[furcas-discussion] Reference Re-Evaluation Algorithm

[Stephan Erb]

Hey all,

I have to revert an earlier statement of mine: The IA should NOT update
token values. It may only perform updates which do not directly affect
the textual representation (e.g., it may adapt the referecedElements). 

My current understanding implies that it has to be the editor that calls
the ShortPrettyPrinter, as I cannot think of another sane
synchronization  model for background parsing.  I will elaborate all my
current findings (and their implementation) on our next meeting

Kind Regards,
Stephan


On Sun, 2011-02-06 at 23:09 +0100, Stephan Erb wrote:
> Hey everybody,
> 
> here the result of our last whiteboard session:
> 
> > Create a collect expression only consisting of lookupScope and referenceBy. (It does not contain ?, prefix or postfix).
> > Register the IA on this expression
> > Register the IA on the lookupScope expression
> >
> > If the IA triggers for one of these two registrations, then:
> >
> > 	Filter the affected elements. Only those may remain for which a TB is 
> > 	documenting that the above triggered expression may be executed.
> >	(See IAExpressionInvalidationChangeListener in the MOIN code base)
> >
> > 	For all remaining effected elements and their associated document nodes:
> > 		
> >		Determine if the given reference is bound. Probably this works by inspecting
> >		the 'referrencedModelElements'. The context element at hand may be visible
> >		in several textual views. If at least one of these documents that the reference is bound
> >		then we consider this to be true. Views indicating otherwise are considered to be stale.
> >
> >		evaluate the lookupScope
> >
> >		If bound and the bound element is still in the current lookup scope then:
> >			
> >			Update all corresponding token values. The tokens are set to
> >			the evaluation result of referenceBy on the bound element.
> >			Prefix and postfix are substracted. If prefix and postfix do not match
> >			the given name then the bound reference will be unset.
> > 
> >		else if bound but the bound element is no longer in the lookup scope then:
> >			
> >			Retrieve the name of the previously bound element (overruling the
> <			respective token value). Use this name to perform a fresh lookup
> >			using lookupScope, referenceBy, prefix and postfix. Set the reference
> >			based on the result or break it, if nothing was found.
> >		
> >		else if not bound:
> >
> >			Perform a fresh lookup. The lookup is performed with the current token value
> >			at hand, using lookupScope, referenceBy, prefix and postfix. Set the reference
> >			based on the result or break it, if nothing was found.
> >
> 
> Tasks:
> 
>       * Axel will implement a ModelUpdater using to the given algorithm.
>       * Sebastian will write several test cases:	
>               * Tests covering all branches of the algorithm.
>               * A testcase where the model is constructed in memory. We
>                 will use it to test the algorithm if no TB model is
>                 available.
>               * A set of testcases operating on the text level:
>                 search/replace in the TB model with a subsequent call of
>                 the incremental parser.
> 
> 
> Happy Hacking!
> Stephan
> 
> PS: Thomas, we might need your help with all this textblocks magic (e.g.
> find out if a reference is still bound). Expect some questions on the
> mailing list rather soon :-)
> 
> 
> 
> 
> 
> _______________________________________________
> Furcas-discussion mailing list
> Furcas-discussion at lists.furcas.org
> http://lists.furcas.org/cgi-bin/mailman/listinfo/furcas-discussion