i was discussing this in one of the classical threads--i repaste my suggestion here:

 ack--let me try again here--i'll ignore whether these releations are relfexive--either we'll treat them all as reflexive or none of them as reflexive.

in the database, in addition to having RECORDINGS and WORKS we need "WORK VERSIONS"

where a work is a set of work versions (such as the set of all different versions of greensleeves) which can have properties of its own, and the versions inherit properties from the work unless overidden.

for simplicity's sake, a user can simply say that recording y is a recording of work x, rather than saying y is a recording of version v of work x, although the dedicated will be allowed to do the latter.

i'm picturing WORKS as large circles, and when precision is not required all relations and work can be done between works, ignoring work versions.

then i'm picturing WORK VERSIONS as dots within those circles.

revision and version relationships are all realtionships inside one circle.  includes material from and based on relationships are relationships accross circles.

x "is a recording of" y   is neither transitive nor symmetric

x "is a part of" y is transitive, but not symmetric
x "is a revision of" y  is transitive but not symmetric

x "is a version of" y is symmetric, and transitive

x "is (merely) based on" y

x "(merely) includes material from" y  is symmetric, but not transitive 

the thing i can't decide is if "is a revision of" shoudl be a version/work relation or a version/version relation.

if i revise a symphony three times, is the third revision a revision only of the 2nd revision? or of all three prior versions? is it a revision of the more abstract work?  or are different situations possible, signaling different things about the revision?

Musicbrainz-style mailing list

Reply via email to