#musicbrainz-devel

hey ijabz

I was wondering what needs to happen now you've added those punctuation characters to the search server

hi

Well the search server code has to be rebuilt and released on test

ah, who usually does that?

but I dont do that, I dont think this is enough of a change to bother doing that

ocharles I think

and does that mean the live server won't get updated?

the problem is mostly that one of the guidelines got changed to allow punctuation like that, so quite a few people are being quite... vocal about it :/

I think it's probably more that they just don't like the characters in general, but having the search not treat them the same doesn't help matters

so I'm just trying to keep track of when things are likely to get updated. I can't make anyone do anything, after all :)

Nikki, no Live will not get updated, I mean the whole codebase has changed and live still has the bigger problem of not decoding the freeedb index data correctly

ok

I've got a couple of hours to kill. anyone need anything?

But apart from that, not really

hmm...

yeeees?

in http://bugs.musicbrainz.org/ticket/5491 navap suggested using [[Page_Name]] instead of [Page_Name] in edit notes to link to the documentation pages... but is there a way we could do that without breaking all the old notes?

garrr

test is giving internal server error

hi warp!

nikki: not sure. we could convert the old notes in the NGS conversion.

'cause if we allow spaces like the ticket suggests, I think that would end up turning things like "I think you should use [Some Unofficial Name]" into a link to the docs, which it shouldn't, since we use square brackets for special purpose stuff

(if that bit made sense...)

it does.

ffs.

I passed through TWO security checkpoints already.

and now a team of people rolled in for a THIRD one.

fucking security theater!

in NGS is a recording always associated with at least one release or not ?

no

non-album tracks are converted into standalone recordings, which are recordings not linked to a release

Do you have an example

hehe

That reminds me. Is it technically feasible for NGS to somehow make Picard believe that a standalone recording is a 1-track release?

ijabz: well, we lose the fact that a track has been a NAT in Mason

When it is released

So no change there

(well, yeah, the change that we don't lose all the ARs and have to enter them again)

and we don't lose the mbids either

I'm not sure whether there's much sense in recording that "this was once a standalone track"

do you gain much from that?

prob not, just trying to understand

we can already create one-track digital releases if we want to think of the online track as a proper release

ocharles: any idea how I can get http://test.musicbrainz.org/static/scripts/test... working?

warp: with the dev server?

ocharles: yes.

warp: http://search.cpan.org/~mstrout/Catalyst-Plugin...

nikki, that was more my thinking, so they are put into their own release are they

that mentions that html is by default ignored

ijabz: yeah

ijabz: you can just add a 1-track digital single

warp: or perldoc Catalyst::Plugin::Static::Simple if you're saving bandwidth ;)

which is a single release

ocharles: yes, I know Catalyst::Plugin::Static::Simple is ignoring .html by default, but I cannot figure out how to make it stop doing that.

you just stuff ignore_extensions => [] into the config

one sec

So you just dont do that when the sonh has'n't been released/made available publicly

warp: add to __PACKAGE__->config (MB::S line 45): static => { ignore_extensions => [] }

ocharles: I already tried setting a new value in lib/MusicBrainz/Server.pm. tried ignore_extensions => [ ] and ignore_extensions => [ qw/php/ ] (just in case it would consider an empty list as a falsy value).

ijabz: although whether people do that depends on whether it seems like a proper release, there are plenty of cases where it would be really weird to call something that's currently a non-album track a digital single

warp: you have to put it in static => {} inside the config

ok, I'll do that again.

ocharles: still 500.

warp: i'll try then

Say a band put a song on their website available to download, but it is not actually a single what would u do

ijabz: normally, I add a track that is for sale as a single. A track in Bandcamp -> single. A track in a less final-user-oriented site, like Soundcloud -> NAT

warp: wait, you want this on test.mb.org or your server?

because test serves static with nginx, not plugin::static::simple

for me it's also has cover art -> single, tracks on myspace/facebook -> non-album track

ocharles: aaah, why is nginx 500-ing then? :)

warp: 500 seems odd, but I can see it giving permission denied or something

warp: let me look at the config a sec

rewrite or internal redirection cycle while internal redirect to "/static/scripts/tests/index.html"

warp: 2011/02/01 13:27:01 [error] 733#0: *38544 rewrite or internal redirection cycle while internal redirect to "/static/scripts/tests/index.html", client: 10.1.1.14, server: test.musicbrainz.org, request: "GET /static/scripts/tests/ HTTP/1.0", host: "test.musicbrainz.org"

oh, heh

you win!

:)

maybe:

rewrite ^\/(.*\/)index\.html$ /$1 ;

?

yeah, that's probably it.

but I'm not that familiar with writing nginx rewrite rules.

that's in /etc/nginx/mbserver-rewrites.conf if you want to try messing around

it's in git too, so you can go crazy

easiest is probably to prevent that rule from matching anything which starts with /static.

yea, that's what I'd try first

ocharles: I've read your email about the testing.

I dont really have any valuable feedback for you. :-(

warp: what do you think of this new testing approach?

ruaok: yea, I'm not really sure what feedback I can expect from people :)

ruaok: er. email? I'll have to get back to you.

how pervasive are the changes to the codebase, ocharles ?

267 files changed, 6519 insertions(+), 6095 deletions(-)

warp: speaking of email, did you get the email about watching the servers tonight?

oh fuck.

do we really want to do this before NGS?

test has been running it for a week now and while we did have bizzare hiccups, it's been for 4 days since

ruaok: the changes are mostly variable renaming

if it's the acid2/test-routine stuff I veto.

:)

warp: yes.

are kidding about that?

The one thing that does change is that now we connect once, and only once, and pass the same dbh around. I think I may change that to pass a DBIx::Connector around. This means that every query will do a ping on the dbh first, but it will try and reconnect if it fails

that's pretty much exactly what it was doing before (except it might ping once if there are 3 transactions coming up)

ruaok: yes, I'm tethered though so cannot do much watching. but I am near computers all day.

warp has changed the topic to: http://musicbrainz.org/#devel | call warp if servers are crashing, telephone number on http://frob.nl

thanks! :)

warp: I would love to get your feedback about ocharles test stuff too. when you have a chance to review/whatnot.

ruaok: not kidding. test-routine introduces the weird redirect bug I've been fighting last week. I don't want that stuff in master until we know what that is.

ah!

if you can try and trace that today I'd really appreciate it, because I can't reproduce on any of my machines

ok, if its introduced crazy bugs that have already caused us to lose many hours hacking, I would really prefer to wait with these changes til post NGS.

we need stability now, not crazyness.

yes. I just put up my guess case stuff on code review. I'll start stepping when I'm going offline again in a few minutes.

this is for stability so I can actually run the test

+s

as it stands now half of them just fail

ruaok: in general the goal of the patch sounds very good, so I'm not opposed to having that in for NGS. just need to find that bug.

but I understand the reservations, for sure

warp: and see if that bug really is introduced by my direct changes, or not something latent that's just come to surface

ok, lets compare notes on this again tomorrow.

ocharles: how should I go about debugging this again?