#musicbrainz-devel

which isn't what we want

Why we do not want use user language for sorting? (If I understand your anwers correcly... )

adhawkins: Ok, git remote update ocharles && git cherry-pick 7d64388

Alright, now to DROP DATABASE and try again with an older version of libicu and make sure it all still works

ocharles: Continuing. Will let you know if it catches up Ok :)

Surely it needs to be fixed properly in case anyone else does something similar in the future though?

http://tickets.musicbrainz.org/browse/MBS-3917 Localised sorting

adhawkins: the proper fix is to make sure we don't try and insert such huge data into indexes

I made a ticket for investigating that

Yep, that's what I was envisaging.

Have you done a catch up on your slave? Or might there be more?

Yea, my slave got all the way to now

Ok cool.

Hjum. maybe I should add git update before each replication packet import... :)

I'm sure mine will one day soon :D

InitDb.pl is a bit confusing... It really does not tell user what difference is --createdb, --clean and --import (and can they used together). Also --help gives outdated information (it contains options that not available on GetOptions())...

And I also do not like way that it does mixes database admin and musicbrainz using on postgresql interaction... (I've copied admin privileges needed commands out of it to separate script)

(and that was just comment why we may have so many initing database -related question on there)

I also yesterday installed (partially) search server. I didn't found any way define index_dir outside WEB-INF/web.xml. Something I wanted is some way define it on tomcat Context config....

(And I do not liked that search server servlet tests need port 8080 and it's hardcoded so many places...)

(I founded way define some random port on maven pom.xml, but tests code has that port also hardcoded.... And I do not really familiar with java)

Jozo: complaining about it on irc isn't going to fix any of that however. I agree with many of your complaints, but it's probably more helpful to either fix them or report them in our issue tracker.

kuno: I was just about to ask is some my notes worth for reporting it to jira......

musicbrainz jenkins is down?

yes

the whole machine is down

Oh. (maybe I install jenkins to my local machine just for fun... I've not used it for while)

that's a very warped sense of fun

:)

luks: inspired your mbdata/mbslave solr things I created musicbrainz core to my solr and used solr's DataImportHandler for importing and solr's velocity for browsing... works pretty well (I also used only one index..)

ocharles: All caught up now, thanks.

Why is artist_deletion not used ?

I am wondering why we have not deleted that artist from our db 259b602f-7f41-47dc-aecc-a5a75307b219

everything doesn't go though artist_gid_redirect i guess

but then i thought, maybe for deleted artists it goes in artist_deletion

but no sign of that gid in either

Hadora: artist_deletion is only used for subscription emails

(right now)

it helps us keep the subscription tables consistent, while also being able to do subscription summaries at mignight

midnight*

I see, ok

But i still wonder where that artist went

The deletion is a result of one of these edits http://musicbrainz.org/release/a6db3a98-7822-4e...

But how can i know automatically it has been deleted ?

You can add your own triggers to the artist table, and log deletions

yup i see, it's not something i missed

ty!

Jozo: np :)

ijabz: http://tickets.musicbrainz.org/browse/SEARCH-333 - mvn install builds war, so I assumed I can use it with tomcat. I'm not really familiar with servlets and paramaters (I always hated when I needed install some random app)...

ijabz: (I always wondered that I've must doing something wrong... ) When install instructions says that you have to modify some things source code before building... (or after that war is unpacked to somewhere) ... It's a bit painful to upgrade software when you have to repeat those steps everytime... (just for some directory paths or parameters)

http://musicbrainz.org/ws/2/tag/?query=sixteen%... authorization required?

Guest2323: I think the tag search only works for tags you used, so it needs to know who you actually are :)

Only tags you used?

Mineo: http://musicbrainz.org/ws/2/tag/?query=electronic has many public tags http://musicbrainz.org/tag/electronic

wel,, in that case it seems I was wrong

Maybe query is just wrong.. http://musicbrainz.org/doc/Development/XML_Web_... not mention any query like that

searches are documented at http://musicbrainz.org/doc/Development/XML_Web_...

(Oh, I would excepted that when you search something url is also contains search)

The wiki page gives an example query, which also asks for authorization.

So it's not, I assume, a matter of a poorly formed url.

http://musicbrainz.org/doc/Development/XML_Web_... do not mention anything that you have to be logged on... so bug

Jozo yes you can use Tomcat, whats your question ?

ijabz: Just what I said on ticket... (I thougt that you thinked that it's not bug but just some intentional thing to do so)

ijabz: What is more generic way? (modify source and build or build something and then after unpack something and then modify?)

ijabz: Just say me that I've wrong and there is way to define parameters to that war (without unpacking it and modifying some files)

Well no you are right, you have to modify the web.xml file if using a nonstandard location, whther you do this before building or after unpacking depends on your own circumstances

ijabz: And that is something that I have to tolerate?

