EMF Facet, a new project for model customization and extensibility

Add a Comment

Comment by Mark Dalgarno on April 29, 2010 at 19:23

Let's keep the debate healthy and informative people.

Comment by Rui Curado on April 29, 2010 at 17:39

There's an interesting discussion here:

"How to explain EMF?
Have you ever tried to explain EMF to nonbelievers? I find it difficult to explain what EMF is and why it makes sense to use it."
http://michaelscharf.blogspot.com/2009/04/how-to-explain-emf.html

On this discussion we have comments that range from

"I think EMF and some of the other Eclipse modeling technology are the hidden gems of the Eclipse community."

to

"I slightly disagree with Ian Skerret. Its not only hard to explain to newbies, its freaking hard to explain for people who already know something about it. [...] I have found the learning curve for EMF and associated technologies like oaw to be amazingly steep."

A recommended read.

Comment by Frédéric Madiot on April 29, 2010 at 17:19

>Like I said, Vlad works for a tool vendor. His job is to "not like" free alternatives. ;)
I don't think that's a problem : I am product manager for a tool vendor too ;-)
The problem is that Vlad is stucked with his business model. And he is completely misunderstanding what is EMF.
EMF is not a competitor for tool vendors. At the opposite they can benefit from integrating EMF within their proprietary tools (as we have done with Mia-Studio).
Vlad, rather than writing cons-truth, you should spend your time looking at how you could integrate or contribute to technologies like EMF or compete with them ... on a technological field.

Comment by Xavier Seignard on April 29, 2010 at 17:06

EMF can't be a part of the 20% Rui. EMF is the basement of the future devolopments of the eclipse platform itself. Have a look on E4 : http://wiki.eclipse.org/E4.

Anyway, thanks for your reply.

Regards

Comment by Rui Curado on April 29, 2010 at 15:54

"Every time we speak about open source tools in this modeldrivensoftware network, you just shit on them, why?"

Like I said, Vlad works for a tool vendor. His job is to "not like" free alternatives. ;)

Personally I had a quick look at EMF and it doesn't really look complex. The Ecore meta-metamodel looks quite simple actually. But because EMF integrates and relies on other Eclipse technologies, EMF's average complexity increases because these other Eclipse subsystems throw their own complexity at it.

Unlike Vlad, I don't see Eclipse disappearing. However, I see the 80/20 rule here: 80% of Eclipse users will just use 20% of Eclipse modules (projects). EMF may well become a part of those 20%...

Comment by Xavier Seignard on April 29, 2010 at 15:25

"Do you think that EMF technologies should be manipulated by end users such modelers or by tool providers ?"
Nobody says that.

"I mean that EMF technology is certainly a great modeling technology but the learning curve is too hard and requires more than 18 month full time training for a senior developer in order to be able to respect the object approach with EMF MDSD technologies."
That's a LIE. Regarding my experience (and the people I worked with), you need around 4/5 months for an intermediate developer.

"Why to use EMF open source projects if you can purchase existing tool from vendors doing the same job ?
Why to invest into an heavy EMF integration if your business is not strictly related to EMF (e.g. modeling tool vendors) ?"
I'll tell you a story, 5 month ago we choose to switch from MIA Transformation to EMF, for a small perimeter in a project, because we knew with EMF that we did will exactly fit our need. And you know what? We use both MIA Studio AND EMF. Incredible? No, that's just a casual day on a MDSD developer's life...

"It seems to me that Eclipse foundation and plugin projects are telling us a dreaming story which is that the community can replace tool vendors but the reality is different. Many companies start modeling eclipse project and almost no success story. Less than 1% of successful project !!"
That's just propaganda...

"My business problem is that during this evaluation stage no company purchase anymore modeling software tools. If they don't purchase then independent software vendors invest less in R&D and in the meantime open source projects do not replace traditional tool vendors R&D investment."
So how do you explain we choose MIA Studio after benchmarking open source and commercial tools?

"It means that we have a lot larger modeling community but also a lot less real R&D investment which are usually replace by integrators. My feeling is that Eclipse open source modeling project will sooner or later disappears or only used by a very qualified but very small group of users."
Come on! You can't believe what you say!

"It is time for a change inside Eclipse and go back to common sense. EMF technologies should only be used for specific purposes and certainly not to replace traditional modeling such as UML. EMF technologies should not replace existing code generators such as MIA, Blue Age, AndroMDA which take an UML model xmi as entry and then generate code. The model should not be EMF but UML !!"
But?! Do you really know what is EMF? EMF is a framework that helps you to define your metamodel (among others) and then create models that conforms this metamodel. So you can define your UML metamodel and then create your UML models. In fact there is already an EMF implementation of the UML metamodel in the Eclipse Modeling projet. And this implementation is widely used by other open source AND commercial tools (such as MIA Studio).

"EMF given to end users which are only looking for modeling (e.g no advanced java code integration) is a non sense for me today !!"
Yes it's true, but it has nothing to do with the blog post of M. Madiot.

Every time we speak about open source tools in this modeldrivensoftware network, you just shit on them, why?

Comment by Rui Curado on April 29, 2010 at 13:43

"I mean that EMF technology is certainly a great modeling technology but the learning curve is too hard and requires more than 18 month full time training for a senior developer..."

"Many companies start modeling eclipse project and almost no success story. Less than 1% of successful project !!"

From where are you basing these numbers on, Vlad? I had a feeling EMF was quite complex but you are putting it on a whole new level. I guess because Eclipse has no direct financial concerns, there are no limits to expansion, no need to make it "lean and mean".

Of course, we both *want to continue/want to be* tool vendors, so we see this through our own coloured glasses... I am sure EMF people (especially Ed Merks and company) completely disagree with you.