In fact, to clarify, there are two object models that I should have better separated, not only with a line, but also in the prose.
- a programming language theory object model (allowing for several object replicas on user machines) as researched in programming language theory, e.g., sth. like the fully functional object model of Pierce, equipped with a history and ɪᴅs (as drafted in WIP research notes here) and
- a natural language object model, roughly a variation of Live Distributed Object, which I have dubbed conceptual object, that does not touch any compiler, but w.r.t which several choices for programming language object models could be discussed concerning adequacy.
The compiler for 1. can be written, but my list of desiderata is that we also establish 2. a canoncial way to write and talk about objects: the conceptual object, coming with local views/replicas, or whatever is compatible with the literature, ongoing research, and disourse elsehwere on the forums, e.g., here.
In summary, there are two object models mixed in this post
- compiler touching (above the line), a triple ⟨T,H,i⟩, s.t. …
- conceptual (below the line), explaining 1. using natural language