#musicbrainz-devel

but yeah, 100 requests per second coming from headphones and beets-via-headphones

since the latter wasn't being throttled, it wasn't getting ratelimited *at all* until fifteen minutes ago or whatever it was, because it was under the 50/s limit for python-musicbrainz/0.7.3

heh

which is where it was getting binned, though apparently clients *actually* using that UA are now only about 3 requests per second :P

So tomorrow we'll have some people dropping by like "you broke our stuff again!"

to which I think we respond "no, that was you, with ours" :P

We respond "implement a latest-changed in MB and then we'll see"

(if nobody wants to, bad luck for them :p)

heh

well, we have that partly done, I think rob's working further on it

Yeah, I guess it is useful for others too, not just headphones

this doesn't solve the problem completely, of course -- ideally we'd be able to support our capacity

which means? :)

Not break up as we approach 10?

down rather than up :p

ideally we wouldn't need to ratelimit people because we'd have enough money to pay for enough bandwidth :P

yes

Ideally people wouldn't make absolutely ridiculous requests and that wouldn't ever be a problem :p

etc.

(and our replies wouldn't be ridiculously detailed without need, and the problem would be smaller)

yeah

there's many ways for us to fit better into our capacity, more money or better use of resources :P

But from what we have, I'm happy with the "throttle and let them pay for free WS" :p

I mean, since they're not paying any artist, they might as well pay someone :p

:P

(or find stuff the slow way like the rest of us)

also: anyone else who uses python-musicbrainz/0.7.3: a.) god damn upgrade your thing, and b.) you're welcome :P

heh, now it is actually blocking more than it's letting through

they're not going to be pleased :P

How many of the requests come from unique sources?

I'd need to do some log analysis to tell you that, heh

Is it a relatively small amount of people going nuts, or a huge amount of people going slightly less nuts?

luckily my copy of the log is going a lot faster now at least

well

any individual IP can still only make one request per second

we are getting 100 requests per second

:P

the per-IP ratelimit comes into effect *before* this one

but they're using python-musicbrainz-ngs which deals with that ratelimit anyway, so

Well, 100 people aren't that many :p

So it can be 100, or it can be 10000 :p

We just know it's not less than 100

yeah, fair enough

Which would be absolutely nuts

So it's good it isn't :p

heh

reosarevok: RE: "latest-changed", if the Bottle ends up caching WS data anyway, the "latest-changed" could be utilised to selectively update the cache as-needed.

Certainly, that's what I mean with it having other uses

(Of course, being able to just cache properly in the first place would also be good...)

Speaking of proper cache

Any ideas on how to do the wikipedia thing?

the wikipedia on entity pages on MB thing?

there's a patch up for that already :P

ocharles: speaking of which, please respond to my comment on http://codereview.musicbrainz.org/r/2393/

Oh, cool

ianmcorvidae: Are we out of polar season btw? Or won't it get tropical until the release is out?

yeah, the release needs to happen

hopefully it'll be before thursday or whatever insane day it was last time :/

Also, I'd really like to get closure about raptekster.dk - but I keep missing Rob on IRC. Just a quick yay/nay would be good. :/

what's the situation there?

ianmcorvidae: The RFV has passed, but I need a legal yes/no from the style grand chief, aka Rob, before putting it in code.

ianmcorvidae: As it's a lyrics site.

Mail him then

Hopefully he'll see it

(and if he does not you didn't lose anything since you can tell him when he's around)

Night!

I would mail him if my mail was behaving. :p

I guess I'll try again in the morrow.

I wonder how exactly does headphones use mb

Badly

Well, fine, that's not "exactly" :)

Code is open though, you can check it

so I take the answer is "we don't know" :)

my understanding from the conversations I've seen here was that it rechecks stuff every couple of days looking for changes

morning

where 'stuff' is your entire music library

which is why I keep saying we should finish this recently updated entities thing :P

ahh good, the track table is missing the 'delete_orphaned_recordings();' trigger

ocharles: I believe I have found the problem with the failing test. master is building now on jenkins.

yay!

fingers crossed

+@edits = nsort_by { $_->id } @edits;

warp: Sort by edit ids correctly in Controller::Work::Create test.

yea, i saw it on github

ocharles: btw, a full test (with HTML5 validation) takes 16 minutes here.

right, and hobbes is a bit slower than 'here', so call it 20 minutes. and a full test takes 30 minutes, so we're at almost an hour still :)

so a faster and less memory constrained build server would help a lot.

oh wait

you meant 16 minutes in total

indeed

time prove t/tests.t > all.log 2>&1

real16m6.793s

ocharles: anyway, build FIXED, when do you want to release? :)

oh good, tis release time

i can release when ever

alright, let's do it now then.

I'm logging in on lenny

ok

hm, 19380 orphaned recordings

I'm ready!

ok, go ahead

astro is out.

hmm, what a mess

what happen?

that one again

so this DBDefs stuff is not working.

or configured wrong

well, DBDefs/Default attempts to set this automatically. that not working is what I'm saying.

setting it to a simple value in DBDefs.pm should work.

yea, but you can set it manually

that looks better

warp: astro done!

yay!

astro back, pingu out.

ah

I think astro might have broke because I did a 'git stash'

but there are, as always, unstaged changes :)

(this time to DBDefs/Default)

uh oh

so should I take astro out again?

nah

i did remember doing work on that, and when I opened it the file i was like 'huh, I thought this was changed?'

so i'll get that commited to beta soon

ok

bah, pingu did the same :)

[pingu] out: make: Entering directory `/home/musicbrainz/musicbrainz-server/po'

[pingu] out: make: *** [attributes.de.mo] Error 127

matters

oh

pingu done

no idea

pingu back, asterix out.

we don't have translations enabled on the main servers, so it shouldn't