#musicbrainz-devel

Project full-import build #104: SUCCESS in 4 hr 59 min: http://ci.musicbrainz.org/job/full-import/104/

CatQuest: The "Import to MB" button is only sometimes displayed due to a bug somewhere. murdos, however, insists there is no issue: https://userscripts.org/topics/89514 (I also always have to do some workaround to get it to appear - but I also use Adblock)

does anyone know if there's already a ticket for http://musicbrainz.org/edit/23540403 (the non-thumbnail version of the picture is fine)? I could swear I've seen a ticket for that somewhere :/

Mineo: http://tickets.musicbrainz.org/browse/CAA-63?

that's probably it, thanks :)

LordSputnik: just a shame it's a poor choice, financially

at least for stuff running 24/7

(re: amazon cloud stuff)

ocharles: it's free for a year though :)

ocharles: what's a financially better cloud service?

LordSputnik: hetzner and linode are both good choices

not exactly "cloud"

but if you're just running one server 24/7, that isn't really how the cloud works

but if I need to I can scale

but I really have no clue about cloud hosting and just want a cheap replacement for my raspberry pi which is slightly faster, and amazon seemed easy

ruaok, ruaok_ - ping

ijabz - ping?

actually, nm

pingu is to blame

one perl-fcgi process , started 16:05 seems to be unresponsive

ok, i restarted pingu but it still seems to have the same behaviour

i've lowered the weight on that server so hopefully it won't impact things too badly

possibly it's a specific sort of request?

ianmcorvidae: could be that, not sure

i'd rather go and cook my dinner ;)

but it's weighted less than lolo now and stuff seems stable enough, so i'm gonna see what happens

fair enough :)

http://tintin.mb/render/?width=1220&height=... is what i'm using for monitoring this atm

so you might want to keep an eye on that too

ocharles: if you're around, astro's now 500ing a bunch

(well, 5xxing, but)

oh good lord.

looks like it's 502/504, as presumably expected

astro is out now

how do we debug this?

not sure, it feels like something else is broken :/

all back in, more weight on pingu

are we running with four frontends now? or is one still an upgrade-work-in-progress situation?

four at 12.04

i think this might be a bit better now

:)

it seems like there's some trigger, I guess

ocharles: haskell is hard!

I can't imagine it's load, given we were fine with three frontends before 12.04

warp: correct is hard!

ianmcorvidae: yea, i think what's happening is a lot of processes are taking a long time to deliver responses

which means one of our dbs is slow - search, wiki, totoro, memcached, or redis

or they're getting stuck somewhere else

but yes

wiki could be slow due to hoser being loaded in some capacity, how is that doing since we made pinoprime and more stuff is hitting tintin?

caa-indexer on pinoprime is pretty borked too :/

i restarted it and it seems to be actually processing events now

luna/forums have been periodically flickering, according to nagios

ok, gonna get back to non mb stuff. highlight me if things go down again

k

warp heh, I do wonder whether haskells primary aim was to be as a test to prove to people that you are clever

it started out as a language for doing programming language research

so that's why it's designed to be able to make so many high-level concepts fairly integrated

conveniently though, understanding those concepts tends to help you use those concepts to write better code, so :P

Decided I should read all the iain banks books I havent read, now that he has died

Just read Excession and all the minds taking to each other, reminds me a bit of the musicbrainz irc channels

hehe

ianmcorvidae the few times I look at it, I just think wtf, there is no easy way into it

hm, I suppose

I guess it's a language focused on making you think before you write code, which a lot of languages don't really enforce

and the same sort of thing tends to apply when learning it, which is a barrier

hopefully you've at least been linked to http://learnyouahaskell.com/ before though, that's a pretty good introduction to the language

there's always more things to learn with haskell, of course, but that gets you far enough you can typically figure out what things are doing and bootstrap your way further

yeah I did look at it, but got a but bored with it.

fair enough

heh, yes almost anything is more fun than extended periods with the inlaws

warp: watcha working on?

ocharles: I'm trying to port a python script which does some magic and then downloads a resource.

ocharles: https://gist.github.com/warpr/6406768 <-- I'm currently trying to get rid of the do-notation on resolvePrefix

(and failing)

Freso: patches are welcome for the Discogs userscript

warp: where are you stuck?

oh, getting rid of do notation

yes

if you write it as "a >>= b" and then move 'a' and 'b' to separate functions, you can ask ghci there types, and gradually fill them in

which is sort of "hole" driven development

hm

I get lost when it starts complaining about monads :)

ocharles: http://paste.ubuntu.com/6052616/

murdos: The workaround work fine for me. :p

*works

hmm

warp: write the types out for step1 and step2

you see to be mixing monads

or more, step2 is in a different monad than you expect

(you expect IO but it's in Maybe)

ocharles: I was kind of hoping the transformation from do-notation to >>= would be straightforward, and I was doing something silly wrong.

warp: well it is, because that transformation is completely mechanical

but I think you've tried to go too fast

:)

it may help to look up how do syntax works in the ghc manual

and then manually expand it

and then try and tidy that code up

because right now you've dropped a return out

lookup should be 'return . lookup'

ah

oh indeed.

so if you replaced the ';' (that do notation would add) with >>=, then you might find the refactoring easier

but this stuff you just develop an eye for eventually

presumably understanding monads would help.

hmm, the ability to remember exactly what the type of >>= helps

but that's mostly memorizing the "monad" type class

ocharles: unrelated, I don't understand what happens with the curly braces. is there a way to do that inline?

what do you mean by curly braces?

noRedirect request = request { redirectCount = 0 }

oh

not really, that's specific syntax

that says "replace the 'redirectCount' field with 0, in the 'request' record

the possible fields are specified by the type of request

though i don't think conduit exports them

though it *does* export the fields themselves, that's why you can update the record

at least iirc

yes, the documentation says to create the request with parseUrl, and then change fields afterward if neccesary.