#musicbrainz-devel

ocharles: ah, I managed to get youtube to show up by adding it to Data/URL.pm, but the super() thing is still giving me empty strings

goodmorning!

moin moin

tsk tsk

ruaok: u there ?

ok sent email

ijabz: pong

I should be in bed, but I'm not a wise man

re-fetched head, built and its running again.

indeed

MAX_BUFFERED_DOCS = 5000

instead of 10k

%T warp

time-of-day ruaok :)

how are things in .nl land today?

sunny

very nice

hi djce

ok roobarb is busy, hopefully it will build the full index tomorrow.

bed time now.

nn

oh hi ruaok, maybe your like me, if Im on my own I cant get to bed until Im just about falling aslep

ah, missed him

doh, missed ruaok too

warp: ping

pong

!

warp: s'ok now, I forgot how to validate the xml but remembered we use jing

ocharles: ok :)

ijabz: ping

well ijabz, warp: ws/1/release?inc=counts was failing the test as currently it adds a <track-list count="12" /> type element to the <release> element. /ws/1 does not currently do that, but it seems harmless to add it in (passes schema etc)

any objections?

hmm. surely the schema validation should fail if there's an element that isn't expected?

alastairp: <track-list> is a valid element for the <release> element

so schema validation passes. the xml tree comparison fails though, because the test doesn't expect to see it

right

something else needed it, and I guess it's too liberal adding it in, but I personally don't mind it being there

if it's backwards compatible (I don't see why it shouldn't be) then it can stay

I think it's because inc=count for artist gives you releases with a tracklist count, but inc=count for releases currently doesn't (which is a bit odd)

alastairp: right, all other tests pass, and it passes schema validation

I could maybe see a library not expecing it and exploding

ocharles, yeah leave it in

great

but if it's in an element where it could appear anyway, then that should be fine

actually gotta go, speak to you later if need be

hopefully we'll be at 100% pass rate for master today

first time we'll see the green light on ci!

ncie

ocharles: I agree with adjusting the test, not the /ws/1 output in this particular instance.

ocharles: also, looking forward to 100% pass on master :)

cool

ocharles: poke me if there's anything I broke and need to fix.

warp: well your qunit and selenium tests still fail on next

but nothing on master is your fault, that's all mine :)

I merged them into master today, they passed here (on master).

excellent

running CI on master now, lets see what we get!

how long does it take these days?

a good 10 minutes I think

boo, warps selenium test fails and gives 8% pass rate

and controllers.t still fails more than it does on my machine

ocharles: I'll have a look after dinner

ocharles: hm, the test is correct, something is broken.

ocharles: hm, a Schrödinbug. the offending line is from october 2010, from the reading code I am convinced it was broken since that time.

yet it passes on your machine?

no, it doesn't in master.

(it did this morning after I merged those branches though).

oh... odd

yes, very.

I could look into it further, I didn't delete those branches yet. but that doesn't seem like a productive way to spend my time, so I wasn't planning to.

still two out of twelve selenium tests breaking on master.

yay the recording index built!

I can work on getting ngs.mb. org setup

ijabz: just one second before you arrived I said the recording index bullt!

ruaok: awesome!

ruaok: will me and kuno have access to these machines?

maxdocs at 5000 made it work and it didn't seem much slower.

at least during testing period

maybe if you make a compelling argument. :)

i can do it through you, but it seems more efficient if I can have access to the postgres user myself

that one we should do together.

right, but I mean more than just general settings, I might need to alter statistics on certain tables (rawdata.edit comes to mind)

I can send you the config and you can send a tweaked one back.

ruaok: Good Morning

but I need to be able to do analyzed explains too. I don't really need root access to the machine, but being able to run those queries would be nice

I'll give you a shell account on astro, that lets you connect to the DB.

ok, that works for me

but postgresql.conf we work on collaboratively.

sure

k.

gonna be around for a bit?

if so, nudge me when I get to to the office.

yep, i'm back for the day now

meeting in an hour too

ok

2 even

yeah, I gotta get a move on.

2?

i thought it was 21cet?

oh wait, that's an hour away

timezone fail

MBS-1599 is the only one I know of.