Print

Print


On 11 Dec 2012, at 10:27, Lou Burnard <[log in to unmask]>
 wrote:

>> the first is just confusing and contradictory in the usual manner of the Guidelines. But the second
>> seems to be simply impossible to understand. IF <quotation> is present, with no @marks,
>> the value 'marks=all' is assumed - but we have no way to express that in RELAX NG schemas,
> 
> What does a schema have to do with it? The purpose of the <quotation> element is not to help a validater -- by the time you get to processing this element it's too late! Its purpose is primarily to document how the (valid) markup should be interpreted, i.e. whether or not the absence of quote marks in the encoding implies that they were also missing from the source. What you do with that information is up to your processor: both situations are valid.

yes but no but yes but no.  I suspect that when you designed that <defaultVal> element back in the day, you had in mind
the facility of SGML to carry a default through to the DTD, and for the DTD processor to insert that into the
stream of markup presented to the processor.  RELAX NG broke this connection between
validation and processing, which muddies the waters. the presence of defaultVal means that
processing _with_ a DTD gives different results from without - not good.

> 
>> so how can I implement it? And what if <quotation> is NOT present? what is the default then?
> 
> If the <quotation> element is missing then the encoder has chosen, for whatever reason, not to tell you about their practice in this respect. So you're at liberty to decide for yourself what to do about it.
> 
but if she had said

  <quotation>
     <p>quotation marks have not been preserved</p>
  </quotation>

I'd be in a Pickle :-}

> 
>> 
>> I suspect that default on <quotation>/@marks should be removed, as it
>> is implying interpretations that it cannot fulfil.
> 
> I don't follow your logic here, though I agree that in general default values on optional elements are a bit of a pain.

its ok when the effect is localized. but in this case, where the effect of this attribute is across the whole document,
I suggest its not defensible.

--
Sebastian Rahtz      
Director (Research Support) of Academic IT Services 
University of Oxford IT Services
13 Banbury Road, Oxford OX2 6NN. Phone +44 1865 283431