Discussion:
[mb-users] ngs csg
(too old to reply)
David Hilton
2011-11-16 03:08:46 UTC
Permalink
I worked today on 'Hans Hotter singt Lieder', a compilation album (
http://musicbrainz.org/release/205cfe1a-0f49-48dd-9ada-ecd816cb38ed).

This is my second foray into adding a release in ngs; the first time was
pretty involved. This time took a good while; about 6 hours of, I think (I
like to get all of the metadata I can).

Does anyone see significant issues with the data as it has been entered?

Here are some specific questions:
1. the new csg - work names, recording names, and track names are all
distinct values. I'm presuming that work names follow the old csg, while
track names follow the cd? I followed this, but then started to change to
the csg (tracks 2 and 4) before second-guessing myself. I have no idea
which value recordings should have.
2. artist tags: on tracks, I believe these now point to the primary
performing artist, instead of the composer.
3. part-of work relationships: I would think that parent relationships not
found at a child level are inherited (see track 14 as an example). I added
Robert Franz' Op. 10, putting Robert Franz as the composer of the top-level
work, created sub-works for individual pieces and put the lyricist of each
piece on the works they apply to. Picard doesn't seem to interpret it this
way, though, so perhaps I should have done it differently.


David
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111115/9819ee8d/attachment.htm
Rupert Swarbrick
2011-11-16 08:21:32 UTC
Permalink
Post by David Hilton
1. the new csg - work names, recording names, and track names are all
distinct values. I'm presuming that work names follow the old csg, while
track names follow the cd? I followed this, but then started to change to
the csg (tracks 2 and 4) before second-guessing myself. I have no idea
which value recordings should have.
You're not the only one on the recording names. I don't think there's
been any consensus yet. In my opinion, this is because it doesn't really
make sense for the recording to have a name at all (although how one
would display it without a name is also a problem...)
Post by David Hilton
2. artist tags: on tracks, I believe these now point to the primary
performing artist, instead of the composer.
Most people (including me) seem to still be pointing to the composer as
the track artist. Was there some decision to change that? Does it also
apply to, say, orchestral releases? Concertos? (Presumably in the latter
case we should attribute the track to at least three different artists...)
Post by David Hilton
3. part-of work relationships: I would think that parent relationships not
found at a child level are inherited (see track 14 as an example). I added
Robert Franz' Op. 10, putting Robert Franz as the composer of the top-level
work, created sub-works for individual pieces and put the lyricist of each
piece on the works they apply to. Picard doesn't seem to interpret it this
way, though, so perhaps I should have done it differently.
There's an RFC going through at the moment about how inheritance for ARs
in works should be interpreted. What you've done agrees with the RFC I
think (and seems sensible). Presumably Picard will eventually follow
suit.


Rupert
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 315 bytes
Desc: not available
Url : http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111116/34e9137f/attachment.pgp
symphonick
2011-11-16 09:35:04 UTC
Permalink
2011/11/16 Rupert Swarbrick <rswarbrick at gmail.com>
Post by Rupert Swarbrick
Post by David Hilton
1. the new csg - work names, recording names, and track names are all
distinct values. I'm presuming that work names follow the old csg, while
track names follow the cd? I followed this, but then started to change
to
Post by David Hilton
the csg (tracks 2 and 4) before second-guessing myself. I have no idea
which value recordings should have.
You're not the only one on the recording names. I don't think there's
been any consensus yet. In my opinion, this is because it doesn't really
make sense for the recording to have a name at all (although how one
would display it without a name is also a problem...)
No consensus. I'd like to see track names, some editors want work names.
(I'm afraid that won't work because tracks & works don't always match).
Post by Rupert Swarbrick
Post by David Hilton
2. artist tags: on tracks, I believe these now point to the primary
performing artist, instead of the composer.
Most people (including me) seem to still be pointing to the composer as
the track artist. Was there some decision to change that? Does it also
apply to, say, orchestral releases? Concertos? (Presumably in the latter
case we should attribute the track to at least three different artists...)
AFAIK there has been no suggestions regarding track artists. I just leave
them at the default value (=release artists).
Post by Rupert Swarbrick
Post by David Hilton
3. part-of work relationships: I would think that parent relationships
not
Post by David Hilton
found at a child level are inherited (see track 14 as an example). I
added
Post by David Hilton
Robert Franz' Op. 10, putting Robert Franz as the composer of the
top-level
Post by David Hilton
work, created sub-works for individual pieces and put the lyricist of
each
Post by David Hilton
piece on the works they apply to. Picard doesn't seem to interpret it
this
Post by David Hilton
way, though, so perhaps I should have done it differently.
There's an RFC going through at the moment about how inheritance for ARs
in works should be interpreted. What you've done agrees with the RFC I
think (and seems sensible). Presumably Picard will eventually follow
suit.
Rupert
_______________________________________________
MusicBrainz-users mailing list
MusicBrainz-users at lists.musicbrainz.org
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-users
--
/symphonick
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111116/6fc6f36a/attachment.htm
David Hilton
2011-11-16 09:50:39 UTC
Permalink
Post by symphonick
2011/11/16 Rupert Swarbrick <rswarbrick at gmail.com>
Post by Rupert Swarbrick
Post by David Hilton
1. the new csg - work names, recording names, and track names are all
distinct values. I'm presuming that work names follow the old csg,
while
Post by David Hilton
track names follow the cd? I followed this, but then started to change
to
Post by David Hilton
the csg (tracks 2 and 4) before second-guessing myself. I have no idea
which value recordings should have.
You're not the only one on the recording names. I don't think there's
been any consensus yet. In my opinion, this is because it doesn't really
make sense for the recording to have a name at all (although how one
would display it without a name is also a problem...)
No consensus. I'd like to see track names, some editors want work names.
(I'm afraid that won't work because tracks & works don't always match).
Right, like the last movement of Beethoven's 9th, or operas where track
divisions sometimes vary wildly. I can definitely see a 'work-name
override' as being a useful option in picard, though. I'm not really
satisfied with limiting myself to either option, to be honest.

