Depending on your access patterns, maybe you could have a hash table mapping entities ids to indexes in your SoA. Perhaps that's viable if looking up a single e

Depending on your access patterns, maybe you could have a hash table mapping entities ids to indexes in your SoA. Perhaps that's viable if looking up a single entity is not typical to your use case? > Which leads to my theory… I feel like Bevy could be implemented on top of an in-memory DuckDB and get away with it Haha, it certainly does sound viable.