Optimisation: Produce Delete Events from GreenInterpreter?

[anthonyanjorin]

The green interpreter knows in some cases that a certain match is now invalid. It could produce a delete event in such cases so that the operation is informed and can directly invalidate the match.

[Arikae]

Not sure about this.
Does this means that the event is directly delegated to the green interpreter?
Wouldn't that bypass the whole repair process?

[anthonyanjorin]

If the green interpreter gets a match m that creates an edge between two objects, then it knows already that this match is invalid and can be discarded (no double edges in emf). The green interpreter could therefore invalidate the match directly.

At the moment we don’t do this: we forget the match is invalid and do nothing. Later on the match could be tried again, will be inspected and identified as invalid and then removed. I’m not sure how inefficient this is - but it is certainly redundant.