Having tracks listings reflect the album cover (or the current listing, as
long as it is internally consistent), and adding a work-name metadata field
would work for me - but standardizing on this just adds another dimension
of cleanup/ngs data upgrading that we have to do.


David
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111116/2a3e4cae/attachment-0001.htm
Rupert Swarbrick
2011-11-16 10:01:59 UTC
Permalink
Post by David Hilton
Having tracks listings reflect the album cover (or the current listing, as
long as it is internally consistent), and adding a work-name metadata field
would work for me - but standardizing on this just adds another dimension
of cleanup/ngs data upgrading that we have to do.
"A work-name metadata field": Isn't that usually called a recording-work
is-a-performance-of relationship? :-)

Rupert
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 315 bytes
Desc: not available
Url : http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111116/553043a8/attachment.pgp
David Hilton
2011-11-16 10:11:25 UTC
Permalink
Post by Rupert Swarbrick
Post by David Hilton
Having tracks listings reflect the album cover (or the current listing,
as
Post by David Hilton
long as it is internally consistent), and adding a work-name metadata
field
Post by David Hilton
would work for me - but standardizing on this just adds another dimension
of cleanup/ngs data upgrading that we have to do.
"A work-name metadata field": Isn't that usually called a recording-work
is-a-performance-of relationship? :-)
:) Yeah.

What I /meant/ to say was, saving the work-name field in picard would work
for me.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111116/601dfaf2/attachment.htm
caller#6
2011-11-16 17:38:14 UTC
Permalink
David Hilton <quercus.aeternam at gmail.com
Post by David Hilton
1. the new csg - work names, recording names, and track names
are all
Post by David Hilton
distinct values. I'm presuming that work names follow the old
csg, while
Post by David Hilton
track names follow the cd? I followed this, but then started to
change to
Post by David Hilton
the csg (tracks 2 and 4) before second-guessing myself. I have
no idea
Post by David Hilton
which value recordings should have.
No consensus. I'd like to see track names, some editors want work
names. (I'm afraid that won't work because tracks & works don't always
match).
I've been trying to steer a middle-course on this. In my mind

Work Title: the ur-title (in the ur-language), referring to the
"platonic ideal" composition, normalized using CSG title style. (use
aliases for transl*tions)

Recording Title: the track title (in the language of the tracklist),
normalized using CSG style. This might differ from the Work Title based
on instrumentation, track splits, etc. (in the case of multiple
languages, I don't think there's a one-size-fits-all answer for which to
use (wait for recording aliases?))

Track Title: the track title normalized using non-CSG-specific MB style.
i.e. capitalization, subtitle style, part style. (use release language;
use official transl*tion for multi-language tracklist)

Alex / caller#6
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111116/8b822961/attachment.htm
symphonick
2011-11-16 18:59:27 UTC
Permalink
I think I agree with you, but I'm not sure what you expect the "csg-style"
normalization to do with the track titles for recording names? Can you give
an example of when the usual MB normalization is insufficient?

/symphonick


2011/11/16 caller#6 <meatbyproduct-musicbrainz at yahoo.com>
Post by symphonick
**
Post by David Hilton
1. the new csg - work names, recording names, and track names are all
distinct values. I'm presuming that work names follow the old csg,
while
Post by David Hilton
track names follow the cd? I followed this, but then started to change
to
Post by David Hilton
the csg (tracks 2 and 4) before second-guessing myself. I have no idea
which value recordings should have.
No consensus. I'd like to see track names, some editors want work names.
(I'm afraid that won't work because tracks & works don't always match).
I've been trying to steer a middle-course on this. In my mind
Work Title: the ur-title (in the ur-language), referring to the "platonic
ideal" composition, normalized using CSG title style. (use aliases for
transl*tions)
Recording Title: the track title (in the language of the tracklist),
normalized using CSG style. This might differ from the Work Title based on
instrumentation, track splits, etc. (in the case of multiple languages, I
don't think there's a one-size-fits-all answer for which to use (wait for
recording aliases?))
Track Title: the track title normalized using non-CSG-specific MB style.
i.e. capitalization, subtitle style, part style. (use release language; use
official transl*tion for multi-language tracklist)
Alex / caller#6
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111116/bebcf8b5/attachment.htm
David Hilton
2011-11-16 19:10:43 UTC
Permalink
Post by symphonick
I think I agree with you, but I'm not sure what you expect the "csg-style"
normalization to do with the track titles for recording names? Can you give
an example of when the usual MB normalization is insufficient?
/symphonick
Mostly I think that the csg should be replaced with a wsg (work style
guide).

