Re: [wtp-incubator-dev] RelaxNG OM checked in

[David Carver <d_a_carver@xxxxxxxxx>]

In regards to your ModelQuery question, that might be the best approach 
to take.

Nitin can probably chime in as well, if he sees any problems with that 
approach to tying into the existing content assistance.

Dave

Martin Schmied wrote:
> Hi Dave,
>
> thank you for commiting the sources & binaries. I have checked
> everything out and made some modifications to make everything compile
> & run. I had to redo some modifications of rng-om in order to use it
> (basically just relieving some access restrictions to make use of
> rngom.binary package).
>
> Some features will not work because they require further modifications
> of rng-om (e.g. displaying documentation comments from schema) or both
> rng-om & jing (datatype library contribution). Content completion &
> very basic validation works OK within my test project.
>
> I've also looked over the WST content model and here are my observations:
>   - it's XSD-centric: it has the semantics of "group" and "any" taken
> from XSD, atributes have "required" flag...
>   - it most probably expects a-priory knowledge of element's content
> model based only on it's name
>   - you cannot construct such a content model from RNG - you can only
> derive such specific content model using context from an instance XML
> document, not just from the RNG schema itself
>
> However, I've also found that most (all?) of the content assist stuff
> is shielded from the content model through ModelQuery. It might be
> possible to implement ModelQuery for RNG - such an implementation
> would derive element's model every time content model for such element
> is requested. There are many other requirements for the RNG content
> completion to work smoothly - e.g. that the results of XML content
> assist processor are not cached anywhere as the content model of RNG
> element can change even after adding / removing an element, but it
> might be worth exploring as using the built-in content assist
> processor is a much cleaner way to go than offering my own.
>
> The bottomline is that I might be able to make classes that use
> ModelQuery work with RNG through a new implementation of ModelQuery.
> Classes that use CMDocument directly won't probably play well with RNG
> as you cannot construct a model this specific a-priori.
>
> Martin
>
> On Mon, Nov 2, 2009 at 11:54 PM, David Carver <d_a_carver@xxxxxxxxx> wrote:
>   
> > Martin, I checked in the last dependency that we were waiting for under
> > parallel IP.  The RNGOM package has been checked into CVS.  Please check
> > everything out and make any modifications that you need to make.  Also
> > verify that I have the JING binary plugin checked in correctly, I think I
> > might have messed that one up and might have to redo it.
> >
> > Dave
> >
> >
> > _______________________________________________
> > wtp-incubator-dev mailing list
> > wtp-incubator-dev@xxxxxxxxxxx
> > https://dev.eclipse.org/mailman/listinfo/wtp-incubator-dev
> >
> >     
> _______________________________________________
> wtp-incubator-dev mailing list
> wtp-incubator-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/wtp-incubator-dev
>
>