On Apr 28, 2005, at 7:24 AM, Stefano Mazzocchi wrote:
> conor dowling wrote:
>> I've been reading this mail group over the last two weeks with great
>> interest. After using the current generation of XUL templates to
>> build an RDF-driven display (see:
>> http://www.the325project.org/factlog/325Factlog.xul), I could see a
>> lot of functionality described here and in the Fresnel spec that XUL
>> templates sorely lack. I think that their main omission, besides the
>> lack of remote queries, is support for wildcards and defaults - they
>> need a fallback to grab all properties and use OWL labels to enable
>> compact but powerful displays.
>> There's a new XUL template proposal (see:
>> http://wiki.mozilla.org/XUL:Templates_Plan), ...
> I had not and I'm very happy that you pointed me in that direction.
>
> After a complete-but-not-super-detailed read, I see strong overlap in
> the design methodology and partial (but very interesting!) overlap in
> the functional capabilities.
>
> I personally think that the mozilla folks used RDF for a long time but
> never really got it, they use it as a sort of mergeable property files
> on (big!) steroids. Their templates reflect that: they are pretty solid
> on the UI side (actions, bindings, UI widgets and all that) but lacking
> in the RDF selection and graph->tree generation (which doesn't
> surprise me: you hit that wall only after a while).
you can still do a lot with the Query Processor as is - it's just very
verbose and you need XBL bindings for some workarounds. Opening up the
Query Processor part to the likes of Fresnel addresses most of the
shortcomings.
>
> If XUL templates were to support, say, XPath queries over a DOM (as
> they
> mention), then my vision of Fresnel would fit perfectly: fresnel would
> do the graph selection and graph->DOM manipulation, and XUL templates +
> mozilla will do the rest. Having a fresnel XPCOM component in mozilla
> that does graph->tree rendering would be the only thing needed for this
> to happen.
exactly - a Fresnel Query Processor perhaps one that would leverage and
specialize more basic XPath or SPARQL processors.
>
> XUL templates, to me, represent a perfect example of why we should stop
> illuding ourselves that we can declare everything people will need from
> RDF presentation and just work to bridge existing solutions between the
> two worlds.
which makes the "tree representation", the interchange format from
processor to builder so important. It has to be rich enough for all
envisaged processors. It will be interesting to compare the Mozilla
"interchange tree"
(
http://wiki.mozilla.org/XULTemplatesAPI#Result_Objects) with the one
being spec'ed right now for Fresnel. As I said, I think Mozilla's needs
to be enriched. After the next pass on the Fresnel spec, there should
be a thorough comparison with the template spec to compare Mozilla's
current RDF Query Processor to Fresnel and to make sure that the
Builder and Tree support all of Fresnel's nuances. Coalescing on one
set of terminology (Query Processor, Builder, "Tree Representation")
would be a good thing too. I think Neil Deakin who wrote the V2 XUL
Template spec is on holidays for another week
(
http://www.xulplanet.com/ndeakin/article/299) but he's been very
responsive in the past to any mail I've sent him. Not to be nothing but
a leech, I'm definitely going to write factlog queries using both and
compare and contrast.
Any way, let's see. Sooner Fresnel and remote querying with the likes
of SPARQL go mainstream, the better. Firefox is one route.
>
> --
> Stefano Mazzocchi
> Research Scientist Digital Libraries Research Group
> Massachusetts Institute of Technology location: E25-131C
> 77 Massachusetts Ave telephone: +1 (617) 253-1096
> Cambridge, MA 02139-4307 email: stefanom at mit . edu
> -------------------------------------------------------------------
>
>
Received on Thu Apr 28 2005 - 16:28:38 EDT