But even if you dont unpack the war yourself the servlet container usually unpacks it for you anyway

ijabz: So I have to do that everytime I update search server code?

Hi all - looking for some Picard insight - anyone here who can help?

No, if you are checkng out direct from svn like me your modified files will not get overwritten, so you only have to modify web.xml once

ijabz: Why you/we want keep that hardcoded thing and not making it to some more elegant way?

Its not hardcoded in the code, in a configuration file

ijabz: It's hardcoded if you have to modify SOURCE before building.

You dont have to, you can modify the deployed file

I dont like the way the instructions are written make it sound more difficult than it is

Modifying the source file is just useful, because then you only have to do it once avoiding the annoyance you are encountering

ijabz: That's why I hate java and every java-coder.... (no offence)

hardcoded paths, hardcoded variables, hardcoded everything... and java is everywhere still... haha

Its not hardcoded, it is a configuration file, you could keep your own configuration file somewhere and just overwrite the existing one with that if you want

its anot a java issue

But your solution is to use tomcat specific features to have it working on tomcat, use jetty specific features to make it work on jetty ectera

how does that make it easier

ijabz: Why many project let you define context PARAMATERS: http://tomcat.apache.org/tomcat-7.0-doc/config/...

ijabz: And that is not tomcat specific (if I understand correcly)

ijabz: What is difference init-param and context-param?

(I readed some answer, but I want hear your answer)

I dont know

conguring servlet can be a real pain, which is why I prefer to keep it simple

ijabz: Why we use init-parm instead of context-param?

(I've not read any java specs, but they may give answer that)

Dunno, cant remember why, who cares ?

ijabz: I cared

ijabz: One can be overwritten other one can't

Why can one be overwritten and the other cannot ?

And yeah, everytime I see java, it's just real pain...

what language do you like then ?

ijabz: I do not know. But tomcat do not provide any way overwrite them... And I'm pretty sure that jetty do not allow either (You can prove that I've wrong just prove that Jetty can do something that tomcat can't)

Do you mean the file, you can overwrite the web.xml file

ijabz: It's not language what I like. Everytime I come across Java there is some hardcoded things or some other weirdes...

ijabz: If jetty provides that functionality I happily which to jetty...

As I understand it context xmls file are vey similar to web.xml so I dont really get the improvement

Feel free to add a more concrete propsal to the ticket

ijabz: So next I have to paste some Java servlet documentation to you?

ijabz: What version do you want?

jozo ffs you need to explain why having it moved out of one configuration file into another is going to improve things fo ryou

ijabz: I do not want It to moved anywhere, I just want way to overwrite it. And I suggested that context-param may be better than init-param. (And yeah that makes me to read java specs...)

Hi all - looking for some Picard insight - anyone here who can help?

Ive already said i dont know the difference, going for some noshg

ijabz: Do you have some well written java servlet your mind?

ijabz: https://github.com/jenkinsci/jenkins/blob/maste... <!-- if specified, this value is used as the Hudson home directory -->

ijabz: That is pretty common way to do that... (What I've read jira, confluence and other sources)

Sophist-UK: Asking your question will allow people to see if it's something they can help you with.

ijabz: And why we stuck on jira 4.0.1 (has some security bugs)... I just guess that It's just too painful to update

ijabz: And /home is not correct directory to save some application data. http://www.pathname.com/fhs/

Jozo: ijabz is not in any way shape or form responsible for the version of Jira we're running. There's no reason to ping him about that.

Freso: That's not just symbol what some "do this and this and this" on actual program data instead of proper configuration files...

Freso: s/not//

Jozo: That did not make any sense.

Ok - I have started to see a recordingtitle tag appearing in the metadata box. Doesn't happen in 1.2, does happen in latest github. Cannot track down how this is set - I raised a PR a couple of months ago (merged) which adds a ~recordingtitle, but I have NO idea where the recordingtitle tag comes in. Have searched the source code and cannot find anything which would cause this.

maybe someone wants to share server installation tutorial from scratch?

Freso: Basically: "You have that .war file, but you have to modify this and this and this file before you can upgrade"

Freso: (Just for define some directory or some other thing)

Jozo: I have no idea what you're getting at or how it's related to ijabz (not) being responsible for our Jira install.

Freso: http://tickets.musicbrainz.org/browse/MBS-7033

Freso: Oh wrong

Freso: http://tickets.musicbrainz.org/browse/SEARCH-333

Jozo: How is that related to the version of our Jira install?

Freso: (Not really relate) Jira was just one example where that "edit that, edit that, edit that" methods is used (instead of defining those on some your own configuration file)

Freso: Who is responsible of MusicBrainz's Jira?

Freso: What you think hardcoded paths in general? (Do you use them on python programs?

I suggest that we should hardcode all musicbrainz related data paths to /var/lib/musicbrainz =)

Freso: And if you use hardcoded path do you except that users overwrite them on source code?

I think you're being intentionally daft, Jozo. Have fun with that.