" :)

heh

ocharles: btw, the artist and label merges probably need similar changes to what I'm doing now.

(in Data::IPI::merge)

yep

wanna go ahead and do that?

but I'll do that in a separate patch.

sounds good

yeah, sure. I'm going to push this to code review now, then I can do those.

bah. new code review is very buggy.

deleting your comments too?

yes.

well, I pasted my comment, and I need to trigger a keystroke in the textarea before code review is aware that there is data there. just the paste event is not enough.

yea

i found that

I had some issues before where certain keys just didn't work in the textarea, I couldn't type some letters.

(but I just wrote/deployed an ~/.XCompose file which is also causing some issues, so I'm not always certain where the problem originates :)

ocharles: The image IDs for cover art, I assume they are always unique?

Or are they unique per release?

adhawkins: they are unique

(over all ids)

Ok.

Just thinking about how to handle multiple images in flactag.

Currently it only has one per FLAC file.

However if I'm going to put all of them in, I need to be able to tie up which one is which when checking for changes.

Is there any concept of 'editing' an image? Or do they just get added and deleted?

The metadata can be added, the artwork cannot

when we add support for replacing images, we will be using a new ID

Ok, so (for example) uploading a new front cover will cause a new ID to be generated?

yea

Ok, think I've got it straight in my head then. Need to come up with a way of mapping the CAA image ID to an image in a FLAC file then.

There is a description field for each image, I guess I could hijack that.

hi..after upgrade and startup of mb servers..getting this memcache error in logs though request is getting processed.. [error] Caught exception in engine "Can't locate object method "forget_dead_hosts" via package "Cache::Memcached::Fast" at local/lib/perl5/Cache/Memcached/Managed.pm line 765."

is it of any concern?

Swarup: yea

Change the reference of Cache::Memcached::Fast to Cache::Memcached in your DBDefs

we should get that out our default config, because it's broken

ha ok

4 unit tests to go and then I can actually start trying to fix this bug

what a productive day...

ruaok has changed the topic to: agenda: review, artist page fixes (ruaok)

Mineo: since rika uses data dumps, nobody's set to AE -- if you need that on the slave DB I can make it happen, otherwise set it yourself on the static one

Mineo: and until yesterday nginx was off on rika since the databases were updating; should be back now

ruaok has changed the topic to: agenda: review, artist page fixes (ruaok), forums (ruaok), bug triage (ruaok)

ruaok: scheduling.ocharles is offline atm

k

just to make sure you're aware of that

another good reason to address it in the meeting.

though I could bring it up on my other server actually

alrighty

have seen mr 2099 freaking out on our blog?

si

k

don't really see point in responding

yeah, its just froth.

but, a solution should be very high on your priorities list.

but I suspect you know that. >)

:)

ianmcorvidae: I'm kinda stuck with the forums box.

I ordered a raid card that should've been compatible with the box, but it didn't fit.

have you got any experience with raid cards?

not really, no

the one I've ever tried to use didn't actually hold configurations :P

ruaok: data loss is higher on my priority list

so i'm backlogged for now

more on that in the meeting :)