[tei-council] Fwd: the new <constraint> element

[Laurent Romary]

Hi all,
A message from Syd.
Laurent

Début du message réexpédié :

> De : Syd Bauman <Syd_Bauman at Brown.edu>
> Date : 9 avril 2009 05:57:06 GMT+02:00
> À : Sebastian Rahtz <sebastian.rahtz at oucs.ox.ac.uk>, Laurent Romary <laurent.romary at loria.fr 
> >
> Objet : Rép : the new <constraint> element
> Répondre à : Syd_Bauman at Brown.edu
>
> [LR -- please re-post to the Council; thank you.]
>
> Whoo-hoo!
>
> I am thrilled to see this.
>
> Overall, it looks great. This will be a powerful step forward. A few
> thoughts jump to mind:
>
> * Why limit values of scheme= so severely? While I don't actually
>  know anyone who uses DSD, I don't see any reason to exclude it.
>  Would be nice if someone could develop a list of all the languages
>  that would make sense to use inside <constraint>.
>
> * Why isn't <constraintList> permitted inside <macroDef>? There may
>  well be a reason that hasn't jumped to my mind -- I haven't thought
>  this through very carefully -- but at first blush it would seem
>  useful.
>
> * Why is it "constraintList" rather than "listConstraint", which
>  would be analogous to <listPerson>, <listBibl>, etc.?
>
> * On the organization of constraints:
>  As currently envisioned, constraints are expressed grouped by
>  language; my instinct is that it would be better to organize these
>  by function. I.e., if I have two constraints I wish to express, A
>  and B, the current scheme would have me use:
>    <constraintList scheme="schematron">
>      <constraint> schematron 1.6 for A </constraint>
>      <constraint> schematron 1.6 for B </constraint>
>    </constraintList>
>    <constraintList scheme="isoschematron">
>      <constraint> ISO schematron for A </constraint>
>      <constraint> ISO schematron for B </constraint>
>    </constraintList>
>    <constraintList scheme="private">
>      <constraint> CLIX for B </constraint>
>      <constraint> CLIX for A </constraint>
>    </constraintList>
>  But I think it might be a lot more logical to use:
>    <constraintList>
>      <constraint scheme="schematron"> schematron 1.6 for A </ 
> constraint>
>      <constraint scheme="isoschematron"> ISO schematron for A </ 
> constraint>
>      <constraint scheme="private"> CLIX for A </constraint>
>    </constraintList>
>    <constraintList scheme="isoschematron">
>      <constraint scheme="schematron"> schematron 1.6 for B </ 
> constraint>
>      <constraint scheme="isoschematron"> ISO schematron for B </ 
> constraint>
>      <constraint scheme="private"> CLIX for B </constraint>
>    </constraintList>
>
> * What are the semantics of the att.identified attributes? I.e., what
>  does the ident= *mean*? (Does it have to be unique; do two
>  <constraint>s with the same ident= have to be the same? In which
>  case could I use copyOf= instead? I presume mode= means the usual
>  "add this", "delete this", or "modify this" kinda stuff. Does the
>  module have to be declared? What does it mean to specify module="A"
>  if the <constraint> is in module B?
>
> * What is the difference between <content> and <constraint
>  scheme="RELAXNG">?
>
> Well, I have to get some sleep now. But again, this is exciting
> stuff.
>
> P.S. My P5 build process on my Mac is now completely broken. Will try
>     to look at why over the weekend.
>