The ModVersionResolver currently uses a very naive, greedy algorithm that fails easily.
In general the problem is quite hard, but I'm sure we could do a bit better than now.
It's not clear whether we actually need a better algorithm, because most use cases will probably focus on the latest releases of mods, which makes the problem much easier. It would be good to see real-world failures as a starting point for improvements.
The ModVersionResolver currently uses a very naive, greedy algorithm that fails easily.
In general the problem is quite hard, but I'm sure we could do a bit better than now.
It's not clear whether we actually need a better algorithm, because most use cases will probably focus on the latest releases of mods, which makes the problem much easier. It would be good to see real-world failures as a starting point for improvements.