if a dataset has multiple edges between the same source and target node, the user should provide an ID for the edge (or else)...

[hdngr]

or else things won't work in edge captioning and filtering...

Realistically, Alchemy.js as a client application should not be burdened with teasing out unique edges. If someone was smart enough to get a data set with rich, diverse relationships between the same source and target node, they should be smart enough to provide Alchemy.js with a unique id for the edge.

Therefore, all edge ids will be defined with the logic:

if edge.id
    edge.id = edge.id
else
    edge.id = "#{edge.source.id}-#{edge.target.id}"

sound good?

cc: @mdcox @ifjorissen

[ghost]

Sounds good to me, we just need to make it excessively clear in the documentation.

[ghost]

Another idea we were talking about was having the ID be source-relationship-target, since it's unlikely (except with temporal or chemical data I suppose) that there would be duplicates.