In contrast you have the data base itself (you can just use reverse engineering to get the ERD of the tables you are interested in)
and a pretty good documentation, both general and for every single package (some of them even with an ERD).
For the reverse engineering you have plenty of tools, I would recommend 'Case Studio' and 'Power Designer'.
I'm currently re-evaluating OpenACS. Without an ERD it would be hard to do. I can't believe things have change that much since the days of Arsdigita. Whether the schema has 200 or 500 tables it doesn't really matter. The value of a toolkit like OpenACS lies in its data model. There must be an ERD somewhere.
Don, are you around?
As for the changes, they are numerous, but the basics should still be fine. What is more important though for the evaluation as a Toolki is not the datamodell but the API that is provided to access the data modell (in my opinion).
I you want to create an ERD of the core functionality, install OpenACS in PostgreSQL and use http://www.rbt.ca/autodoc/ PostgreSQL autodoc to create a DIA script that you can use for the ERD. I have not tried this way, but I think Joel did it once.
Here is some of the core data model
And some probably too detailed documentation of every table
HTML descriptions http://openacs.org/storage/view/miscellaneous/oacs_5_kernel.html
Graphic of tables and relationships http://openacs.org/storage/view/miscellaneous/oacs_5_kernel.png
and the same graphic as a PDF http://openacs.org/storage/view/miscellaneous/oacs_5_kernel.pdf
Thank you so much for the links. I think this should do. Do you know if this data model and ERD apply to the latest version of OpenACS?
On the other hand I have to agree though, that if you want to evaluate the *internals* of OpenACS the ERD should give you the best insights (I'm just thinking about the help this ERD will give us to better understand how to make Dynamic Types work ).