To be perfectly honest, I've had this capability for several years (the XSD schema says 2009-01). The design works as you say here (add members to a group, control the colors collectively for all members in that group). Currently, each color-enabled group has 8 color categories (background, outline, etc.). Colors can apply to fact types and object types, as well as individual roles, constraints, and role player connector lines. The settings can be applied globally or limited to specific diagrams. Teams who have had this feature use it extensively and end up with some truly garish color combinations:)
Here is my dilemma: the extension that manages this capability is not part of the open source code. The core NORMA components have been open source since the project started at Neumont University, which discovered that private institutions needs to respect a balance sheet (in this case by cutting all research funding). Since then, I've been incrementally working on extensions to make a 'pro' version of the tool. Simply put, I can't feed my family (horse, cars, tractor, etc.) if I give away all of my work. The 'pro' work was largely on hold for several years while I was working salaried for another company writing NORMA extensions to target their proprietary platform, but I am now happy to be in a position to accelerate the pro extensions.
The shape grouping extension, which covers colors and tracking items that are missing a shape, is only part of the picture. Most of the 'pro' work relates to formal derivation rules. However, at this point, the derivation rules don't map to any broadly available physical target. Although the formal rules are extremely useful for discussion and writing a formal specification, until I'm actually generating something for the rules (starting with a relational mapping), I don't think that I have a saleable product. The internal approach is to move the majority of NORMA development (model specification, validation rules, plus RMAP and other generation) into rules and models written directly in NORMA. Once this declarative work and the corresponding code generators are in place, remaining tasks such as full RMAP (including all constraints and derivations) become declarative modeling exercises instead of a coding exercises. This will drastically accelerate the rate of development.
I'm making steady progress on this goal, but am realistically still a couple of years away from making a V1 pro product with full physical mapping a reality.
The unfortunate part about the broad vision is that there are several pieces (like shape coloring, a multi-page relational view, and formal fact and subtype definitions) that are stable but do not have a release vehicle until there is enough product to successfully market (color capabilities are nice but are not a product). I also can't guarantee that there will not be significant changes between now and a final release. For example, basic derivations are stable and have been in heavy use for about three years, but handling of functions and aggregations is changing, subqueries and outer join semantics are not complete, and existential rules (creation of entities in additional to deriving facts between asserted entities) are specified but unimplemented.
So, I'll leave it as a question to the community as to the level of interest in a 'tracking license' system that would allow you to purchase access to the 'pro' extensions before a full release is available. I think many of you would find the current capabilities extremely useful.
Let me know what you think.