#musicbrainz-devel

those are not the droids you're looking for.

:-)

who messed it up, you or the bank?

me, technically.

technically?

but they accepted it. then realized they accepted something that was incomplete.

then they GUESSED and sent it out.

they should've rejected it.

tsk

both, then :P

and really, if you guess are you going to send GBP to UK or the EU?

what's the difference? it's all weird foreign shit :P

haha

everyone knows the UK loves the EU

probably canadian! :P

seems freaking obvious to me, but then again I can tell the difference between two countries.

so obviously you'd send the UK euros

LOL. yes, canadian!

ianmcorvidae: wrt to the replication roll up stuff.

canadian euros!

and, yes? XD

ruaok: haha, brilliant.

it should not just be a tarball of the other replication packets.

okay

that's what I figured, but the concern is then we're generating really huge replication packets, and hit our existing performance problems with replication

every 24 hours we should take all the replication packets from the last 24 hours, concatenate the files in them and make a new -daily packet.

so I guess I should also be looking into being smarter about that?

and once again on every 24 * 7 to make a -weekly packet.

actually, the really huge replication packets for the time delayed feeds are fine.

its a sliding scale of timeliness and convenience.

the cheaper options are harder on the consumer.

the more expensive ones are easier.

fair enough

so on the LoadReplicationPackets scheme we should have a -hourly -daily and -weekly flags.

-a

so I guess the code I'd need to write is the code to roll together packets, to calculate which ones to roll together, and updates to the replication scripts to deal with one packet jumping several numbers in the sequence

sounds about right.

possibly also migration measures like "give me hourly packets until I'm at a day boundary so I can switch to daily packets"?

yes, good thought!

LoadReplicationChanges already has a --limit option, but making people figure out how many they need seems like unnecessary work to impose :)

presumably it should default to hourly to not break existing stuff :P

nikki: yes.

yeah, for sure

but we need the ability for users to specify which feed to use.

somehow.

yeah, I was just saying/confirming 'cause you listed all three as flags

warp: is the json ws live yet?

ruaok: no, only for labels I think.

ruaok:reminds me we need to do a simple release for search server, to provide the search server json part

warp: ok.

however there is one piece of work on relations that I need to do first

ijabz: ok.

early next week acceptable for that ijabz ?

ruaok: I have shippits from ian, but I'd like to get approval from oliver as well.

ok.

ocharles: poke. can you look at the json webservice reviews? :)

once that is done I can finish the "rss" stuff.

I'll try and get it done for then yes

dont rush on my account.

warp as I understand it , this json release is going to be 'this is the json, but it may change a bit if users find problems with it, right ?'

everything in my life is a mess right now and will take me some time to straighten out.

right :(

ijabz: yes, I think we should allow some time for people using it in production to give feedback. I expect we'll find a few bugs and things which can be improved.

yeah, agreed

ijabz: and then in a few weeks or months, depending on feedback, we'll mark and announce it as stable.

warp: not until monday, but I can then

warp: might be worth renaming the ticket you're fixing to 'preliminary JSON support' or something, to clarify the release notes

ruaok: btw, two other things we should do at some point: a.) schedule the wiki switchover, probably a blog post telling people it'll be read-only for a bit would be good

ruaok: and b.) I'd like to catch up with you re: the sysadmin stuff for the paid WS, but that's on no real schedule since I still need to come more fully to grips with the varnish setup/the VCL/etc. before it's actionable

does anyone here have a valid iswc so i can test my lookup

thanks

iswc not isrc

or they're shown in recording searches, so just search for anything and it'll list some

oh

:P

http://beta.musicbrainz.org/iswc/T-003.047.088-2 then

yeah its a pain to find 1

and s/recording/work/

and thanks

http://beta.musicbrainz.org/artist/9bffb20c-dd1... take your pick :P

T-000000001-0 is easy to remember.

ianmcorvidae: re: wiki switchover. yep, a blog post would be good.

when do you want to do that?

I'm up for whenever, really, as early as tomorrow :)

I don't know how much advance warning we want to give

tomorrow is a bit scattered for me.

how much time do you think this project will take?

most of the time is running the database upgrade script, which takes a couple hours

how much of my attention do you need?

probably almost none -- I'll need you at the end for doing the nginx switchover and some sanity-checking

mostly just want you around in case something goes sour, but the past two imports on perry have gone without a hitch

doesn't *need* to be tomorrow, either, that's just the earliest it can happen :)

CallerNo6: http://wikiport.musicbrainz.org <-- that's what the new wiki will look like, after we do the switchover; will take a couple hours of read-only wiki but that's no huge deal :)

I'm free from 5pm to 6:30pm pacific tomorrow.

is AZ the same as PDT right now?

yeah, it is

k

so maybe I'll start the process tomorrow around 3:30 or 4 so the end is within your availability?

so, what if we start the data conversion at 3:30pm tomorrow.

hah

ha. :)

same wavelength

sounds good.

k, want me to post a blog entry?

I can do that, should be quick

ok. sounds good.

I will appear starting at 5pm. I'll be around in the morning too, but lunch -> 5pm will be dicey.

okay

ok, in the calendar.

worst that happens is the wiki's read-only a bit longer or we say fuck it and use wiley for a bit longer

k

ok, on to point b.

catching up on the paid WS.

what loose ends do you have for that?

mainly what I want to know is what sysadmin-y things need doing other than setting up the VCL and varnish itself

I know there was some stuff in the ratelimiter, obviously

I don't know everything we were hoping to launch off with -- I guess just the no-waiting one to start, though

yes, the ratelimiter is the tricky portion.

basically I just want a picture of what needs doing between where we are and when we can start getting people paying us :)

:-)

you've gotten all the info from ocharles on this, right?

I think the next thing to do is to have a discussion with djce.

he's had some thoughts and was going to run off an do some ratelimiter things.

yeah; I have the VCL and such which I need to play with (and my own 3scale account to do that with)

we need to see where he is at and if he has anything to pass off to you.

that sounds like it could be right

ok, so the next time the three of us are here, lets chat.

sounds good

I'm really glad you're going to own this an drag it across the finish line.

so. long. overdue.

heh

the next project I need to kick off is the MB showcase.

helps that more money means more chance I get to keep my job ;)

we need to find everyone who is using MB and make a list of how and what they do.

ianmcorvidae: :)

and then start shaking them down for money.

afterall we seem to have 2M inbound links. that is "monetizable", right?

you'd think :)

haha

;)