I came across this paper on “Generalized Reversible Computation” from the Nop Platform project: https://github.com/entropy-cloud/nop-entropy/blob/master/docs/theory/paper/generalized-reversible-computation-paper-en.md My rough understanding is that it treats semantic model deltas as first-class citizens. That seems to be what distinguishes it from differential/incremental computation: the delta is not primarily over runtime data collections, but over a DSL-defined semantic/model space. The proposed construction is something like:

App = Generator<DSL> ⊕ Δ

So the DSL provides a semantic coordinate system, the generator produces a base model/application structure, and Δ describes structured changes over that model space. The author also seems to be trying to formalize this as a general software construction paradigm, rather than just as an implementation technique for customization. I’m curious how people here would situate this. Is this basically delta-oriented programming / feature-oriented programming / model-driven engineering under another name? Is it related to semantic patching, lenses, or bidirectional transformations? Or is there something meaningfully different in treating semantic model delta as the primary abstraction for software evolution?