Johannes Dewender

When we change names like that (or add aliases), there are possibly other things we should change. As mentioned in http://lists.musicbrainz.org/pipermail/musicbrainz-devel/2013-April/005223.html `artist-credit` is a misleading attribute name, since it is...

We already have a previous issue: #108 where I suggested to create some form of tree of elements/attributes. That would be great and easier to maintain, but more difficult to...

My list: - #94: 404 errors currently give `ResponseError`, returning `None` would be an API change (that is what alastair mentined as his first point) - #100: fix function names...

Yes, this package is in debian, even with a version in stable: http://packages.qa.debian.org/m/musicbrainzngs.html With isrcsubmit I was using 0.x numbers, too and I just released this as 1.0 and started...

I forgot one important thing: **musicbrainzngs (0.x and 1.x)** could live happily besides **musicbrainzngs2 (2.x)**. So one could have both installed. That would technically also be true for _musicbrainzngs (0.x)_...

> No need for a new Python module name. See, for example, requests, which recently broke its API for the 1.0 release. I count 10 rdeps for python-requests on Debian....

Alastair Porter wrote at Thursday 04 July 2013 04:23:49: > > Personally, I see no problems with breaking the API when going from > > 0.x to 1.x, or 1.x...

I consider this an apichange. People might run into problems expecting a bytestring at some point, trying to do some decoding and failing since you can't decode a (non-ascii) unicode...

I would only disallow the includes if the server doesn't allow them (or deprecates them as "not intended").

Well, yes. "kind of" related to #80. Error code 400 is a real exceptional case though. Code 404 is a standard occurance when you try to check if an ID...