From: David Alan Black <dblack@...>
Date: 2002-02-24T06:36:07+09:00
Subject: Re: rdoc/xml questions

Hi --

On Sat, 23 Feb 2002, Dave Thomas wrote:

> james@rubyxml.com writes:
>
> > >          <class name="A::B">
>
> >
> > I'm unhappy with name="A::B", as it hides the relationship of A to B
> > inside the attribute value, using non-XML notation. Perhaps
> >           <class name="B" container="A">
> >
> > or something along those lines
>
> I thought about this a while before originally posting. The problem is
> that the relationships can get pretty complex pretty quickly (as there
> is arbitrary nesting, and classes can be inner classes in singletons
> and so on. So I wouldn't want to capture the interrelationships using
> the attributes at all--I'd view the attribute here as simply a
> label. That's why I was wondering if the nested representation might
> be better. I was also showing a somewhat simplified set of
> attributes. In reality a class or module would likely have its name,
> its fully qualified name, and the name of a parent class as
> attributes.

Would you consider something that mimicked the Ruby syntax more, such
as:

  <class name="A">
    <description>
    comment 1
    </description>
    <class name="B">
      <description>
      comment 2
      </description>
    </class>
  </class>

and then use processing to establish the relationships?  Or would that
mean backpedaling (from the amount of into you could put into
attributes) and displacing too much work onto XPath (or whatever)?
(Speaking of backpedaling: I know I'm not thinking through all the
complexities (singleton methods, etc.), so this may be a
non-question.)


David (who is on an XSL[T] kick :-)

-- 
David Alan Black
home: dblack@candle.superlink.net
work: blackdav@shu.edu
Web:  http://pirate.shu.edu/~blackdav