Ring. Hat. Toss.
In my opinion Sparx is too complex to use for large companies (Gartner's target audience). Multinationals want cheap generic employees doing cheap generic work as cheaply as possible, and just good enough to get by (especially in IT). To use EA to anywhere near its capabilities you need to REALLY understand UML AND SQL (for model queries) AND its API.
If you are a professional, Sparx is a great (maybe the best) tool. It's just that it's not part of Gartner's target audience.
See, now that's interesting. I see it a little differently.
In my view EA is an enthusiast's tool --
not a professional's. In order to make good use of EA you need someone in the organization who knows the ins and outs of the tool. It's not enough to know the modelling language.
EA is a toolkit. It's a rag-bag of features that don't work well together, and which sometimes are different attempts to implement the same functionality. Unless you intend to work only within the tool, it comes with nothing to help you in any given usage scenario: no usable project setups, model structures, or document templates. What's there is all demo-level stuff. On the IT side of things, EA integrates poorly and clunkily. And even the extension mechanisms are limited: you can access the most basic functions through the API, but only a very small number of the high-level ones.
Navigating this quagmire requires deep insight, which must either be built up within the organization, or purchased from a third party. All of this makes EA a far more expensive tool to use than the purchase price would suggest.
As to this Gartner stuff, I don't think it's necessarily a big deal in and of itself. It's just marketing. If Sparx have taken the positive decision not to cooperate with Gartner for this report, that can be simply because they haven't see enough return on that in previous years.
But I am concerned for EA's future.
To some extent, the failure of EA is the failure of UML: it promised effective communication in all things software, but very quickly turned its back on all real world issues involved in software development and instead disappeared up its own backside, issuing new versions of the language standard with ever more complex syntax and semantics, but making no attempt at implementing profiles for any architectural patterns, design principles, or even source code languages. So if you take the trouble of teaching all stakeholders a completely alien symbolic language, you can then use it to express... nothing that actually matches the reality you encounter at any stage of the software development process.
The failure in both cases is one of attitude. Instead of facing inwards and trying to come up with more clever bits to bolt on, there needs to be a 180
o shift of focus outwards to what the tool is actually used for. No one is ever going to do their project planning in EA, but they will use it to draw UML diagrams, and those diagrams need to be made easily available in other contexts.
I firmly believe that what EA needs is
not yet more superficial features spread like manure across a weedy field. What it needs is, at the front end, packaging of sets of features for different real-world scenarios, with profiles, document templates, project structures, user security roles, version control setup, integrations with other tools, and
locking off of extraneous features. At the back end, it needs optimization in the data layer, a complete overhaul of the IT integration principles and a substantially expanded API.
I also believe that we the community could help Sparx in this. We have the expertise in using this tool in the wild; Sparx don't. They don't supply their own consultants, and internally, they only develop one piece of software so even if they do eat their own dog food it's only one flavour of one brand.
So we could actually help in prioritizing what improvements to make and what features to focus on for a given goal, eg reducing cost of adoption in mid-to-large organizations. But that would of course also require a higher level of community commitment from Sparx.
Well, that's what I think anyway.
/Uffe
Show Posts
