#musicbrainz-devel

just ws/1 I think

ah. that should probably changed then

to cover all of /ws I think

please throw a ticket at me and I'll get that sorted

will do

interesting, components/relationships.tt was never tested by html_ok

ocharles: I get conflicts merging master into next, is that expected?

it's possible

depends what's conflicting

CONFLICT (content): Merge conflict in lib/MusicBrainz/Server/Data/CoreEntity.pm

CONFLICT (add/add): Merge conflict in lib/MusicBrainz/Server/Data/Role/GetByGID.pm

mm, yea, I know why that is

let me finish this work and I'll merge master into next and resolve that

awesome, thanks

warp: next updated

warp, djce: do you know if mail headers are case sensitive?

We add the 'BCC' header sometimes, but I'm wondering if it's not working because it should be 'Bcc'

http://stackoverflow.com/questions/6143549/are-... seems to say it shouldn't matter

I'd say it wouldn't matter

Hm

Well somewhere in our mail setup BCC isn't getting stripped out from the headers

nor is an additional mail being sent

send me your evidence and I'll take a look when I can

djce: https://gist.github.com/1014328 is the result of me dumping the email right before it gets to the sendmail function

ocharles: are you sure the sendmail function you're calling parses that from the headers?

well that's where To and From comes from so I can't see why not

ocharles: "If you don't specify recipients, it will use all the email addresses it can find in the To and Cc headers by default."

where are you reading this?

ocharles: http://search.cpan.org/~rjbs/Email-Sender-0.110...

via the documentation for Email::Sender::Simple

so Email::Sender::Simple parses the headers for you, but is limited in what it does for you.

hm

but looking at the source code, it should still be passing all headers through to sendmail

also, there's no way to do { bcc => 'blah' }, so I don't think that's it

/usr/sbin/sendmail doesn't do anything with headers

sure, but the whole message gets passed to the MTA I thought, which would look at headers?

ocharles: well, the /usr/sbin/postfix has a '-t' option to set recipients from headers, but in general /usr/sbin/sendmail did not generally do that for you.

hum, well I dunno what to do then :)

ocharles: normally you would have to set the recipients and the envelope sender on the command line.

i'll ask rjbs and see what he says

ocharles: I expect there's a CPAN module which does all that stuff for you, but Email::Sender::Simple is not it.

Email::Sender::Simple is part of Email::Sender, but that doesn't look like it has a method either

ocharles: Mail::Sender seems to do it, but that's probably using SMTP as the transport.

mm

i'll leave it til rjbs gets back to me, probably have better things to do than chase this down

ocharles: I'm looking into some performance issues. what would be the best way to profile mbserver?

oh, a new index helps quite a lot.

ocharles: I'd like to discuss MBS-2347 with you at some point, I think it needs a new indexes on artist_name and label_name.

I like more indexes. :)

I wonder how many other 502s can be taken care of with more indexes. :)

hah

haha

Quite a few I hope (but maybe we could cut down no of queries, clever queries as well)

ruaok: I guess our benefit is that many writes which deal with those tables are done through modbot, and thus don't need to be super fast.

ruaok: Did you do a new search deploy for me ?

but, modbot is grouchy enough. maybe some indexes will help his (her?) disposition. :)

ijabz: its 50% done. :)

lol

new indexes should be out, I just need to update the search servers.

So the new index builder code is already done, and indexes have been built with that code ?

Coz if that is the case my fix hasnt worked :(

new indexes have been built, but no new code pushed out.

I'll do that in a minute.

I'm heading up to a friends now for an hour or so

warp: we can discuss it when I get back, if that's ok with you?

(then we can do the release, etc)

will try and be back for 7:30 bst, toodles!

ocharles: sure

oh right, (there will be no fix until the indexes have been built with the new code)

the indexes have already been built with new code.

ruaok: emailed design doc last night, let me know when you've had a chance to look over it

Batsy: will do. you're next on the list after updating the search server.

and thx! :)

okay :)

eww. stinky girl. :)

ijabz: one server update and live now.

sanity check please?

both servers updated now.

Yeah, should have fixed http://tickets.musicbrainz.org/browse/SEARCH-70, but hasn't

ruaok: why this agenda? Is there a meeting today?

My fix must be wrong then , if te ewn code was used to build indexes, cant see it though

Oh well, Im off on a bike ride now to lose a few pounds

I see "agenda: SEARCH-65, MBS-390, MBS-2294" on the room description…

feck. indexes have not been updating at all. :(

(an artist I added 10 hours ago is still not indexed)

oh, ok i'll catchup with u later

yay. travel visa. sweet!

23 */3 * * * root ulimit -n 10240 ; exec su -l -c 'build.cron' search

*/3

wow.

yup

any number divisible by three

I guess now we need to wait 63 minutes to find out if the stuck cron process was the culprit

well, the build process works. not sure why cron got stuck.

is logging

Users on #musicbrainz-devel: MBChatLogger zazi aeontech ppawel ruaok voiceinsideyou reosarevok bignoise dinog shultz luks Mineo warp Batsy murdos Leftmost kurtjx nikki bitmap davitf flamingspinach ianmcorvidae Muz enjayhch sampsyo ocharles lfranchi

Batsy: I'm reading the design doc now

okay.

I like what I see, good work.

but there are a few things we ought to address.

what is the &callback=<identifier> argument on the URL?

jsonp callback

presumably

yes, that

so thats the name of the function that will wrap the json data?

yup.

does it need to by dynamic?

can't we just pick one and use it consistently everywhere?

probably

it needs to be if it's jquery calling the jsonp

needs to be what?

dynamic

why?

jquery autogenerates the name

you don't provide a name to it

huh. so jquery tells you what the callback name needs to be on a per call basis?

it generates a name to use, so

interesting. I'm sure it has a good reason to do that.

or at least I hope so.

looks like you can pass it a callback name, looking at it, but it's not recommended

"It is preferable to let jQuery generate a unique name as it'll make it easier to manage the requests and provide callbacks and error handling."

(from http://api.jquery.com/jQuery.ajax/)

ok

lol