A few months back I was discussing (unhappily) self-repairing polymers, the problems with them and my general dislike of their future. Compare that with this paper on "mendable" polymers. Not a subtle semantic difference, but a totally difference approach - working with (thermally) reversible reactions, either in the polymerization and/or the crosslinking. They are largely relying on the Diels-Alder (DA) reaction, but also other reactions such as disulfide bridges and photodimerization to a lesser extent. When the material needs repair, the reaction is reversed to the monomers, and then re-initiated. Instead of a two-component system in the "self healing" polymers, there's only one component.
The problem here is the limited number of polymer systems that can be easily reversed. You're not going to get too far with plain vanilla polyethylene. Certainly that is a real concern with the monomers, but with the crosslinkers it is a much less concern, as you can have the crosslinking groups as moieties hanging off the backbone. The challenge then becomes a numbers game. There have to be sufficient number of each group so that they can find each other without relying on extensive reptation.
One possible advantage of the DA reaction is that it require two different reactants, the diene and the dienophile. If half of the polyXXX backbone is modified with the diene and the other half of the polyXXX is modified with the the dieneophile, then the reaction within the blend is assured to be intermolecular, not intramolecular. That would not necessarily be assured with the disulfide bridges or photodimerization.
Certainly these are not perfect systems, as they require a separate processing step. The "self-healing" materials do just that - they heal themselves. I doubt that this will be a winner-take-all contest (very little is). I'm just glad there is a second horse in the race.