I see 3 big areas the old csg effects current stuff: release title,
release/track artist, and track formatting.

Of the 3, I only think it should be retained for the release title, as
follows:

1. release title formatting (retain for classical works, because all-too
frequently there /is/ no real title)
2. release/track artist field (go to MB-wide standard, performer(s))
3. track formatting (use release info for listing)
4. work names (use csg conventions for works - possibly with some minor
logic, such as prepending the has-work information to reduce information:
ie. 'Moerike-Lieder' and '22. Seufzer "Dein Liebesfeuer"' automatically
turning to 'Moerike-Lieder: 22. Seufzer "Dein Liebesfeuer"').

I don't see a compelling argument for having recording names, other than
the fact that it's hard to reference them by number.


David
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111116/b671d247/attachment-0001.htm
David Hilton
2011-11-16 19:13:03 UTC
Permalink
slight correction: s/old csg effects current stuff/csg effected pre-ngs
data/

On Wed, Nov 16, 2011 at 12:10 PM, David Hilton
Post by David Hilton
Post by symphonick
I think I agree with you, but I'm not sure what you expect the
"csg-style" normalization to do with the track titles for recording names?
Can you give an example of when the usual MB normalization is insufficient?
/symphonick
Mostly I think that the csg should be replaced with a wsg (work style
guide).
I see 3 big areas the csg effected pre-ngs data: release title,
release/track artist, and track formatting.
Of the 3, I only think it should be retained for the release title, with
1. release title formatting (retain for classical works, because all-too
frequently there /is/ no real title)
2. release/track artist field (go to MB-wide standard, performer(s))
3. track formatting (use release info for listing)
4. work names (use csg conventions for works - possibly with some minor
ie. 'Moerike-Lieder' and '22. Seufzer "Dein Liebesfeuer"' automatically
turning to 'Moerike-Lieder: 22. Seufzer "Dein Liebesfeuer"').
I don't see a compelling argument for having recording names, other than
the fact that it's hard to reference them by number.
David
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111116/6b41e5eb/attachment.htm
Rupert Swarbrick
2011-11-16 23:57:03 UTC
Permalink
Post by caller#6
Recording Title: the track title (in the language of the tracklist),
normalized using CSG style. This might differ from the Work Title
based on instrumentation, track splits, etc. (in the case of multiple
languages, I don't think there's a one-size-fits-all answer for which
to use (wait for recording aliases?))
Ok, but what do you do if a recording is used on two different releases
with massively different track names? And different languages?

Rupert
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 315 bytes
Desc: not available
Url : http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111116/9aa4b249/attachment.pgp
David Hilton
2011-11-16 09:36:41 UTC
Permalink
Post by Rupert Swarbrick
Post by David Hilton
1. the new csg - work names, recording names, and track names are all
distinct values. I'm presuming that work names follow the old csg, while
track names follow the cd? I followed this, but then started to change
to
Post by David Hilton
the csg (tracks 2 and 4) before second-guessing myself. I have no idea
which value recordings should have.
You're not the only one on the recording names. I don't think there's
been any consensus yet. In my opinion, this is because it doesn't really
make sense for the recording to have a name at all (although how one
would display it without a name is also a problem...)
Heh, that does make sense. What about the track listing vs. work title?
Post by Rupert Swarbrick
Post by David Hilton
2. artist tags: on tracks, I believe these now point to the primary
performing artist, instead of the composer.
Most people (including me) seem to still be pointing to the composer as
the track artist. Was there some decision to change that? Does it also
apply to, say, orchestral releases? Concertos? (Presumably in the latter
case we should attribute the track to at least three different artists...)
The main reason I see for changing this is to make the work-level
performance listing usable:
http://musicbrainz.org/work/19c3772e-be6f-3302-a7e0-59c5cda97dfb

If every recording of a work is performed by Beethoven, that field becomes
pretty useless.
http://musicbrainz.org/work/f76ea5fb-9579-31ce-817f-d3b820d9ef0f

Which one was performed by Kim Borg?
(also, that recording attributed to Liszt reminds me: I haven't seen
arrangement/transcription work relationships as an option just yet, though
I seem to recall there being some discussion about it)

Anything that could go in the artist field should also be listed under a
more advanced relationship, so it's not a matter of data loss, but rather
which values make the most sense. I /am/ approaching this from the work
perspective. I'm having trouble coming up with any reason to have the
field contain a composer. What am I missing?


