#musicbrainz-devel

yay, hostel booked

bring on Icicle, Teebee & Calyx and Rockwell!

ocharles: I gave up on the appears_on query

want me to take it over?

yes, but not yet.

ok

I did try to incorporate the other bits of your feedback, sI'm almost done with that

I'll poke you again when I've pushed the branch,.. then you can have a look.

you broke the build!

http://ci.musicbrainz.org/app/projects/report_d... :)

the t/tests.t one is failing at search, I think

ah, Country search results. that indeed was me.

+in

ocharles: I've pushed mbs-1517, so if you want to take a look, you can do that now :) (there's no hurry though, it's also on code review)

warp: cool :) I'm just doing some SoC reading/replying at the moment

ocharles: I did change it to get all release groups in one go for all the recordings, but not any truncating / sorting of them in the query itself.

[error] Couldn't render template "undef error - The 'epoch' parameter (undef) to DateTime::from_epoch did not pass regex check at /usr/local/lib/perl5/x86_64-linux-gnu-thread-multi/DateTime.pm line 486

odd

which server?

that means that the registration date didn't get serialized to your session when you logged in

maybe you logged in from a branch that didn't have that code added in yet

yes, session b0rk, visiting /logout fixed it

I've seen it before, but it still takes me a few seconds to realize what's going on :)

LOL at the "1911" mode in youtube today

warp: where is your use tracklist work?

I think http://tickets.musicbrainz.org/browse/MBS-1637 is related to it, and as I can't see it in master, maybe you want to take that issue?

ocharles: oh, yes, that sounds like something for me.

ocharles: you still have MBS-78 on code review btw, my branch is based on that..

ah, right

i'll try and get that problem fixed today

great :)

ok, pushed warp

ooh, that was quick

ocharles: oh, now that I'm reading the changes on code review, I have a related question.

ocharles: I've skipped adding discids in the new 'Add Disc' dialog for now, because I didn't understand how to do that. can we create discids from freedb imports?

or should that only work for cdstubs?

heh, "one" through "four" hardcoded (disc one)

ocharles: so... are the cron job thingies running on test yet?

ocharles: ping

ijabz: I'm going to run the indexes on the new hardware soon.

before I do, I am tempted to up the number of filehandles a machine can have...

ooh, be interesting to see how long it takes and if it fails on filehandles

yes please do

Also, funny issue with works, dont now if there was a problem building this index before http://tickets.musicbrainz.org/browse/SEARCH-62

yeah, I saw that bug.

I'll keep my eyes open for that issue as I build

Cant reproduce it locally

hmm

797953

323433

hmmm.

pingu (new index machine) has way more available file handles.

but I cant think that we're using 323k open files.

I wonder what the per process limit it.

either way I suspect that this will not be a problem

Perhaps just try with default

yep, I agree.

Also, I should have beta of my ngs tagger available next week to test against test.muiscbrainz,org , if thats okay

then then I hope to have ngs.musicbrainz.org available which should have more throughput that musicbrainz.org :)

lets coordinate that so that your users can use the new servers

oh cool, I was wondering about them

I've got 3 machines nearly configured.

the database is converted and ready to go.

Yes will do, also ther eis something I would like to test on the search servers

I need to configure another 2 machines, but that isn't as urgent.

k

tis the time to test things. :)

They run using file i/o or memory mapped, by default they use file i/o but memory mapped should be quicker ..

I would like to enable one to use memory mapped and compare performance and stability with the others

its just a simple http request to the server switch between the two

interesting.

we may want to wait on that until we do some real load testing.

I'm going to tee the web traffic we get on the live site and play it back against the new site.

the output will just get tossed, but it should simluate real load on the servers.

making the switch then should give us a better clue.

agreed

does anyone know why hobbes only knows about 2 ASINs?

ijabz: ping

only knows about?

seems like the universalchardet stuff has moved.

nikki: there are two rows in the release_meta table that contains asins.

though the converted data on totoro has a lot more.

not sure what happened there.

hmm... I think I added two asins

I know cover art on test isn't working though

yeah, that would make sense.

know nothing about this package, when did you get this error

just now.

yeah, but doing what

on a completely clean build on a virgin machine.

mvn package

ijabz: complete error: http://paste.pocoo.org/show/364203/

so you know 1 is because you have to build the mmd stuff first

Hmm I get [WARNING] Missing POM for org.mozilla:universalchardet:jar:1.0.3

but it still builds okay

building mmd now.

I'll let you know if it passes.

it didn't last time.

I had to follow the manual install instructions it gave

500 for me nikki

500? o_O

nikki: "Caught exception in MusicBrainz::Server::Controller::User->do_login "Can't call method "epoch" on an undefined value at /home/musicbrainz/musicbrainz-server/script/../lib/MusicBrainz/Server/Authentication/Store.pm line 32.""

huh. haven't seen that

ijabz: same error, sans the mmd one. build failed.

Looking in maven repository now, remember what it was for now (freedb charset matching)

thx

ijabz: should I install by hand or should I wait for you to try and fix it?

I have another clean machine to test with later...

I would try and do my hand now

k

Im trying to fix it, but its a bit late and Im confused

no worries.

I can proceed.

You can download from http://code.google.com/p/juniversalchardet/

i did. thx

building now.

Ah, I dont think it ever was in maven , http://code.google.com/p/juniversalchardet/issu... !

dont remember writing this though !

too many pints while coding?

I know I've done that. :)

Sure thing

build successful \ø/

Ah, I created a makeinstall.sh script for this, and updated the EADME

so, that needs to get into the maven repo before it works smoothly?

yeah, there is a local copy of the jar that the script bungs into the repo