Modelling questions and details general

[reinierdevalk]

What?

This issue lists various questions and details regarding the modelling of any kind of tablature tablature that are still to be addressed.

Questions and details

• What is our policy on xml:ids? Empty IDs (xml:id='') are not valid XML (see here: an xml:id must have a value, an NCName, and be document unique).
  • We place them everywhere by default, adding an ID everywhere.
  • We only place them when the thing it identifies is acually referred to, and leave them out completely otherwise.
  • What will an ID look like? Some random character string, or something with more meaning, which makes things easier to interpret for people going through the encoding (e.g., I have used xml:id='m1.n.3' for the third note in the first bar).
• When modelling, I find it difficult to decide when to use an element ("if it feels like data" is an argument I've found) or when to use an attribute. In general, we should follow the practice from the MEI Guidelines as much as possible.
  • RdV: It seems to make sense to use
    • An element if the technique is something that may affect multiple <tabGrp>s in a row, e.g., let ring (cf. <beam>).
    • An element if the technique is something that affects all notes inside a single <tabGrp>, e.g., palm muting.
    • An attribute if the technique is something that affects only a single note.
  • DL: It would be good for MEI to have some guidelines for expansion writers for this. I suspect that Perry, Johannes and Laurent have other criteria. I'd add
    • Use an element if information is multidimensional.
    • Use a generic element with types if the alternative is a million new attributes.
    • Prefer subclassing something that already exists, so if pull-off looks like a slur, maybe subclass <slur>.