Thanks,
David
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111116/6253e1f1/attachment.htm
Rupert Swarbrick
2011-11-16 10:09:16 UTC
Permalink
Post by David Hilton
Post by Rupert Swarbrick
Post by David Hilton
2. artist tags: on tracks, I believe these now point to the primary
performing artist, instead of the composer.
Most people (including me) seem to still be pointing to the composer as
the track artist. Was there some decision to change that? Does it also
apply to, say, orchestral releases? Concertos? (Presumably in the latter
case we should attribute the track to at least three different artists...)
The main reason I see for changing this is to make the work-level
http://musicbrainz.org/work/19c3772e-be6f-3302-a7e0-59c5cda97dfb
If every recording of a work is performed by Beethoven, that field becomes
pretty useless.
http://musicbrainz.org/work/f76ea5fb-9579-31ce-817f-d3b820d9ef0f
Which one was performed by Kim Borg?
Oh, absolutely. But this is a perfect example for me too :-) I can ask
two questions: (1) Which one was sung by Kirsten Flagstad? (2) Which one
has Edwin McArthur playing the piano?

Both of these questions are reasonable and, for a violin sonata (say)
there isn't really a primary performer at all. My solution would be to
improve the interface so that the track artist wasn't the only thing
shown.

As far as I can see, changing the guidelines for the track artist may or
may not be worth doing for other reasons, but it won't fix the problem
you're showing here.
Post by David Hilton
(also, that recording attributed to Liszt reminds me: I haven't seen
arrangement/transcription work relationships as an option just yet, though
I seem to recall there being some discussion about it)
Yeah, I think that's still slightly up in the air too. I'm in the
lots-of-works camp, but I'm not sure how others feel.
Post by David Hilton
Anything that could go in the artist field should also be listed under a
more advanced relationship, so it's not a matter of data loss, but rather
which values make the most sense. I /am/ approaching this from the work
perspective. I'm having trouble coming up with any reason to have the
field contain a composer. What am I missing?
Well, just the historical one. People like their jukebox applications to
have performances of Beethoven's 9th under "Beethoven". And these
jukebox applications can only deal with having one artist on a file (and
don't respect the composer tag, even if there is one).

Rupert
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 315 bytes
Desc: not available
Url : http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111116/e421604b/attachment.pgp
David Hilton
2011-11-16 19:01:42 UTC
Permalink
2011/11/16 Rupert Swarbrick <rswarbrick at gmail.com>
Post by Rupert Swarbrick
Post by David Hilton
1. the new csg - work names, recording names, and track names are all
distinct values. I'm presuming that work names follow the old csg, while
track names follow the cd? I followed this, but then started to change
to
Post by David Hilton
the csg (tracks 2 and 4) before second-guessing myself. I have no idea
which value recordings should have.
You're not the only one on the recording names. I don't think there's
been any consensus yet. In my opinion, this is because it doesn't really
make sense for the recording to have a name at all (although how one
would display it without a name is also a problem...)
No consensus. I'd like to see track names, some editors want work names.
(I'm afraid that won't work because tracks & works don't always match).
Post by Rupert Swarbrick
Post by David Hilton
Post by Rupert Swarbrick
Post by David Hilton
2. artist tags: on tracks, I believe these now point to the primary
performing artist, instead of the composer.
Most people (including me) seem to still be pointing to the composer as
the track artist. Was there some decision to change that? Does it also
apply to, say, orchestral releases? Concertos? (Presumably in the latter
case we should attribute the track to at least three different
artists...)
Post by David Hilton
The main reason I see for changing this is to make the work-level
http://musicbrainz.org/work/19c3772e-be6f-3302-a7e0-59c5cda97dfb
If every recording of a work is performed by Beethoven, that field
becomes
Post by David Hilton
pretty useless.
http://musicbrainz.org/work/f76ea5fb-9579-31ce-817f-d3b820d9ef0f
Which one was performed by Kim Borg?
Oh, absolutely. But this is a perfect example for me too :-) I can ask
two questions: (1) Which one was sung by Kirsten Flagstad? (2) Which one
has Edwin McArthur playing the piano?
Both of these questions are reasonable and, for a violin sonata (say)
there isn't really a primary performer at all. My solution would be to
improve the interface so that the track artist wasn't the only thing
shown.
As far as I can see, changing the guidelines for the track artist may or
may not be worth doing for other reasons, but it won't fix the problem
you're showing here.
I suggest that one of three things happen:
1. The artist field should contain the primary artist(s) - in your violin
sonata case; the violinist, the conductor and the orchestra, in the next
example; King's Singers and The Consort of Musicke, or in the original
case; Hans Hotter and Hans Dokoupil.
2. As you suggest, all performance ARs should be displayed.
3. The work listing show the recording's Artist field, as well as the
release's Artist field. One of the two fields, lets say the recording's
artist field, should contain the composer, while the other should contain
at least one primary artist.

