[tei-council] TEI model classes future directions

[Sebastian Rahtz]

James Cummings wrote:
> Sebastian Rahtz wrote:
>   
>> So what do the rest of you feel? do you
>>
>>   A.  not understand the technical argument
>>   B.  not care, it's nerdy detail the editors should sort out
>>   C.  feel violently that this is bad technological change for change's sake
>>   D.  just _love_ the idea
>>   E.  think that the immediate problems it may solve can
>>         be better dealt with another way, so we don't need
>>         to decide.
>>     
>
> F. I care, and believe it to be a good idea, but would like to see more
> explanation of how it works.
>   
thats A, then :-}

OK:

model.foo, with members a, b and c,
has a <classSpec> which says
"generate='alternate sequence sequenceOptional'".
On processing that generates three patterns/entities:

  model.foo  = a | b | c
  model.foo.sequence = a, b, c
  model.foo.sequenceOptional = a?, b?, c?

in the content model for any given element you can choose which of these
patterns to refer to.

There are two downsides:

 a) in RELAXNG, it is idiomatic to say
         model.foo |= bar
    to extend model.foo with the bar element as an alternate.
    Obviously if you say model.foo.sequence | = bar you end up
    with
        a, b, c | bar
   which is not what you meant :-}

   Answer: our class models are not RELAXNG patterns which are extensible

 b) the order of a, b, and c is determined by the order
   in which they are declared in the source ODD
   (ie the source of P5)


-- 
Sebastian Rahtz      

Information Manager, Oxford University Computing Services
13 Banbury Road, Oxford OX2 6NN. Phone +44 1865 283431

OSS Watch: JISC Open Source Advisory Service
http://www.oss-watch.ac.uk