[news.eclipse.tools.emf] Re: Ecore2Ecore data migration when reference's

[Ed Merks <Ed.Merks@xxxxxxxxx>]

Louis,

Comments below.

Louis Rose wrote:
> Hi all,
>
> I'm trying to use an Ecore2Ecore mapping to perform data migration as 
> per the EMF Recipe here:
>
> http://wiki.eclipse.org/EMF/Recipes#Recipe:_Data_Migration
>
>
> I have the following Ecore model:
>
> @namespace(uri="faultsets", prefix="faultsets")
> package faultsets;
>
> class System {
>    val Tuple[*] tuples;
> }
>
> class Tuple {
>   attr String name;
>   val IdentifierSet[*] contents;
> }
>
> class IdentiferSet
>  val Identifier[*] contents;
> }
>
> abstract class Identifier {}
>
> class Wildcard extends Identifier {}
> class NoFault extends Identifier {}
> ...
>
>
> Which has evolved to this Ecore model:
>
> @namespace(uri="faultsets", prefix="faultsets")
> package faultsets;
>
> class System {
>   val Tuple[*] tuples;
> }
>
> class Tuple {
>   attr String name;
>   val Atom[*] contents;
> }
>
> abstract class Atom {}
>
> class Wildcard extends Atom {}
> class NoFault extends Atom {}
> ...
>
>
> The contents of a tuple have been simplified. A tuple no longer 
> contains a set of sets, just a set. 
That sounds like the remove of levels of nesting in the XML.  I don't 
think the mapping can handle that...
> In addition, the Identifier class has been renamed to Atom.
>
> The recipe described on the EMF wiki suggests constructing an 
> Ecore2Ecore mapping (and corresponding Ecore2Xml mapping) and then 
> processing the getEObjectToExtensionMap() in the postLoad() method of 
> a custom ResourceHandler.
>
> I'm not sure how to define the Ecore2Ecore mapping such that I can 
> process the IdentifierSets for instances of the original Ecore model. 
> When I define a mapping between Identifier and Atom, the 
> getEObjectToExtensionMap() contains an entry for each Tuple for the 
> contents feature, but the value is null. (I was expecting the value to 
> contain data representing an instance of the now absent IdentifierSet 
> EClass).
It would be easier to comment if you showed the XML...
> When I don't define a mapping between Identifier and Atom, I receive 
> the following exception when trying to load the model:
>
> org.eclipse.emf.ecore.xmi.IllegalValueException: Value 
> 'org.eclipse.emf.ecore.xml.type.impl.AnyTypeImpl@945ff9 (eClass: 
> org.eclipse.emf.ecore.impl.EClassImpl@3d648c (name: Atom) 
> (instanceClassName: null) (abstract: false, interface: false)) (mixed: 
> null, anyAttribute: null)' is not legal. (Before.faultsets, 5, 48)
>     at 
> org.eclipse.emf.ecore.xmi.impl.XMLHandler.setFeatureValue(XMLHandler.java:2662) 
>
>     at 
> org.eclipse.emf.ecore.xmi.impl.XMLHandler.setFeatureValue(XMLHandler.java:2647) 
>
>     at 
> org.eclipse.emf.ecore.xmi.impl.XMLHandler.createObjectFromFeatureType(XMLHandler.java:2166) 
>
>     at 
> org.eclipse.emf.ecore.xmi.impl.XMLHandler.createObject(XMLHandler.java:2034) 
>
>     at 
> org.eclipse.emf.ecore.xmi.impl.XMIHandler.createObject(XMIHandler.java:134) 
>
>     at 
> org.eclipse.emf.ecore.xmi.impl.XMLHandler.handleFeature(XMLHandler.java:1813) 
>
>     at 
> org.eclipse.emf.ecore.xmi.impl.XMLHandler.processElement(XMLHandler.java:1023) 
>
>     at 
> org.eclipse.emf.ecore.xmi.impl.XMIHandler.processElement(XMIHandler.java:87) 
>
>     at 
> org.eclipse.emf.ecore.xmi.impl.XMLHandler.startElement(XMLHandler.java:1001) 
>
>
>
> I don't fully understand the way in which the Ecore2Ecore and 
> Ecore2Xml mappings interact with EMF to load instances of the original 
> Ecore model, so I'm sure I'm doing something silly because of my 
> ignorance. If anybody could shed any light on how I can proceed, I'd 
> be very grateful.
It sounds likely that your changes are more complex than what the 
mapping approach can handle...
> Many thanks,
> Louis.
>
> ----
> Louis Rose
> Research Student
> Department of Computer Science,
> University of York,
> Heslington, York, YO10 5DD, United Kingdom.
> +44 1904 434762
> Twitter: @louismrose