Re: Comments on XML-Link Spec

Multi-directional links and EXTENDED/LOCATOR
--------------------------------------------
<Q>Is it correct that SIMPLE links can never be part of multi-directional
links? </Q>

On this assumption, multi-dimensional links are defined by an EXTENDED
element containing (?how many) LOCATORs.  Please comment on my assumptions.
Let's assume the first example (unchanged) is in 

threebears.xml:
 
<!ATTLIST A XML-LINK CDATA #FIXED "SIMPLE">
 
<CHAPTER>
Once upon a time there was a 
<HOUSE ID="h1">
in which lived
<BEAR TITLE="Father" ID="b1"><BED XML-TYPE="hard"/></BEAR>
<BEAR TITLE="Mother" ID="b2"><BED XML-TYPE="soft"/></BEAR>
and
<BEAR TITLE="Baby" ID="b3"><BED XML-TYPE="OK"/></BEAR>
</HOUSE>.  One day
<GIRL TITLE="Goldilocks"/> came to a
<A HREF="#h1" ROLE="A" TITLE="houseLink">charming little house</A> but the
<A HREF="ROOT,DESCENDANT(ALL,BEAR)" ROLE="B" TITLE="bearsLink">three bears</A>
were not at home.
</CHAPTER>
 
and let us assume a commentary on this story (i.e. with out-of-line links), 

comment.xml:

<!ELEMENT XL XML-LINK CDATA #FIXED "EXTENDED">
<!ELEMENT LOC XML-LINK CDATA #FIXED "LOCATOR">

<COMMENTARY>
Before the <A HREF="threebears.xml#ROOT,(1,CHAPTER)">story</A> started,
we assume a 
<XL ROLE="MARRIAGE" TITLE="Bear Wedding" ID="marriage">. Long ago
<LOC HREF="threebears.xml#ID(b1)" ROLE="male spouse" ="Father Bear"/> and
<LOC HREF="threebears.xml#ID(b2)" ROLE="female spouse" TITLE="Mother Bear"/>
got married.
</XL>
After a few years, they became a 
<XL ROLE="FAMILY" TITLE="Bear Family" ID="family">
<LOC HREF="threebears.xml#ID(b1)" ROLE="father" ="Father Bear"/>
<LOC HREF="threebears.xml#ID(b2)" ROLE="mother" TITLE="Mother Bear"/>
<LOC HREF="threebears.xml#ID(b3)" ROLE="child" TITLE="Baby Bear"/>
</XL>
<XL ROLE="Parent-child" TITLE="Father-Baby" ID="father">
<LOC HREF="threebears.xml#ID(b1)" ROLE="father" ="Father Bear"/>
<LOC HREF="threebears.xml#ID(b3)" ROLE="child" TITLE="Baby Bear"/>
</XL>
<XL ROLE="Parent-child" TITLE="Mother-Baby" ID="mother">
<LOC HREF="threebears.xml#ID(b2)" ROLE="mother" TITLE="Mother Bear"/>
<LOC HREF="threebears.xml#ID(b3)" ROLE="child" TITLE="Baby Bear"/>
</XL>
</COMMENTARY>

The COMMENTARY contains 4 EXTENDED links.  
The first I assume to be bidirectional (and in this case symmetric).  It 
asserts that the resources b1 and b2 (each are BEAR elements) are the 'ends' 
of a link.  Each end can locate its other end.  (The housekeeping for this is 
not completely trivial, which is why I need to be sure).  Presumably b1
'knows' it is involved in the link commentary.xml#ID(marriage) [though it
would be kept in  a hashtable or something].  If b1 wants to know about
its 'spouse' it asks the 'marriage' link and finds that the other end is b2.

Links 'father' and 'mother' are similar.  After links are  added to the 
database, b1 knows that it is involved in two bidirectional links with 
different roles.  b3 is also involved in 2 bidirectional links, but it
could also be said to be in a multidirectional link where it was one resource
and the 'other end' was a resourceSet with 2 resources b1 and b2.  (This
could only be deduced by adding these semantics to the ROLE.  But this is
one possible defintion of multidirectional link, i.e.

	parent <------------> parent
                      |
                      |
                      V
                    child

If this is what is meant, presumably it can only be created by another
EXTENDED link with contains HREFs to the 'father' and 'mother' XLs.  This
seems complicated.

Alternatively 'family' could be seen as a multi-directional link.  All three
components have equal status.  (What is meant by traversal in this case 
I do not know).  Is this 'three-way' link what is intended?

<Q>Are all the links displayed above (a) legal interpretations of the 
draft and (b) useful interpretations?
</Q>

Terminology:

multi-directional link.  The spec talks of traversal and resources.  I assume
that all the XLs define resources, 'marriage', 'father' and 'mother' defining
2 and 'family' defining three.  I assume that if there are N LOCs in an
XL, then there are N resources and they all have symmetric status.

out-of-line links.  I assume that the four XL elements (NOT the contained
LOCs) are out-of-line links.


	P.



-- 
Peter Murray-Rust, domestic net connection
Virtual School of Molecular Sciences
http://www.vsms.nottingham.ac.uk/

Received on Tuesday, 20 May 1997 19:11:44 UTC