#musicbrainz-devel

And database people about ?

ok here goes

CREATE INDEX artist_name_idx_page ON artist_name (page_index(name));

why is the second one needed ?

ijabz: I haven't looked at the db schema for a few years,

but if it hasn't changed since then,

then the answer is that the "page" column is used for browsing by name.

but I have to say, that SQL looks wrong. Or, it has changed in the last few years.

heh, you haven't seen the ngs schema then? :P

correct.

ok, so ist just for browsing artist by names, it wont speed up any other sql queries

I would assume so, if "page" has the same meaning it used to pre-NGS

it looks like it, looking at http://bugs.musicbrainz.org/browser/mb_server/t...

ah, so the page logic is moved into the db now. IIRC it used to be in *.pm somewhere.

it's probably just another of the things which has changed for ngs

thx, I didnt undertsand the syntax, thx djce and nikki looks like you are a database person ;0

ruaok: Quick 1-minute update on web stats?

shoot

hullo!

I'm half way through writing an email now. I'll send it in a mo.

But in short:

ok.

see http://stats.musicbrainz.org/webstats/wiley/

I should really get to the office anyway.

and http://stats.musicbrainz.org/webstats/zaphod/

woah. 7.3M daily avg.

That'll be /ws/ then :-)

yeah. :-)

mornin folks

or new zealand or wherever it's morning

i wish :P

his connection seems to be in france.

wow. the web service accounts for 89.3% of our overall traffic.

and the old web service *still* gets 3.3M hits per month.

<BANG>

MBChatLogger has changed the topic to: http://musicbrainz.org/#devel

hello!

welcome everyone!

ruaok has changed the topic to: agenda: review, b1 review, warp planning

moin warp.

warp planning \o/

aCiD2: you here?

oui

:-)

kick off the review form last week, aCiD2 !

greetings everyone

ok :) As somewhat anticipated work slowed down this week

But I still managed to get a good 10 hours fixing some bugs that are hitting test.mb

Sadly, the most noticable problems (/edit/open) is still yet to be fixed

:-(

The biggest work I did was refactoring our database handling, -- mostly a development issue, but this runs our test suite about 300% faster (and I think it will actually make the web site faster too)

yeah, next time it works, ping me. I've never seen it work. :)

essentially, we now connect to both databases once, per process, and share that handle over the life of the process

(before we were reconnecting and disconnecting per request)

good. that wasn't the case before?

Newp

what was it before?

Connect on a request (lazily), and then disconnect on completion

I think under apache that might have used a pool, but it was still a needlessly complicated way of doing things

ah, on a per request basis.

yes, apache::DBI provided that.

ok. how is your week looking? when can we expect to have /edit/open working?

This week it will be working, that has to be fixed now

this week is so-so, I should be able to hammer out another 15 hours work I think at least

good.

did you look at my discid dilemma at all?

I'm stuck on that for finishing the WS.

Yea, sorry I haven't got round to replying yet, I was hoping to talk to you about that

lets do that after the meeting then.

warp: when querying for the data, too much data is returned and its not easy to pick out which ids belong to which release.

join us in the post meeting discussion. :)

ah

warp: how was your week? busy getting out of your old job?

yes, didn't do anything mb related really.

ok. I noticed this weekend was very quiet. I needed a break too, so it was hard to stay focused.

I kept working on the web service. today I won't be able to work on it -- today is set aside to close out 2009 for the foundation. loads of BS paperwork and crap like that.

:)

fun

maybe i can get to the annual report.

ah

dave just finished his work on the web stats, so I can write the report.

:)

one of my main tasks for this week is to get the WS working (finish all core elements)

and to think about how to integrate warp into the team. but more on that on our next topic.

djce: please give us an update on the web stats.

erp. looks like he's run off. :(

stats.musicbrainz.org has been updated....

this is our front end that collects stats about all of our services.

ruaok probably meant ' musicbrainz-full takes into account all of the traffic to musicbrainz.org '

musicbrainz-full takes into account all of the traffic to mb.org

somewhat interesting. 90% of our traffic is web service traffic, which isnt very surprising.

but its a clear case for making sure the next WS is *tight*.

and on a business note...

SO which value do we look at to work out the 90%

there are more people wishing to have unfettered access to the WS. (read: no rate limiting)

I think making a amazon EC2 hosted WS that has not rate limiting, but is $$$ for access may bring in more money. certainly to pay for warp. :)

but that is a lot of work.

oo, interesting idea

ijabz: http://stats.musicbrainz.org/webstats/zaphod/mu...

look for : Top 30 of 2511791 Total URLs

evening murdos

hi

aCiD2: yes, but there are plenty of engineering issues to make that work.

77% web service in that list.

ijabz: yes, this would be part of that.

oh yea, sure