Re: UPA example

On 25 Jun 2008, at 01:59 , Boris Kolpackov wrote:
> ...
> Radu Preotiuc-Pietro <radup@bea.com> writes:
>
>> Maybe it's also interesting to mention that, while in XML Schema  
>> 1.1 the
>> Schema becomes valid, the instance
>>
>> <apple/>
>> <apple/>
>> <apple/>
>>
>> won't be valid still (based on my understanding, the third <apple/>
>> element will map to the element particle and the validator will  
>> still be
>> waiting for an element to map to the wildcard).
>
> Yes, I also think it won't be valid unless the parser in 1.1 is  
> expected
> to do backtracking. I am also wondering if the authors of this  
> change to
> the spec considered how hard it will be to explain something like this
> to a user.

Speaking for myself, yes, but in this case I am not entirely certain
what aspects of the example are important for 'something like this'.

I think explaining things to users would be somewhat simpler if
we lost the Unique Particle Attribution rule entirely, but I have
not succeeded in persuading the rest of the working group.

As regards this particular example, though, I may be missing
something; it doesn't seem to me hard to explain that when you
have a content model which

   1 allows an infinite number of repetitions of subsequences
   2 starts a new subsequence each time an 'apple' element would
     be the third element, or later, in a subsequence
   3 requires each subsequence to have two or more elements

then a sequence of three 'apple' elements is not valid.

If the user is surprised by the fact that some apple elements match
the wildcard, not the element declaration, then perhaps the wildcard
is overbroad and should exclude 'apple'?  If the user is surprised
by the fact that the wildcard does not capture all 'apple' elements
after the first, then perhaps the outer sequence ought not to be
repeatable?

--Michael Sperberg-McQueen
   World Wide Web Consortium / MIT CSAIL