[news.eclipse.technology.emft] Re: [EMF Compare] Similarity in ordered l

[laurent Goubet <laurent.goubet@xxxxxxx>]

Hi Guillaume,

yup, that's exactly it. EMF Compare is based on a generic algorithm for 
the metrics to determine objects' similarities... With specific objects 
such as yours when the generic engine doesn't manage to properly match 
the objects, isSimilar() is the way to go.

Laurent Goubet
Obeo

Guillaume Chatelet a écrit :
> Ok, looking closer to the code, I understood I had to write my own 
> MatchEngine (extending from GenericMatchEngine) and override the 
> GenericMatchEngine#isSimilar(EObject obj1, EObject obj2) function.
>
> Guillaume Chatelet wrote:
>
> > Hello Laurent,
>
> > I'd like to compare two objects, say o1 and o2. Those objects contain 
> > an EList<IMyObject>.
>
> > The compare is pretty difficult because the lists might be shuffled 
> > and the object might change a lot.
>
> > In order for the diff ordering to be efficient I need rules to ensure 
> > that two IMyObject are similar. Otherwise the diff algorithm will 
> > conclude with a lot of additions and deletions in the list. So I need 
> > to precise what makes two IMyObject similar or not. I already played 
> > with ReferencesCheck#shouldBeIgnored(EReference) and 
> > AttributesCheck#shouldBeIgnored(EAttribute) as you proposed but it's 
> > not enough. I need to compute intermediate metrics and maybe to give 
> > some attributes more importance than others.
>
> > Do you think this is something that fits in EMF Compare's framework ? 
> > How can I do this ?
>
> > Best regards,
> > Guillaume

begin:vcard
fn:Laurent Goubet
n:Goubet;Laurent
org:<a href="http://www.obeo.fr/";>Obeo</a>
email;internet:laurent.goubet@xxxxxxx
url:http://www.obeo.fr
version:2.1
end:vcard