Re: [modeling-pmc] Sven's comment on Comment to Ed's new main text for t

[Miles Parker <milesparker@xxxxxxxxx>]

On 2012-08-18, at 3:03 PM, Sven Efftinge <sven@xxxxxxxxxxx> wrote:

> your view on the whole topic is so different to mine, that I decided to not go into a more detailed discussion in my first mail.
> However I'm glad Ed did so and I agree with everything he said. I fear that there is to little common ground to find a compromise on. 

It sounds that way to me as well. It does make sense to have a page talking about Eclipse Modeling as a standards driven (driving, really!) platform for MDA -- and people with that motivation can work at driving eyeballs to such a page -- but you get the worst of all worlds when you start designing by committee. Eclipse projects are by design not democratic, but they are certainly participatory. I'm sure that Ed will take into account the feelings of the community, but it's actually more important to me that there is a strong single vision than that it happens to agree with mine, and that's what I'll support. The worst thing is a muddled vision.

> I've added some comments below, but the bottom line however is: 
> I totally respect the work people did at OMG but it is just not relevant for me nor to any of the technologies I've been working on in the past years. 
> And it never was. I never shared those visions nor do I know that anybody I work with does.

+1. In fact, by far the greatest barrier to me in convincing colleagues and organizations to adopt EMP technologies is the word "Model". MDA with most of these folks is the kiss of death. I'm not being hyperbolic. We shouldn't even be getting into the kind of conversations about how model-driven approaches aren't about UML; you've already lost the battle at that point.

We all know the history of that and how it could have been better -- and we don't need to get into a debate about that :) -- but the bottom-line is hubris and obsessive over-generalization. There is something fundamentally broken about trying to develop a single over-arching schema that rules them all -- it actually misses something deep about fundamental CS issues and the way the world works -- the best you can hope for are M3 and M2 representations that allow many schemas to interact consistently and that in my mind is what EMF facilitates.

Phillip has a lot of interesting thoughtful things to say, and we all have different perspectives. There are many domains for which it is very natural to think in OMG and Standards language, but the vast majority of developers only care about whether a) a tool makes it possible to do more things more easily and hopefully more enjoyably, and b) other people can adopt it without having to learn too much new stuff, lock-in to a particular vendor or be left high-and-dry when the next new wave crashes ashore. I've been struck actually by how conservative most Software Engineers are and as I see more and more of the real world issues that they face, I can understand why.

This is the audience that has to be won over. These tools are all powerful and very cool -- regardless of what they do-- and people just need to use them; the greatest barrier to adoption is when people talk about them. The fact that everything that they do in that tool is secretly a "model" (Ed has convinced me that even that is a bad word :) ), well in that case it will just be a happy discovery when a few years later they discover that that is what they really wanted all along.

>> We had a commonly agreed message in the old text, and if we are not able to improve it along the lines Miles described, we should simply keep it.
> 
> I understood that Miles supports Ed's proposal but suggest to add mentioning the support for industry standards, which
> I think is a reasonable approach.

Exactly, in fact if anything I think Ed's current text has far too many three syllable words. Sorry, Ed. ;D

cheers,

Miles