Who is saying the object model is cast in stone? What's being said is that some of us disagree with you on the wisdom of adding new fields which will only be of use to a subset of objects. In part my objection is due to the fact that in PG unused fields carry a per-row space penalty and people already worry about the cost of objects as they exist today.
If you want to rewrite the object model to use one of Guy Harrison's superior approaches, feel free. Just remember, though, you'll need to write upgrade scripts so folks can migrate existing sites in situ ...
I guess in this sense, yeah, the object model *is* somewhat cast in stone for OpenACS 4. We inherited it from aD and for better or worse chose not to rewrite it when we started our migration project. I doubt you'll find a single member of the OpenACS implementation team that would argue that the model we inherited is perfect or even close to it. However, when we started our assumption was that aD would stay in business and support ACS Tcl 4.x for Oracle and that we'd just be maintaining the Postgres version. So we had very practical reasons for not rewriting the object datamodel.
In practice it seems to work quite well as is ...
.
