> > Given that the feature is still labeled experimental, do our backwards > compatibility constraints apply? >
Actually, it doesn't look like it is marked with @Experimental or is described as experimental in the docs. That said, I think maybe it *should* have been marked as experimental because I don't think it's completely baked or in common use. I'm not sure how we want to deal with that. I'm generally in favor of switching this model for storing hashes and sets, because I think that's the more common use case. Maybe we just need to document that 1.2 breaks the upgrade from 1.1 for redis data? -Dan