Modifying the web interface to list all performance ARs could become pretty
overwhelming (here's a moderate example with 7 artists) - and I don't think
coming up with heuristics to determine which performance ARs to display is
worth the development time. Well, not yet anyway.
http://musicbrainz.org/recording/df2af9f5-699c-4b3f-bb5a-3b4626a02579

I don't love having each artist field mean something different, either. I
just see it as being a half-way point.
Post by Rupert Swarbrick
Post by David Hilton
Anything that could go in the artist field should also be listed under a
more advanced relationship, so it's not a matter of data loss, but rather
which values make the most sense. I /am/ approaching this from the work
perspective. I'm having trouble coming up with any reason to have the
field contain a composer. What am I missing?
Well, just the historical one. People like their jukebox applications to
have performances of Beethoven's 9th under "Beethoven". And these
jukebox applications can only deal with having one artist on a file (and
don't respect the composer tag, even if there is one).
Yeah, I'd kind of forgotten about some applications not having advanced
tagging support. Picard could have a flag saying: if composer is one of
these values, use composer as artist.

I'm all for improving the data, particularly when the old behavior can be
retained through 1 minor software change.


David
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111116/97ef21a8/attachment.htm
SwissChris
2011-11-17 00:59:31 UTC
Permalink
Post by David Hilton
I worked today on 'Hans Hotter singt Lieder', a compilation album (
http://musicbrainz.org/release/205cfe1a-0f49-48dd-9ada-ecd816cb38ed).
This is my second foray into adding a release in ngs; the first time was
pretty involved. This time took a good while; about 6 hours of, I think (I
like to get all of the metadata I can).
Does anyone see significant issues with the data as it has been entered?
1. the new csg - work names, recording names, and track names are all
distinct values. I'm presuming that work names follow the old csg, while
track names follow the cd? I followed this, but then started to change to
the csg (tracks 2 and 4) before second-guessing myself. I have no idea
which value recordings should have.
While we seem to agree that there is a release called "Hans Hotter singt
Lieder" (or "Greatest Arias by Maria Callas") which should be credited to
the Performing artist, I'm strongly against crediting tracks to performing
artists for the time being: There is no such thing as e.g. "Wandrers
Nachtlied, Op. 96 No. 3, D
768<http://musicbrainz.org/recording/7d92e167-fd42-4ca9-872b-14c34ddae30d>
" by Hans Hotter: this is nonsensical.

2. artist tags: on tracks, I believe these now point to the primary
Post by David Hilton
performing artist, instead of the composer.
No. The rules have not (yet) changed: Artist Credit at track level still is
the composer.

3. part-of work relationships: I would think that parent relationships not
Post by David Hilton
found at a child level are inherited (see track 14 as an example). I added
Robert Franz' Op. 10, putting Robert Franz as the composer of the top-level
work, created sub-works for individual pieces and put the lyricist of each
piece on the works they apply to. Picard doesn't seem to interpret it this
way, though, so perhaps I should have done it differently.
The inheritance guidelines are still at proposal status, and even if this
proposal passes (as it should), it's still a long way till it's implemented
at UI level and on picard. So we still should apply CSG at track level -
with artist credit to the composer

