[xml-h] LMNL

Gavin Thomas Nicol gtn at rbii.com
Sun Jan 19 17:28:58 GMT 2003

On Sunday 19 January 2003 03:56 pm, Elliotte Rusty Harold wrote:
> At 3:24 PM -0500 1/19/03, Simon St.Laurent wrote:
> >(LMNL is a markup language that is designed explicitly for cases where
> >markup needs to overlap, like annotation.  For more information, see
> >http://lmnl.org or http://xmlhack.com/read.php?item=1790 .)
> Moving off on a tangent, how does this relate to Patrick Dursau's
> Just In Time Trees (JITTS), which is also designed around a need for
> potentially overlapping markup?

JITTS is (now) probably closer in philosopy to Attributed Range Algebra (ARA). 
LMNL is a particular (elegant) syntax, a data model, and a reified form that 
offers a means to support alternate syntaxes using the same data model. The 
data model is based on ARA.

JITTS, at least initially, were about being able to choose between a number of 
alternate potential trees in a given document. I think they're now looking at 
supporting alternate syntaxes and to still be able to instantiate trees as 
needed. At that level it is exactly the same as Range Constructors in ARA, 
though in ARA, the canonical data model is not trees, but rather ranges and 
sequences. I've been working to finish of the ARA paper, and also a thing I 
call RAQueL, a range-based query language. At some point I'd like to show 
that all the node-centric data models can be layered over ARA (hard to find 
the time to do things...).

This does have some relevance to hypertext too though, precisely in the area 
of overlapping links, and also ranges as link-ends. It also has some 
connection to XSLT... making XSLT 1.0 support XPointer instead of XPath in 
match and select expressions is possible (I did it with XT), but it raises a 
number of interesting issues (like priority... like Norm, I choose "innermost 
match wins). This also has a pretty significant impact on linkbase design: if 
the linkbase contains XPointers instead of XPaths, it's going to be very hard 
to use XSLT to use a linkbase to add links as a document is transformed.

FWIW. One of the *hardest* things to think about is whether a link authored 
over an XHTML rendition of an XML document should be stored in a linkbase 
against the XML original, or the XHTML rendition. I've never come to a 
reasonable conclusion on that one..

More information about the xml-hypertext mailing list