Re: [henshin-dev] Proposal to remove direct creation of edges

[Gregor Bonifer <gregor.bonifer@xxxxxx>]

I think the term cycle was misleading. To exemplify what I meant:
A TrafoSys transfoms an ecore model. There is a Rule with a LHS-graph 
matching two classes. Consider this rule to be an injectively matched 
kernel. There might be a multiRule that extends over one of the matched 
classes. The multiRule demands that the matched class holds a reference 
to some class. If the multiRule is injective there has to be a third 
class in the package. If the multiRule is non-injective this allows 
self-references or a reference to the second class that is matched by 
the kernel.
So what I meant is some kind of optional cohesion between the 
kernelMatch and a multiMatch that is not mandated by a multiMapping.

Regards,
Gregor

Am 14.12.2011 18:59, schrieb Enrico Biermann:
> According to the theory closing a cycle would still not be possible with
> nested amalgamation. At least from a theoretical point of view each
> higher layer completely embeds the last layer whether it is displayed as
> such or not.
>
> Regards,
> Enrico
>
> On 14.12.2011 14:29, Gregor Bonifer wrote:
> > Unique construction it is. The setting for injectivity should be taken
> > from each rule. This allows a (non-injective) multiRule to close
> > cyclic paths by rematching elements matched by a (potentially
> > injective) kernel.
> >
> > However, I think it would be nice to allow some sort of disjoint
> > coverage i.e. global injectivity. I see cases where a globally
> > injective amalgamation would be more efficient than solutions
> > constructed by the means of the existing TUs. But perhaps we should
> > discuss this as a possible extension once unique construction works.
> > Let's save some of the fun ;)
> >
> > Regards,
> > Gregor
> >
> >
> > Am 14.12.2011 09:58, schrieb Christian Krause:
> > > On 14.12.2011 09:49, Stefan Jurack wrote:
> > > > Am 14.12.2011 01:39, schrieb Enrico Biermann:
> > > > > Personally I think the most useful and simplest solution is locally
> > > > > injective matches (depending on the trafo option) and unique
> > > > > construction.
> > > > Agree. The "unique construction" is the strategy that is most intuitive
> > > > and that can be reproduced in ones mind best.
> > > I also think this is the best solution.
> > >
> > > _______________________________________________
> > > henshin-dev mailing list
> > > henshin-dev@xxxxxxxxxxx
> > > https://dev.eclipse.org/mailman/listinfo/henshin-dev
> > _______________________________________________
> > henshin-dev mailing list
> > henshin-dev@xxxxxxxxxxx
> > https://dev.eclipse.org/mailman/listinfo/henshin-dev
> _______________________________________________
> henshin-dev mailing list
> henshin-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/henshin-dev