#musicbrainz-devel

interesting comment\

I was kind of suprised (I start using the module for isrcsubmit2)

right

I guess it's "you eventually get releases"

which is why it's get_releases_by_*

Well, it looks like somebody wrote like "what he needs", not what is provided

So from the point of a user, not a dev of the API

I guess it's because the XML returned from the ws has the <discid> element

or whatever

A specific user, that is. It does get weird when you DO want the discid request

hmm, right

what else does a discid give you?

sectors/toc

and do you need to inc=releases to make it useful anyway?

ok

the thing is, I'm not sure a function get_discid() is very useful

or maybe we need a get_discid, and a wrapper, get_releases_by_discid that returns just a release list

I want to do an api rewrite for 0.5

lots of things I want to change/fix

well, I also need get_releases in terms of actually getting releases, though

but you can just use get_release_by_id for that?

The point is, I access the "release_list" with ["disc"]["release-list"]

yes, from the technical point it works, after I found out that this gives me the discid, not a release-list

So this is only a usage issue

ok, and the function name made you think it should give you a release-list?

A relevant one though, since we have no wrapper, but expose the dicts directly

Yes.

if that's the case then my preference would be to have 2 methods ^^\

get_discid will give a <discid>, get_releases will parse that and just return the list

That would work.

Like I said, I am actually interested in get_releases, it's just that the usage is not at all straightforward.

Do you open a ticket for 0.5 (api change)?

Or should I

if no one else is using it, I'm happy to break api on the next release

file against 0.3\

check if beets or headphones use it

Well, there is samsyo

sampsyo: ?? :)

asampson

beets dev

Mineo has a isrcsubmit variant that uses the function, but I guess that is no big deal

nothing in beets

headphones uses _by_discid

no, wait

it just has its own copy of mbngs

Well, then they can decice if they update it or not.

so I'm happy to put it in the next release, or wait until the api break release

fingers crossed next release will be within in a month

For isrcsubmit2 I could wrap it (test which dict I get). So it's up to you when to include the API change.

k

can you remember who the debian maintainer is?

I have a bunch of emails from him

but can't remember his name

The earlier it is, the earlier people can adapt and/or start using pymbngs with the new API

nope

For debian you have to check if you want to include it in wheezy or not

oh, it was a netbsd guy

just replied to him

Currently the 0.2 API will be in Debian wheezy. (stable)

when's wheezy planning to go stable?

No real plan, but they release RC1 in february

*released

so it's like "anytime now"

right

so 0.3 definitely, probably not 0.4

well, definitely

hopefully

morituri seems to use _by_discid: http://packages.debian.org/changelogs/pool/main...

huh, right

let me email Jonas

he says it's fixed -- is it?

or does your patch fix it?

that is the fix I called "hiding fix"

I can't find the debdiff

I think, checking atm

yes: http://patch-tracker.debian.org/patch/series/vi...

The error is gone, but the includes don't work right, unless given as explicit keyword

OK

I'm tempted to spin a 0.4 in the next few days then with your current fix, no api change

Don't rush things though.

That problem persists for a year already and nobody noticed. So we have enough time for additional review.

I'll email Jonas, see what his status is - if he just replied to the morituri bug, or is now a maintainer

do you know if packages.debian.org gives rdepends output? (I can see depends)

No, I don't know.

I /suppose/ I could install debian in a vm :-P

On Arch Linux only "beets" needs python-musicbrainzngs atm. Morituri doesn't (morituri 0.2.0)

ok

yeah, same in the latest ubuntu

I also don't have an unstable or testing debian VM atm

installing testing now

heh, mirrors on peer'd networks are fast :)

I'm getting a start up error: "encode_base64url" is not exported by the MIME::Base64 module

Any ideas?

navap: yes

navap: http://chatlogs.musicbrainz.org/beta/%23musicbr...

Awesome, thanks

JonnyJD: ok, finally got experimental merged

morituri is an rdepends

0.2.0-1

only that and beets,though

squeeze and sid/wheezy are still at 0.1.2

experiemental is 0.2

are you sure? http://packages.qa.debian.org/m/musicbrainzngs.... says 0.2-1 is in testing

experimental is 0.2-1.1, according do that

unless this mirror hasn't updated yet, apt-cache policy says sid has 0.1.2-2

oh

I mean morituri

not pymb

ah, k

Yes, that is because of wheezy freeze, I guess. Libdiscid also isn't progressing to unstable.

right, yeah

So that means beets and morituri need to be patched or updated when pymbngs gets into debian etc. with a different (backwards incompatible) api.

yes

we can submit an issue to them before the release saying the api is changing

I may consider bumping the major version so make it clear there's a break

Yes, that is a possibility. semver.org says 0.x API isn't supposed to be stable, but pymbngs is already out for some time.

Libdiscid is also stable in 0.x releases ;-)

That also means you are supposed to keep making bugfixes for the 0.x branch then.

that's exactly what I was thinking of

"technically", breaking an api on 0.x is fine :)

I guess I would have lots of angry mails when I would introduce a backwards incompatibe change in Libdiscid after 7 years :-D

So much for "technically". You have to talk about that with the beets and morituri devs, I guess.

For isrcsubmit I kept working on isrcsubmit1, because pymbngs isn't supposed to be stable anyways ;-)

sure

heh, got a nice email from the debian packager

saying "how nice that you email the package maintainers after you make a release"

it turns out the easy way to do it is to just wait for their magic system to realise there's a new version on pypi

blah, I hate aliases. is “サンセット・フライト” Japanese, or is it English (in katakana script)?

kinda wrong channel but whatever

alastairp: yes, they have this "watch" file. So after 1-2 days it turns up in the dashboard and possibly they also get a mail. When you send a mail you can tell them about important things though.

So I also send mails/open update requests for "important releases".

right

I've sub'd to the package, anyway, so I can see any updates that happen there

hawke: try http://nihongo.j-talk.com/ and see if the katakana is anything remotly soundign logic in english http://nihongo.j-talk.com/

CatCat: →#musicbrainz

sunset fright?

what?

but you *Asked* here

yeah I re-asked in #musicbrainz

:-)