Swisschris/chabreyflint
Post by David Hilton
David
_______________________________________________
MusicBrainz-users mailing list
MusicBrainz-users at lists.musicbrainz.org
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111117/0fc4b502/attachment.htm
Nicolás Tamargo de Eguren
2011-11-17 01:06:05 UTC
Permalink
On Wed, Nov 16, 2011 at 4:08 AM, David Hilton <quercus.aeternam at gmail.com>
Post by David Hilton
I worked today on 'Hans Hotter singt Lieder', a compilation album
(http://musicbrainz.org/release/205cfe1a-0f49-48dd-9ada-ecd816cb38ed).
This is my second foray into adding a release in ngs; the first time was
pretty involved.? This time took a good while; about 6 hours of, I think (I
like to get all of the metadata I can).
Does anyone see significant issues with the data as it has been entered?
1. the new csg - work names, recording names, and track names are all
distinct values.? I'm presuming that work names follow the old csg, while
track names follow the cd?? I followed this, but then started to change to
the csg (tracks 2 and 4) before second-guessing myself.? I have no idea
which value recordings should have.
While we seem to agree that there is a release called "Hans Hotter singt
Lieder" ?(or "Greatest Arias by Maria Callas") which should be credited to
the Performing artist, I'm strongly against crediting tracks to performing
artists for the time being: There is no such thing as e.g. ?"Wandrers
Nachtlied, Op. 96 No. 3, D 768"?by Hans Hotter: this is nonsensical.
How is that nonsensical, but having a rock cover credited to the band
that covered it isn't?
Post by David Hilton
2. artist tags: on tracks, I believe these now point to the primary
performing artist, instead of the composer.
No. The rules have not (yet) changed: Artist Credit at track level still is
the composer.
Post by David Hilton
3. part-of work relationships: I would think that parent relationships not
found at a child level are inherited (see track 14 as an example).? I added
Robert Franz' Op. 10, putting Robert Franz as the composer of the top-level
work, created sub-works for individual pieces and put the lyricist of each
piece on the works they apply to.? Picard doesn't seem to interpret it this
way, though, so perhaps I should have done it differently.
The inheritance guidelines are still at proposal status, and even if this
proposal passes (as it should), it's still a long way till it's implemented
at UI level and on picard. So we still should apply CSG at track level -
with artist credit to the composer
Swisschris/chabreyflint
Post by David Hilton
David
_______________________________________________
MusicBrainz-users mailing list
MusicBrainz-users at lists.musicbrainz.org
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-users
_______________________________________________
MusicBrainz-users mailing list
MusicBrainz-users at lists.musicbrainz.org
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-users
--
Nicol?s Tamargo de Eguren
David Hilton
2011-11-17 01:08:37 UTC
Permalink
2011/11/16 Nicol?s Tamargo de Eguren <reosarevok at gmail.com>
On Wed, Nov 16, 2011 at 4:08 AM, David Hilton <
quercus.aeternam at gmail.com>
Post by David Hilton
I worked today on 'Hans Hotter singt Lieder', a compilation album
(http://musicbrainz.org/release/205cfe1a-0f49-48dd-9ada-ecd816cb38ed).
This is my second foray into adding a release in ngs; the first time was
pretty involved. This time took a good while; about 6 hours of, I
think (I
Post by David Hilton
like to get all of the metadata I can).
Does anyone see significant issues with the data as it has been entered?
1. the new csg - work names, recording names, and track names are all
distinct values. I'm presuming that work names follow the old csg,
while
Post by David Hilton
track names follow the cd? I followed this, but then started to change
to
Post by David Hilton
the csg (tracks 2 and 4) before second-guessing myself. I have no idea
which value recordings should have.
While we seem to agree that there is a release called "Hans Hotter singt
Lieder" (or "Greatest Arias by Maria Callas") which should be credited
to
the Performing artist, I'm strongly against crediting tracks to
performing
artists for the time being: There is no such thing as e.g. "Wandrers
Nachtlied, Op. 96 No. 3, D 768" by Hans Hotter: this is nonsensical.
How is that nonsensical, but having a rock cover credited to the band
that covered it isn't?
This.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111116/fc39be92/attachment-0001.htm
SwissChris
2011-11-17 02:24:48 UTC
Permalink
2011/11/17 Nicol?s Tamargo de Eguren <reosarevok at gmail.com>
On Wed, Nov 16, 2011 at 4:08 AM, David Hilton <
quercus.aeternam at gmail.com>
Post by David Hilton
I worked today on 'Hans Hotter singt Lieder', a compilation album
(http://musicbrainz.org/release/205cfe1a-0f49-48dd-9ada-ecd816cb38ed).
This is my second foray into adding a release in ngs; the first time was
pretty involved. This time took a good while; about 6 hours of, I
think (I
Post by David Hilton
like to get all of the metadata I can).
Does anyone see significant issues with the data as it has been entered?
1. the new csg - work names, recording names, and track names are all
distinct values. I'm presuming that work names follow the old csg,
while
Post by David Hilton
track names follow the cd? I followed this, but then started to change
to
Post by David Hilton
the csg (tracks 2 and 4) before second-guessing myself. I have no idea
which value recordings should have.
While we seem to agree that there is a release called "Hans Hotter singt
Lieder" (or "Greatest Arias by Maria Callas") which should be credited
to
the Performing artist, I'm strongly against crediting tracks to
performing
artists for the time being: There is no such thing as e.g. "Wandrers
Nachtlied, Op. 96 No. 3, D 768" by Hans Hotter: this is nonsensical.
How is that nonsensical, but having a rock cover credited to the band
that covered it isn't?
Because in rock we don't use work-related info ("Op. 96", "D 768") on track
titles ;-)
Post by David Hilton
2. artist tags: on tracks, I believe these now point to the primary
performing artist, instead of the composer.
No. The rules have not (yet) changed: Artist Credit at track level still
is
the composer.
Post by David Hilton
3. part-of work relationships: I would think that parent relationships
not
Post by David Hilton
found at a child level are inherited (see track 14 as an example). I
added
Post by David Hilton
Robert Franz' Op. 10, putting Robert Franz as the composer of the
top-level
Post by David Hilton
work, created sub-works for individual pieces and put the lyricist of
each
Post by David Hilton
piece on the works they apply to. Picard doesn't seem to interpret it
this
Post by David Hilton
way, though, so perhaps I should have done it differently.
The inheritance guidelines are still at proposal status, and even if this
proposal passes (as it should), it's still a long way till it's
implemented
at UI level and on picard. So we still should apply CSG at track level -
with artist credit to the composer
Swisschris/chabreyflint
Post by David Hilton
David
_______________________________________________
MusicBrainz-users mailing list
MusicBrainz-users at lists.musicbrainz.org
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-users
_______________________________________________
MusicBrainz-users mailing list
MusicBrainz-users at lists.musicbrainz.org
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-users
--
Nicol?s Tamargo de Eguren
_______________________________________________
MusicBrainz-users mailing list
MusicBrainz-users at lists.musicbrainz.org
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111117/1dea3bfc/attachment.htm
David Hilton
2011-11-17 02:35:17 UTC
Permalink
Post by SwissChris
2011/11/17 Nicol?s Tamargo de Eguren <reosarevok at gmail.com>
Post by Nicolás Tamargo de Eguren
On Wed, Nov 16, 2011 at 4:08 AM, David Hilton <
quercus.aeternam at gmail.com>
Post by SwissChris
Post by Nicolás Tamargo de Eguren
Post by David Hilton
I worked today on 'Hans Hotter singt Lieder', a compilation album
(http://musicbrainz.org/release/205cfe1a-0f49-48dd-9ada-ecd816cb38ed).
This is my second foray into adding a release in ngs; the first time was
pretty involved. This time took a good while; about 6 hours of, I think (I
like to get all of the metadata I can).
Does anyone see significant issues with the data as it has been entered?
1. the new csg - work names, recording names, and track names are all
distinct values. I'm presuming that work names follow the old csg, while
track names follow the cd? I followed this, but then started to change to
the csg (tracks 2 and 4) before second-guessing myself. I have no idea
which value recordings should have.
While we seem to agree that there is a release called "Hans Hotter singt
Lieder" (or "Greatest Arias by Maria Callas") which should be credited to
the Performing artist, I'm strongly against crediting tracks to performing
artists for the time being: There is no such thing as e.g. "Wandrers
Nachtlied, Op. 96 No. 3, D 768" by Hans Hotter: this is nonsensical.
How is that nonsensical, but having a rock cover credited to the band
that covered it isn't?
Because in rock we don't use work-related info ("Op. 96", "D 768") on
track titles ;-)

