Menu

#323 Allow modification of attributes belonging to a class

GREEN
closed-fixed
5
2012-09-15
2011-09-30
No

For constructs such as follows, we should:
1) accept this as perfectly valid for ODD (meaning we document this behaviour in the guidelines)
2) implemented this in Roma
@Seb you alluded to the possibility to make a first pass to check out class memberships before processing other elementSpec fields. Is this complicated?

<elementSpec ident="XXX" module="YYY" mode="add">
<classes>
<memberOf key="att.typed"/>
</classes>
<attList>
<attDef ident="type" mode="change">
<valList>
<valItem ident="xx"/>
<valItem ident="yy"/>
...
</valList>
</attDef>
</attList>
</elementSpec>

Discussion

  • Laurent Romary

    Laurent Romary - 2011-09-30
    • summary: Allow on the fly modification of attributes belonging to a c --> Allow modification of attributes belonging to a class
     
  • Sebastian Rahtz

    Sebastian Rahtz - 2011-10-02

    I agree, this looks like perfectly reasonable syntax, which the odd processing tools should follow. It will not be a trivial change, so may take a while before I have time. Anyone else welcome to try, of course.

     
  • Lou Burnard

    Lou Burnard - 2011-11-01
    • milestone: --> 871209
     
  • James Cummings

    James Cummings - 2012-04-15

    This is the local modification of attribute classes that I've often suggested should be added to a new version of ODD. I would like this feature request implemented because it would simplify the major flaws with the way the TEI class system works at the moment. e.g. if you look at all the uses of @type which are *not* part of att.typed, this is because they have a specific <desc> or suggested <valList>. If you were able to say "This element claims membership in attribute class A, but then has modified that (in a subsetting manner)" then we could purify the TEI's use of the class structure to the way it was intended.

    Of course, we would have to state that if you said you were a member of an attribute class and then modified that attribute so that it was not a pure subset of the original form, that this would be Non-Conformant. (e.g. if you took @when from att.datable and modified it to say allow URIs).

    -james

     
  • Sebastian Rahtz

    Sebastian Rahtz - 2012-04-17

    The Council meeting 2012-04-16 discussed this, and agreed to implement it as soon as possible; with the caveat that it will not be easy, and may take some time. It was noted that there would be an ongoing action to sweep the Guidelines for occasions when this new facility would make an existing element more elegantly expressed. Especially we will look for elements which locally define an attribute instead of using an attribute class.

     
  • Sebastian Rahtz

    Sebastian Rahtz - 2012-04-17
    • status: open --> open-accepted
     
  • James Cummings

    James Cummings - 2012-07-23
    • milestone: 871209 --> GREEN
     
  • James Cummings

    James Cummings - 2012-07-23

    Updating to group 'Green'.

     
  • Sebastian Rahtz

    Sebastian Rahtz - 2012-09-15
    • status: open-accepted --> closed-fixed
     
  • Sebastian Rahtz

    Sebastian Rahtz - 2012-09-15

    I implemented this during DH2012, and have tested it by redefining @type on <abbr> and <title>.