#musicbrainz-devel

there is simply not enough information to present a useful interface.

"This edit was recently cancelled." <-- how do I get that processed on a development checkout?

warp: run admin/CheckVotes.pl

ocharles: great, thanks!

ruaok: so what I think we may have to do is: I write a new data section and make sure it works

next, dump all old merge release edits from totoro, and get them to me

I load them into my database, and play around with a function that tries it's best to turn them into the new data schema

and then reload?

use this data to generate a bunch of UPDATE's and load it all onto ngs

for NGS + 1?

ruaok: needs to be sooner rather than later, probably

it's stopping people voting on merge releases correctly

and I should mention that the data at themoment also can't be migrated into anything useful

how long will this take?

If I put my running shoes on I might be able to make it in the hour

go!

but I need to hand the baton to someone else to dump the data I need

tell me what I need to do.

I need all edits of type 311

in CSV?

in anything I can load

(into postgres)

astro.mb:/tmp/edit_311.csv.gz

thanks!

use \copy edit_311 from '/tmp/edit_311.csv' with csv

great

after unzipping

ruaok: I also have an issue I'd like to make you aware of.

that my pre-coffee time logic sucks?

no :)

shoot

ruaok: when adding a release, seperate "Add Release" and "Add Medium" edits are created (and perhaps some more edits, depending on what you did in the release editor)

ruaok: if the release edit is accepted and the medium edit rejected (in either order), we end up with a release with no medium.

thats to be fixed for ngs + 1?

ruaok: and that will confuse the release editor, it will not be able to add mediums to it.

ruaok: it's not an issue which will happen that often, but it should still be fixed soonish and we need to think about how we fix it.

so, lets note it in the release notes for now

after brekky I'll start that pager where we can collect little bits of info like that

yes, that is fine. the ticket for it is MBS-2035

WIZARD_MEMCACHED

does that need to be a separate instance from the SESSION store?

no, it prefixes things with wizard_session:<SESSIONID>, so it can coexist with our other stuff.

nice, we have even a new NGS data dump out. fresh!!

in that case, lets just use the SESSION store over.

can you fixe that up and not require a new DBDefs setting?

I'll finish reading the review

not right now. SESSION_STORE_ARGS uses a different format and is empty on a typical development checkout.

feh. ok. fine.

phew. update: the new edit type works, working on the display now, then porting over old edits

we just need to add a single MEMCACHED_SERVER or something at the top of DBDefs, and have all the other options use that.

ruaok: I'll create a ticket for it :)

I've got it

I'm adding to MigrationPlan

oh. nm

ok

yes, please do that.

ok

but I have a migration issue to change DBDefs.pm noted

ruaok: Post NGS?

yeah

I see most of the bugs from 8 hours ago still open. :-(

did some get fixed and not closed or are we far behind?

ruaok: a bunch of got fixed, the voting problem took some time but was really important

I've only worked on this session stuff and MBS-1929 (which is the failing release editor related to releases without a medium).

I fixed ModBot too, but now it complains about something different

hence not closing that ticket

lol

modbot can be cleaned up after the release. :)

Music: check. Caffeinated, carbonated drink: check.

I think I'm ready.

djce: we still have a cachet of minor things to tweak.

are we ready to expose the ngs search to the outside world?

Who are you asking? I see no problem doing so.

(It's not being updated yet though, of course)

ruaok: well, there's still bugs. but there will always be bugs.

so here's where i got to last night: http://pastie.org/1921079 when running make installdeps

somethings need to be manually installed using cpan

force install <module>

that usually gets past this

djce: that was directed to you.

and yes, updating is another thing to work on.

My idea was to use cartman for now

tsk. sony. (just up and hacked again).

lol

If the ngs lucene servers are allegedly ready, we could direct some live traffic to them like we did for /ws some 9 hours ago

got for it!

ruaok: looks like it fails in a test at "Can't use an undefined value as a HASH reference at /usr/local/share/perl/5.10.1/Catalyst/Plugin/Session/Test/Store.pm line 93." - any ideas?

I'm not watching logs though. I'm trying to get to the office in the next minute or two.

ruaok: or are you busy right now? :)

force install Catalyst::Plugin::Session

aeontech: rather busy, yea. :)

hehe ko

ok

New edit done, displays correctly. onto the data upgrade now

ruaok: I can't upgrade this data to give a useful format (though the edit can still be applied)

I'm thinking about leaving edit notes on all the edits saying "You are recommend to cancel and re-enter this edit so it can be display better"

there are only 20 open edits. :-(

ah, might not be worth it then

ocharles: but better than nothing.

but going forward the edits are correct?

dora/roobarb now handling 50% of live

ruaok: yea

ok, we'll live with it.

ruaok: Is anything planned to change on dora/roobarb between now and live?

nope.

roobarb is bored. <3

And is dora/roobarb's data set sufficiently close to asterix's so as not to notice the difference?

in theory. :)

but for a load test, that's fine.

ok, heading to the office. back in 15

bwahaha

nikki: you called?

I did?

yes. with your phone. or someone with your phone number.

o_O

19:03 from a +49 number I have in my cellphone as Nikki.

sorry

seems my phone has a mind of its own

indeed!

luks: http://i.imgur.com/cdaHI.png is the best I can do for old edits :(

http://i.imgur.com/MtgHK.png is the new edit

if it's just 20 edits, I think the old display is fine.

perhaps s/old/original/

'medium 1 is now medium 5' also works. so leave off old entirely on the new version.

navap: well, we may change data schema again in the future

this the first edit to have _edit_version in :)

though hopefully future changes will actuall produce an edit_merge_releases table or something so I can actually use SQL to do this update

nice, rob has given me an empty test data set.

ruaok: edit_311.csv.gz is blank :)

huh?

it's has no contents

argh, crap, nvm, I forgot how gunzip works

lol

I'm so used to stuff working in pipes and streams, did gunzip edit_311.csv > 311.csv

ocharles: that's zcat :)