But that's the point. Now that we have works, the track titles can be what
the artist used on their release without any loss of information - and
anyone that wants the old behavior could set a work-title override in the
tagging software.
Post by SwissChris
Post by Nicolás Tamargo de Eguren
Post by David Hilton
2. artist tags: on tracks, I believe these now point to the primary
performing artist, instead of the composer.
No. The rules have not (yet) changed: Artist Credit at track level still is
the composer.
Post by David Hilton
3. part-of work relationships: I would think that parent
relationships not
Post by SwissChris
Post by Nicolás Tamargo de Eguren
Post by David Hilton
found at a child level are inherited (see track 14 as an example). I added
Robert Franz' Op. 10, putting Robert Franz as the composer of the top-level
work, created sub-works for individual pieces and put the lyricist of each
piece on the works they apply to. Picard doesn't seem to interpret it this
way, though, so perhaps I should have done it differently.
The inheritance guidelines are still at proposal status, and even if this
proposal passes (as it should), it's still a long way till it's implemented
at UI level and on picard. So we still should apply CSG at track level -
with artist credit to the composer
Swisschris/chabreyflint
Post by David Hilton
David
_______________________________________________
MusicBrainz-users mailing list
MusicBrainz-users at lists.musicbrainz.org
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-users
_______________________________________________
MusicBrainz-users mailing list
MusicBrainz-users at lists.musicbrainz.org
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-users
--
Nicol?s Tamargo de Eguren
_______________________________________________
MusicBrainz-users mailing list
MusicBrainz-users at lists.musicbrainz.org
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-users
_______________________________________________
MusicBrainz-users mailing list
MusicBrainz-users at lists.musicbrainz.org
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111116/7153b9f9/attachment.htm
Andrew Conkling
2011-11-17 15:54:41 UTC
Permalink
Post by David Hilton
But that's the point. Now that we have works, the track titles can be
what the artist used on their release without any loss of information - and
anyone that wants the old behavior could set a work-title override in the
tagging software.
This should be discussed on the mb-style list in the interest of discussing
some post-NGS changes to the guidelines; it's not a good idea to try to
push forward simply by editing how you think it should be done. :)

You mentioned initially that you thought the track artists were supposed to
point to the performer now; did I miss a current doc that states or implies
that?

Andrewski
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111117/ce24dd47/attachment.htm
Rupert Swarbrick
2011-11-21 14:22:28 UTC
Permalink
Post by Andrew Conkling
Post by David Hilton
But that's the point. Now that we have works, the track titles can be
what the artist used on their release without any loss of information - and
anyone that wants the old behavior could set a work-title override in the
tagging software.
This should be discussed on the mb-style list in the interest of discussing
some post-NGS changes to the guidelines; it's not a good idea to try to
push forward simply by editing how you think it should be done. :)
... except that sometimes you see comments there like "Hmm, we need to
be descriptive rather than proscriptive. We should wait and see what
editors are doing before formulating a style guide that might contradict
them" (This was in the context of what an MB work meant, if I remember
rightly). Frustrating.

