[sc34wg3] CXTM: -reifier- and -reified-

Kal Ahmed sc34wg3@isotopicmaps.org
Thu, 20 Jan 2005 09:01:52 +0000

On 19 Jan 2005, at 16:16, Lars Marius Garshol wrote:

> (Written by, and posted on behalf of, Eirik Opland (from Ontopia), who
> is subscribed to the list, but apparently not allowed to post.)
> I am implementing the latest specification of CXTM, and have a couple 
> of
> questions.
> The RELAX-NG Compact Syntax Schema contains the following rules:
> attlist.reifier &= attribute reifier { xsd:integer }?
> attlist.reified &= attribute reified { text }?
> I assume that 'xsd:integer' is meant to be the sorted index of the
> reifying topic as defined in 5.3 of the CXTM-specification. I find no
> clear indication of this connection in the document. Is my assumption
> correct?
Actually you will see the reifier attribute listed in the construction 
rules  - look where it says what goes in the [attributes] property of 
an element information item. The rules for construction of the 
attribute value are in 5.15.

> What is the 'text' in attlist.reified. I'm assuming that the reified
> property should lists all elements that are reified by this topic. How
> should these be addressed?
Look in 5.16

> To me it also looks as if '&=' could be replaced by just '=' without 
> loss
> in the above rules. What is the significance of the '&'s here?

Makes it easier to add new attributes in the RNG

> Lastly, why is it necessary to both specify on each topic which 
> elements
> have been reified by that particular topic and at the same time 
> specify on
> each element which topic(s) reify that element. It looks to me like
> unnecessary duplication, and that it would suffice with either a 
> reifier
> or a reified property. I would be keen to understand the use of this.

I want to be sure that the processor is doing the right thing and that 
the reifier/reified properties are correctly set - if I omit one I am 
not testing the full model constructed by the topic map processor.