[relax-ng] Re: DOM Abstract Schemas revisited

[Kohsuke KAWAGUCHI <kohsuke.kawaguchi@sun.com>]

> The issue is not whether AS maps RNG concepts directly,
> but whether RNGs can be transformed into ASModels without loss of
> expressiveness.

I think this is where I misunderstood you. I thought you were trying to
map RNG as natural as possible.


> I think they can be, except for the points I mentioned
> previously, of which wildcard naming is the most important.

The element/attribute mixed content model is still a very big problem.
You suggest to split it into multiple content models, but sometimes it
takes exponential time to the size of input. Consider:

<element name="foo">
  <choice>
    <element name="a"><empty/></element>
    <attribute name="a"/>
  </choice>
  <choice>
    <element name="b"><empty/></element>
    <attribute name="b"/>
  </choice>
  ...
  <choice>
    <element name="z"><empty/></element>
    <attribute name="z"/>
  </choice>
</element>




Another problem is the semantics of <interleave>. Firstly, this is not
SGML-&, so the "all" compositor cannot be used for <interleave>.
Secondly, the "all" compositor is limited to the top-level, whereas you
can write <interleave> anywhere. And expanding interleave can again take
exponential time to the size of input. Even worse, this exponential
explosion happens to every-day schemas.



regards,
--
Kohsuke KAWAGUCHI                          +1 607 257 0037
Sun Microsystems                   kohsuke.kawaguchi@sun.com