I have the idea that Sparx tends to release features as soon as possible, even if they are still incomplete. (see component package)
Well, it is all about complexity. EA as a software product has a rather high complexity with functional dependencies and gluing all over the place where even small changes lead to even more changes to be applied in various areas. Keeping track of such things is one of the most challenging tasks in actual requirements engineering and software architecture as I learnt some while ago.
Oliver
While what you say is true, Oliver, it has
always been my view (often expressed) that EA (as a product) is
unnecessarily complex. I have said many times, it is actually more work to be inconsistent than consistent. So many years later, we are still paying for some early inconsistencies in the EA product.
I believe (and I'm prepared to be corrected here) EA doesn't "eat its own dog food". I mean the UI
doesn't use the API/AI to do its work (for example)...
Whether EA was, itself, modelled or not, is - in my view - a
moot point (see my recent postings about the self-inconsistencies with boolean columns). As you said in the post above, "a fool with a tool is still a fool". Whilst not implying that the Sparxians are fools, you must concede that have a
multiplicity of different boolean functionalities is foolish - especially within the
same column!
I suspect that the Sparxians can't make the kinds of changes they might
like/wish to make because they can't be sure of the consequences within the current architecture.
All I know is I have to allow some serious uplift in effort if I'm developing something I haven't used before in EA. It will take me considerable time to make sure I fully understand the "foibles", "gotchas" and plain "bugs" in an area of EA I haven't been into before (or for a while).
My AU$0.05,
Paolo