Rupert
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 315 bytes
Desc: not available
Url : http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111121/1a4f2b8d/attachment.pgp
Nicolás Tamargo de Eguren
2011-11-21 14:55:38 UTC
Permalink
Post by Rupert Swarbrick
Post by Andrew Conkling
But that's the point. ?Now that we have works, the track titles can be
what the artist used on their release without any loss of information - and
anyone that wants the old behavior could set a work-title override in the
tagging software.
This should be discussed on the mb-style list in the interest of discussing
some post-NGS changes to the guidelines; it's not a good idea to try to
push forward simply by editing how you think it should be done. :)
... except that sometimes you see comments there like "Hmm, we need to
be descriptive rather than proscriptive. We should wait and see what
editors are doing before formulating a style guide that might contradict
them" (This was in the context of what an MB work meant, if I remember
rightly). Frustrating.
I can see the benefits of both approaches, but there's a difference
there: we don't have a current way of defining MB works that can
change out of this use, because we don't really have any definition
for them - but we have a current, accepted way of dealing with
classical music. While I do want to change it, I think the difference
is reasonably important for deciding how to deal with it.
Post by Rupert Swarbrick
Rupert
_______________________________________________
MusicBrainz-users mailing list
MusicBrainz-users at lists.musicbrainz.org
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-users
--
Nicol?s Tamargo de Eguren
David Hilton
2011-11-18 17:26:15 UTC
Permalink
On Wed, Nov 16, 2011 at 21:35, David Hilton <quercus.aeternam at gmail.com>
Post by David Hilton
But that's the point. Now that we have works, the track titles can be
what the artist used on their release without any loss of information - and
anyone that wants the old behavior could set a work-title override in the
tagging software.
This should be discussed on the mb-style list in the interest of
discussing some post-NGS changes to the guidelines; it's not a good idea to
try to push forward simply by editing how you think it should be done. :)
You mentioned initially that you thought the track artists were supposed
to point to the performer now; did I miss a current doc that states or
implies that?

I wrote this precisely because I found myself in a quandary; I've not found
docs specifically supporting a position.

I have stopped modifying track/recording/work names, and have explained my
thoughts on the matter. That said, without specific information
contradicting my position, any /new/ entries I add will follow what makes
sense to me.

Quite fortunately, if we decide that track listings should contain work
names following the old CSG, I will change the changes I have to the
standard. This will be much easier than the reverse (looking up each album
and re-entering its track list).

David
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111118/35becdaf/attachment.htm
SwissChris
2011-11-18 18:47:24 UTC
Permalink
Post by David Hilton
On Wed, Nov 16, 2011 at 21:35, David Hilton <quercus.aeternam at gmail.com>
Post by David Hilton
But that's the point. Now that we have works, the track titles can be
what the artist used on their release without any loss of information - and
anyone that wants the old behavior could set a work-title override in the
tagging software.
This should be discussed on the mb-style list in the interest of
discussing some post-NGS changes to the guidelines; it's not a good idea to
try to push forward simply by editing how you think it should be done. :)
+1
Post by David Hilton
You mentioned initially that you thought the track artists were supposed
to point to the performer now; did I miss a current doc that states or
implies that?
I wrote this precisely because I found myself in a quandary; I've not
found docs specifically supporting a position.
As long as we haven't decided on something else through the
RFC/RFV-process, CSG should apply, which clearly doesn't support performers
as track artists:

http://musicbrainz.org/doc/Classical_Style_Guide

(quote: "Exception: Mixed recitals by a performer or group

In cases where a release features a single performer or group and contains
works from multiple composers, that performer or group may be designated
the ReleaseArtist <http://musicbrainz.org/doc/Release_Artist>, with each
TrackArtist <http://musicbrainz.org/doc/Track_Artist> assigned to the
appropriate composer. See
ClassicalReleaseArtistStyle<http://musicbrainz.org/doc/Classical_Release_Artist_Style>
for
full details." )
Post by David Hilton
I have stopped modifying track/recording/work names, and have explained my
thoughts on the matter. That said, without specific information
contradicting my position, any /new/ entries I add will follow what makes
sense to me.
Quite fortunately, if we decide that track listings should contain work
names following the old CSG, I will change the changes I have to the
standard.
Please do so :-)
Post by David Hilton
This will be much easier than the reverse (looking up each album and
re-entering its track list).
David
_______________________________________________
MusicBrainz-users mailing list
MusicBrainz-users at lists.musicbrainz.org
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111118/ecedc139/attachment.htm
David Hilton
2011-11-18 18:55:27 UTC
Permalink
Post by David Hilton
On Wed, Nov 16, 2011 at 21:35, David Hilton <quercus.aeternam at gmail.com>
Post by David Hilton
But that's the point. Now that we have works, the track titles can be
what the artist used on their release without any loss of information - and
anyone that wants the old behavior could set a work-title override in the
tagging software.
Post by David Hilton
This should be discussed on the mb-style list in the interest of
discussing some post-NGS changes to the guidelines; it's not a good idea to
try to push forward simply by editing how you think it should be done. :)
Post by David Hilton
You mentioned initially that you thought the track artists were
supposed to point to the performer now; did I miss a current doc that
states or implies that?
Post by David Hilton
I wrote this precisely because I found myself in a quandary; I've not
found docs specifically supporting a position.
Post by David Hilton
I have stopped modifying track/recording/work names, and have explained
my thoughts on the matter. That said, without specific information
contradicting my position, any /new/ entries I add will follow what makes
sense to me.
Post by David Hilton
Quite fortunately, if we decide that track listings should contain work
names following the old CSG, I will change the changes I have to the
standard. This will be much easier than the reverse (looking up each album
and re-entering its track list).
Post by David Hilton
David
Please vote:
http://musicbrainz.org/edit/15569728
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.musicbrainz.org/pipermail/musicbrainz-users/attachments/20111118/83de961d/attachment-0001.htm
Loading...