#musicbrainz

I didn't intend for people to look at it...

ah, I got it to display the relationships at the bottom, but the track numbers are missing

Maybe I should just do it as an artist credit on the releases/recordings in question.

I had a horribly complicated algorithm for the track number ranges in the old Python NGS code

That seems to fit reasonably well

oh, I'd probably add a tracks variable to the objs with a list of the track numbers

Might confuse people when the link of the name artist A goes to artist B…

So it doesn’t exactly make the situation clear.

But I guess it would work.

heh http://bugs.musicbrainz.org/browser/mb_server/b...

you were porting mbserver to python?

I was writing a new mbserver specifically for NGS

hawke_: it would probably work, yes

oh, why was it scrapped?

Maybe indicate it in an annotation

complicated :)

okay :)

there was some work done on the perl code

and some work by me on the python code

so one of them had to be dropped

I see

I think it was a bad decision now

but it seemed more sensible back then

learn perl then :P

eh, perl isn't that hard

I hadn't touched a line of perl before this summer

probably not, I just never got around to it

and I'm certainly no perl master but I can certainly read through it

I really dislike the "modern" Perl

I have a few perl books I was supposed to read but never did

and, well, perl itself isn't so bad... it's more Moose/Catalyst/etc

which is a different way of saying what luks just did, I suppose

luks: oh?

it's trying to be too magic

ruby syndrome? :P

the funny part is that object creation is the bottleck of the current server code

which is really a wtf

luks: I don't think it's *trying* to be magic, it's more that unless you want to live with a load of gotchas, you have to re-invent the rules

I'm not sure the object system is the bottleneck either, but it's a performance problem in some cases, that's for sure (and I agree, that's a shame and confusing)

Then again, I prefer Moose than Python objects, so I don't think we'll ever share opinions here :)

well, the number of functions to wrap/create methods is really too large

it's hard to compose them in your head to get the global picture

I'd say that's my main reason to not like Moose

which part has too much? our architecture, or moose's api?

moose

you have to read many files to get an idea what the method actually consists of

interesting, I honestly haven't heard complaints about Moose's api before. I mean, it's just Perl still, but with the addition of 'has', roles and method modifiers

right, but how else are you going to remove the repetition?

I think we have different styles too, I think you abstract later than I do

but explicitly calling other methods, like I'd have to do in Python, or Java, or C++

by

Yes, but then you still have to remember to call those methods, it's still repetition

my problem is reading the code, not writing it :)

if I forget to call a method, it's completely my fault

yes, but I'd rather not have that risk

because I do forget things a lot

If the framework can force that for me, then I tend to play to it's strengths

I suppose as I work on this for 6 hours a day, the structure is quite firmly engrained in my head though

the edit methods in MB's controllers is what got me thinking about this recently

the roles?

roles, which dynamically create methods, that wrap other methods

I'm not sure they are done as well as they could be. They worked ok at the time, but now they are becoming a bit hard to work with, I'll agree

it was either spending hours to track it down or write what I needed from scratch

I see that more as a problem with Catalyst though, not Moose. I don't like the controller = class model anymore

I see it as a moose problem

because normally you would have to call the other methods

I prefer that a controller is a collection of objects, one for edit, one for viewing, etc

I can follow such calls in my head just fine

I just haven't managed to decompose things with methods or functions and not have code diverge

Roles split stuff apart, but at least there's usually a single point to go to, you just have to spend more time to hunt it down

I'd rather spend more time understanding the abstraction rather than spending time making sure I've updated all usage of a code, especially as Perl is so notorious to parse

I always struggle with that, but I just decide that I don't care and go with the more practical solution

I don't find it practical in the long run, I find it leads to regressions and sloppy coding

because I start to get very tired of working with the duplication

web code is never going to live for too long

since I started doing websites I was trying to find a way to abstract it, but then I realized it's a waste of time

hmm, /artist/edit which works very similar to /url/edit is not going to go away

it's pretty much why I accepted jQuery as something that I can use :)

all methods on a single class, what kind of API is that? :P

basically, /artist/edit needs a form, then it needs to submit it, then it submits an edit. I could factor that into a higher order function, or I could use a role to guarantee at least the parts fit together get at pseudo-compile time

I'm curious to see how the code you wrote looked though

i'm certainly open to changing things

the code I wrote was for splitting artists

thought so, is the branch public?

not yet, don't have access to the machine right now either

ok, well push it when you get a chance

Hmm, URL relationships will keep an artist from being deleted?

yes

I wonder what happened to the artist entry linked to http://127.0.0.1/

That seems bad: http://musicbrainz.org/artist/d40d4499-73b8-44d...

hawke_: why?

The artist exists, right?

we have a bunch of artists which only have url relationships and nothing else

reosarevok: Because it keeps artists with no useful information around. Isn’t this musicbrainz, not artistbrainz?

it bugs me when people arrive and just add the artist, a url or two but nothing else, even if they *could* do

sometimes they just try and link a bunch of itunes links to the artist ¬_¬

Well, I wouldn't say this is "not useful"

once it's in MB, I wouldn't delete it though

ocharles: you still want some MCs? :p http://musicbrainz.org/artist/d40d4499-73b8-44d...

even if it's not useful right now, at least keep its MBID

reosarevok: ooo

For what, though? What purpose does the MBID serve with no releases, performances, works, nothing by the artist

I should put myself in Musicbrainz, because hey, I might record something some day!

I expect the releases will appear eventually

hawke_: the guy has a youtube channel and a myspace

So he's clearly a musician of sorts

luks: oh you're optimistic :P

well, they have some tracks on myspace

nikki: oh, I don't mean added by the artist :)

heh

I tried keeping those live comments short

this is too long, IMO

gps coordinates, no one can argue with that

maybe just city, country

rather than venue city country (unless of course there are two from the same day in different venues

btu that seems unlikely...

hawke_: you could add Person and UK while deleting the comment ;p

heeelp

was there a decision about the remix works?

or at least, which side is winning? :)

luks: "different ARs or ISWC -> new related work, if not, merge" seems to be it

or if I have just different remix/producer ARs, merge?

s/or/so/

luks: can you check it against a works DB?

no

:(

I imagine you could merge

What's it?

some people might vote no though

I really don't care either way

but I want to know how to vote on such edits :)

I think they're fine

but if you don't care, maybe you should abstain :P